You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by "Matthias J. Sax" <mj...@apache.org> on 2026/05/27 05:43:04 UTC

[DISCUSS] Tightening time-based releases

Hello,

I would like to start a discussion about our time-based release 
schedule. We have a discussion (that I personally find a little bit 
annoying and tedious) about release plan dates on every single release 
(the 4.4.0 discussion, is just another example). And we still often end 
up with none-optimal time-frames (4.4 might also be an example here -- 
not by itself, but the impact on 4.5 release timeline -- cf my latest 
reply on the other thread).

Given that we actually plan to have time-based releases, I think we 
should agree on a more fixed schedule, which allows us to project all 
release dates (including KIP/feature/code freeze) into the future and 
just be done with it. -- And even if we need an exception from the rule, 
we can easily get back on track on future release afterwards.

I was also looking into historic data, and while we overall don't do too 
bad to stay on track for 3 release per year, we had to pay the price of 
a few short release cycles to "shift back" (the previous 4.3 release is 
a good example of it), while we still did shift somewhat forward 
overall. These short releases cause even more pain IMHO.

To avoid problems with the holiday season end of the year, I would like 
to propose the following fixed release cadence (this is not my personal 
favorite, but let's start here):

1. KIP freeze is the second Wednesday of January, May, and September.
2. Feature freeze = KIP freeze + 3 weeks.
3. Code freeze = Feature freeze + 3 weeks.
4. Release target = Code freeze + 4 weeks.

While we never officially formalized the 3+3 weeks between KIP/feature 
and feature/code freeze dates (and we used shorter time windows in the 
past, ie, 2+2 weeks), 3+3 established itself since 4.0 release -- I am 
totally open for other time frames, but believe we should not go shorter 
than 2+2 weeks. I would also not make them longer than 3+3, as it would 
just give us the subjective impression that releases are closer to each 
other.

A 3+2 schedule would actually be my personal preference. For KIPs, 
giving only 2 weeks to complete them is tight, even for simple KIPs. And 
for bug-fixes, I would believe that 2 weeks are enough. Making it 
overall 5 week instead of 6, also helps to keep releases 1 week "further 
apart" from each other (not really, but we know how reality works, given 
our subjective nature).

The above schedule implies that we would have target release dates late 
March, July, and November. Even if we need more than 4 weeks after code 
freeze, we still have some head room in the November release before the 
holidays (we might hit Thanksgiving, but I think that's ok). For the KIP 
freeze in January, it should also give enough time after the holidays to 
complete KIPs and get them accepted (w/o being too close to the 
holidays, eg, "first Wednesday in January" would effectively mean, that 
KIPs must get voted before the holiday break).

If we want to take the holiday season into account even more, we could 
also make the timeline between "release 3" and "release 1 of the 
following year" two weeks longer, and the other two timelines 1 week 
shorter. This would give us the following KIP freeze deadlines:

  - third Wednesday of January
  - second Wednesday of May
  - first Wednesday of September

It also implies, that Thanksgiving would be less of an issue for 
"release 3". This schedule is what I would personally prefer, instead of 
an 100% equal distribution of our 52 weeks across all three releases.



For the upcoming 4.4 release, we should make it a 5-month cycle 
targeting October instead of September, and we would also have a 
slightly longer cycle for 4.6, which we should only start in January. 
However, starting with 4.6 we would enter the above proposed fixed 
timeline, putting us into a much better position going forward IMHO.


Looking forward to your thoughts on this idea.


-Matthias



Re: [DISCUSS] Tightening time-based releases

Posted by Chia-Ping Tsai <ch...@apache.org>.
hi all,

Thanks for driving this. I fully support moving to a fixed 3+2 schedule to eliminate our constant release alignment discussions. Christmas is wonderful, but honestly, Lunar New Year is my personal favorite. I guess Diwali is also a favorite for many of our core maintainers.

Hence, my two cents is a 5 / 3.5 / 3.5-month split:

Release 1: Early May target (KIP freeze in late Feb)
Release 2: Late August target (KIP freeze in mid-June)
Release 3: Early December target (KIP freeze in early Sept)

If we apply 3+2 policy to the current 4.4.0 schedule, it naturally pulls the code freeze forward to July 20. This allows us to target late August for 4.4.0, making it a perfect fit for "Release 2" and smoothly clearing September for the next KIP freeze

Best,
Chia-Ping

On 2026/05/27 05:43:04 "Matthias J. Sax" wrote:
> Hello,
> 
> I would like to start a discussion about our time-based release 
> schedule. We have a discussion (that I personally find a little bit 
> annoying and tedious) about release plan dates on every single release 
> (the 4.4.0 discussion, is just another example). And we still often end 
> up with none-optimal time-frames (4.4 might also be an example here -- 
> not by itself, but the impact on 4.5 release timeline -- cf my latest 
> reply on the other thread).
> 
> Given that we actually plan to have time-based releases, I think we 
> should agree on a more fixed schedule, which allows us to project all 
> release dates (including KIP/feature/code freeze) into the future and 
> just be done with it. -- And even if we need an exception from the rule, 
> we can easily get back on track on future release afterwards.
> 
> I was also looking into historic data, and while we overall don't do too 
> bad to stay on track for 3 release per year, we had to pay the price of 
> a few short release cycles to "shift back" (the previous 4.3 release is 
> a good example of it), while we still did shift somewhat forward 
> overall. These short releases cause even more pain IMHO.
> 
> To avoid problems with the holiday season end of the year, I would like 
> to propose the following fixed release cadence (this is not my personal 
> favorite, but let's start here):
> 
> 1. KIP freeze is the second Wednesday of January, May, and September.
> 2. Feature freeze = KIP freeze + 3 weeks.
> 3. Code freeze = Feature freeze + 3 weeks.
> 4. Release target = Code freeze + 4 weeks.
> 
> While we never officially formalized the 3+3 weeks between KIP/feature 
> and feature/code freeze dates (and we used shorter time windows in the 
> past, ie, 2+2 weeks), 3+3 established itself since 4.0 release -- I am 
> totally open for other time frames, but believe we should not go shorter 
> than 2+2 weeks. I would also not make them longer than 3+3, as it would 
> just give us the subjective impression that releases are closer to each 
> other.
> 
> A 3+2 schedule would actually be my personal preference. For KIPs, 
> giving only 2 weeks to complete them is tight, even for simple KIPs. And 
> for bug-fixes, I would believe that 2 weeks are enough. Making it 
> overall 5 week instead of 6, also helps to keep releases 1 week "further 
> apart" from each other (not really, but we know how reality works, given 
> our subjective nature).
> 
> The above schedule implies that we would have target release dates late 
> March, July, and November. Even if we need more than 4 weeks after code 
> freeze, we still have some head room in the November release before the 
> holidays (we might hit Thanksgiving, but I think that's ok). For the KIP 
> freeze in January, it should also give enough time after the holidays to 
> complete KIPs and get them accepted (w/o being too close to the 
> holidays, eg, "first Wednesday in January" would effectively mean, that 
> KIPs must get voted before the holiday break).
> 
> If we want to take the holiday season into account even more, we could 
> also make the timeline between "release 3" and "release 1 of the 
> following year" two weeks longer, and the other two timelines 1 week 
> shorter. This would give us the following KIP freeze deadlines:
> 
>   - third Wednesday of January
>   - second Wednesday of May
>   - first Wednesday of September
> 
> It also implies, that Thanksgiving would be less of an issue for 
> "release 3". This schedule is what I would personally prefer, instead of 
> an 100% equal distribution of our 52 weeks across all three releases.
> 
> 
> 
> For the upcoming 4.4 release, we should make it a 5-month cycle 
> targeting October instead of September, and we would also have a 
> slightly longer cycle for 4.6, which we should only start in January. 
> However, starting with 4.6 we would enter the above proposed fixed 
> timeline, putting us into a much better position going forward IMHO.
> 
> 
> Looking forward to your thoughts on this idea.
> 
> 
> -Matthias
> 
> 
> 

Re: [DISCUSS] Tightening time-based releases

Posted by Bill Bejeck <bb...@apache.org>.
Hi Matthias,

Thanks for leading a discussion on release dates.  I think it will go a
long way to help with planning and delivery with a more concrete schedule.
I agree with Andrew for 3 releases with the 3+2 schedule.

-Bill

On Wed, May 27, 2026 at 3:44 AM Andrew Schofield <sc...@apache.org>
wrote:

> Hi Matthias,
> Thanks for driving this. I'm a bit less concerned about it than you, but
> stronger rules would eliminate the typical release schedule alignment
> discussions we've seen in all of the recent releases.
>
> I would go for 3 releases of equal lengths in March, July and November. I
> favour the 3+2 schedule.
>
> Thanks,
> Andrew
>
> On 2026/05/27 05:43:04 "Matthias J. Sax" wrote:
> > Hello,
> >
> > I would like to start a discussion about our time-based release
> > schedule. We have a discussion (that I personally find a little bit
> > annoying and tedious) about release plan dates on every single release
> > (the 4.4.0 discussion, is just another example). And we still often end
> > up with none-optimal time-frames (4.4 might also be an example here --
> > not by itself, but the impact on 4.5 release timeline -- cf my latest
> > reply on the other thread).
> >
> > Given that we actually plan to have time-based releases, I think we
> > should agree on a more fixed schedule, which allows us to project all
> > release dates (including KIP/feature/code freeze) into the future and
> > just be done with it. -- And even if we need an exception from the rule,
> > we can easily get back on track on future release afterwards.
> >
> > I was also looking into historic data, and while we overall don't do too
> > bad to stay on track for 3 release per year, we had to pay the price of
> > a few short release cycles to "shift back" (the previous 4.3 release is
> > a good example of it), while we still did shift somewhat forward
> > overall. These short releases cause even more pain IMHO.
> >
> > To avoid problems with the holiday season end of the year, I would like
> > to propose the following fixed release cadence (this is not my personal
> > favorite, but let's start here):
> >
> > 1. KIP freeze is the second Wednesday of January, May, and September.
> > 2. Feature freeze = KIP freeze + 3 weeks.
> > 3. Code freeze = Feature freeze + 3 weeks.
> > 4. Release target = Code freeze + 4 weeks.
> >
> > While we never officially formalized the 3+3 weeks between KIP/feature
> > and feature/code freeze dates (and we used shorter time windows in the
> > past, ie, 2+2 weeks), 3+3 established itself since 4.0 release -- I am
> > totally open for other time frames, but believe we should not go shorter
> > than 2+2 weeks. I would also not make them longer than 3+3, as it would
> > just give us the subjective impression that releases are closer to each
> > other.
> >
> > A 3+2 schedule would actually be my personal preference. For KIPs,
> > giving only 2 weeks to complete them is tight, even for simple KIPs. And
> > for bug-fixes, I would believe that 2 weeks are enough. Making it
> > overall 5 week instead of 6, also helps to keep releases 1 week "further
> > apart" from each other (not really, but we know how reality works, given
> > our subjective nature).
> >
> > The above schedule implies that we would have target release dates late
> > March, July, and November. Even if we need more than 4 weeks after code
> > freeze, we still have some head room in the November release before the
> > holidays (we might hit Thanksgiving, but I think that's ok). For the KIP
> > freeze in January, it should also give enough time after the holidays to
> > complete KIPs and get them accepted (w/o being too close to the
> > holidays, eg, "first Wednesday in January" would effectively mean, that
> > KIPs must get voted before the holiday break).
> >
> > If we want to take the holiday season into account even more, we could
> > also make the timeline between "release 3" and "release 1 of the
> > following year" two weeks longer, and the other two timelines 1 week
> > shorter. This would give us the following KIP freeze deadlines:
> >
> >   - third Wednesday of January
> >   - second Wednesday of May
> >   - first Wednesday of September
> >
> > It also implies, that Thanksgiving would be less of an issue for
> > "release 3". This schedule is what I would personally prefer, instead of
> > an 100% equal distribution of our 52 weeks across all three releases.
> >
> >
> >
> > For the upcoming 4.4 release, we should make it a 5-month cycle
> > targeting October instead of September, and we would also have a
> > slightly longer cycle for 4.6, which we should only start in January.
> > However, starting with 4.6 we would enter the above proposed fixed
> > timeline, putting us into a much better position going forward IMHO.
> >
> >
> > Looking forward to your thoughts on this idea.
> >
> >
> > -Matthias
> >
> >
> >
>

Re: [DISCUSS] Tightening time-based releases

Posted by Chia-Ping Tsai via dev <de...@kafka.apache.org>.
+1 to option 2

I guess the culprit of the delay is still the "must-have backports/CVEs" :)

On 2026/10/31 20:44:28 Andrew Schofield via dev wrote:
> Hi Matthias,
> Option 2 works for me and seems like a good way to transition to
> the more regular release cadence in the future.
> 
> Thanks,
> Andrew
> 
> On 2026/10/29 22:32:18 Matthias J. Sax wrote:
> > Thanks Andrew.
> > 
> > I see that you did keep Jan/13 for KIP freeze, and did not pull it 
> > forward. LGTM.
> > 
> > 
> > > but I think it's very hard to be concrete about 4.6.0 at this point.
> > 
> > Is it? In the end, if we say we want to follow the long-term proposed 
> > timeline in general, we are looking into the following dates:
> > 
> > Proposed KIP-freeze dates for 2027
> > 
> > Release-1: March/3 (week 9)
> > Relasee-2: June/23 (week 25)
> > Release-3: Oct/13 (week 41)
> > 
> > So there is two options:
> > 
> >   (1) We keep March/3, and do 4 release in 2027. Personally, I don't 
> > think this make a lot of sense. With the Jan/13 KIP freeze for AK 4.5 
> > the cycle for AK 4.6 would be way too short IMHO.
> > 
> >   (2) We treat the AK 4.5 release as release-1, ie, pull forward the 
> > March/13 date to Jan/13, and follow with release-2 in June/23 as 
> > planned. -- Here we would still have some wiggle room: release-1 is 7 
> > weeks earlier than the proposed schedule, so it would make sense to pull 
> > forward release-2, too. If we keep June/23 for release-2, we would have 
> > 23 weeks for AK 4.6 release, and only 16 weeks for release-3. It might 
> > be better to even it out and move release-2 forward by 3 week to have 20 
> > weeks for AK 4.6, and 19 weeks for AK 4.7 with Oct/13 KIP freeze date as 
> > originally proposed.
> > 
> > 
> > I would propose to go with option-2 and doing AK 4.6 with KIP freeze 
> > Jun/2 (ie, 3 weeks earlier than the proposed long-term schedule).
> > 
> > 
> > -Matthias
> > 
> > 
> > On 10/12/26 3:17 AM, Andrew Schofield wrote:
> > > Hi,
> > > I have gone ahead with the 4.5.0 schedule as listed, but I think it's very
> > > hard to be concrete about 4.6.0 at this point.
> > > 
> > > Thanks,
> > > Andrew
> > > 
> > > On 2026/10/10 20:33:35 Muralidhar Basani via dev wrote:
> > >> Hi all,
> > >>
> > >> I'm not deeply involved in release management, so please take this as just
> > >> some data in case it's useful for the 4.5/4.6 discussion.
> > >>
> > >> I compared the planned dates on the wiki release plans with what actually
> > >> happened in git for the last five releases:
> > >>
> > >> Version  Planned Feat Freeze  Branch cut  Planned target   Released
> > >>   Late   RCs
> > >> 4.0.0      2024-12-11                  2024-12-16  2025-01-29
> > >>   2025-03-18     48d    5
> > >> 4.1.0      2025-06-04                  2025-06-10  2025-07-09
> > >>   2025-09-02     55d    4
> > >> 4.2.0      2025-11-19                  2025-11-21  2026-01-07
> > >>   2026-02-16     40d    5
> > >> 4.3.0      2026-03-18                 2026-03-18  2026-05-06
> > >>   2026-05-20     14d    3
> > >> 4.4.0      2026-07-29                 2026-07-27  2026-09-09
> > >>   2026-10-12*    33d    5
> > >> (* expected)
> > >>
> > >> Feature freeze is usually on time; the time between code freeze and the
> > >> release is where we tend to need more than planned (median ~40 days after
> > >> the target). With the current 4.5.0 plan, that would put the 4.5.0 release
> > >> around late April 2027, which may be worth keeping in mind when choosing
> > >> the 4.6.0 dates.
> > >>
> > >> I put the small script in this PR, in case it's helpful for future
> > >> planning: https://github.com/apache/kafka/pull/23769
> > >> Added sample runs in the PR.
> > >> If this is not the direction we want, happy to close it.
> > >>
> > >>
> > >> Here is a short snippet from the scripts run
> > >>
> > >> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > >>
> > >> % python3 release/schedule.py
> > >> ....
> > >> ....
> > >> version  KIP freeze    derived from          feature freeze  code freeze
> > >>   first RC        target release       realistic release
> > >> -------     ----------         -----------------         --------------
> > >>       -----------       ----------        --------------
> > >> -----------------
> > >> 4.5.0     2027-01-13   wiki release plan   2027-02-03      2027-02-17
> > >> 2027-03-16  2027-03-17          2027-04-28
> > >> 4.6.0     2027-05-19   history                   2027-06-09      2027-06-23
> > >>    2027-07-20  2027-07-21          2027-09-01
> > >> 4.7.0     2027-09-22   history                   2027-10-13      2027-10-27
> > >>    2027-11-23  2027-11-24           2028-01-05
> > >> ...
> > >>
> > >> %python3 release/schedule.py --release-date 4.4.0=2026-10-12 status 4.6.0
> > >> ....
> > >> ....
> > >> Release 4.6.0 as of 2026-10-10 (dates: history)
> > >> KIP freeze         2027-05-19  (in 221 days)
> > >> feature freeze     2027-06-09  (in 242 days)
> > >> code freeze        2027-06-23  (in 256 days)
> > >> first RC           2027-07-20  (in 283 days)
> > >> target release     2027-07-21  (in 284 days)
> > >> realistic release  2027-09-01  (in 326 days)
> > >> ...
> > >> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > >>
> > >>
> > >> Thanks,
> > >> Murali
> > >>
> > >> On Sat, Oct 10, 2026 at 3:33 PM Matthias J. Sax <mj...@apache.org> wrote:
> > >>
> > >>> What's you rational for doing 4.6 in March already? If we so 4.5
> > >>> early/mid January, it would be a very small release. Given the overhead
> > >>> releases have, I am wondering if it would be worth doing?
> > >>>
> > >>> For 4.5, I am also ok to start a Jan/6 instead of Jan/13 with KIP
> > >>> freeze. If we do 4.6 in July (instead of March) I don't see a need to
> > >>> pull it forward by one a week though. But I am fine in any case.
> > >>>
> > >>> -Matthias
> > >>>
> > >>> On 9/28/26 11:49 PM, Dejan Stojadinović wrote:
> > >>>>> "Anyway, all I want to say is that the original schedule was hard to
> > >>>> follow. The codebase is large and CVEs are common these days."
> > >>>>
> > >>>> As an occasional contributor I can confirm this.
> > >>>>   From my POV committers/maintainers are having a challenging task to keep
> > >>>> up, especially while trying to juggle around with multiple release
> > >>>> branches/versions (trunk/4.5.0, 4.4.0, 4.3.2, 4.2.2).
> > >>>>
> > >>>> Best regards,
> > >>>> Dejan S.
> > >>>> https://github.com/dejan2609
> > >>>>
> > >>>>
> > >>>> On Tue, Sep 29, 2026 at 7:55 AM Chia-Ping Tsai <ch...@apache.org>
> > >>> wrote:
> > >>>>
> > >>>>> Hi Andrew,
> > >>>>>
> > >>>>> +1 to the revised 4.5.0 schedule. I would not pull it a week earlier,
> > >>>>> since that puts the KIP freeze right after the holidays.
> > >>>>>
> > >>>>> Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0
> > >>>>> reaches code freeze. We had too many must-have changes in 4.4.0, which
> > >>>>> caused a serious delay. I try to review all backports but my tokens are
> > >>>>> limited :)
> > >>>>>
> > >>>>> Anyway, all I want to say is that the original schedule was hard to
> > >>>>> follow. The codebase is large and CVEs are common these days.
> > >>>>>
> > >>>>> Best,
> > >>>>> Chia-Ping
> > >>>>>
> > >>>>>
> > >>>>> On 2026/09/25 08:36:23 Andrew Schofield wrote:
> > >>>>>> Hi,
> > >>>>>> I'm starting to look at the release schedule for 4.5.0.
> > >>>>>>
> > >>>>>> In Chia-Ping's table from 29 May 2026, we have:
> > >>>>>>
> > >>>>>> Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
> > >>>>>> 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19,
> > >>> 2026
> > >>>>>> 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
> > >>>>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> > >>>>>>
> > >>>>>> In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP
> > >>>>> freeze of Jan 13, 2027 which gives this.
> > >>>>>>
> > >>>>>> 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon
> > >>> now
> > >>>>>> 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17,
> > >>> 2027
> > >>>>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> > >>>>>>
> > >>>>>> I think we could even pull 4.5.0 a week earlier than that, but I don't
> > >>>>> think there's anything fundamentally wrong with a short 4.6.0 provided
> > >>>>> people know in advance.
> > >>>>>>
> > >>>>>> Are people happy with the 4.5.0 (revised) dates above? Would a week
> > >>>>> earlier be better given that it gives an extra week to 4.6.0 which is
> > >>>>> looking very squeezed?
> > >>>>>>
> > >>>>>> Thanks,
> > >>>>>> Andrew
> > >>>>>>
> > >>>>>> On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
> > >>>>>>> Hi all,
> > >>>>>>>
> > >>>>>>> +1 to the 4.5 schedule, given the delay of 4.4.0.
> > >>>>>>>
> > >>>>>>> I think having 4.5 land right on Lunar New Year should be fine this
> > >>>>> time, since "too close" is the real issue here.
> > >>>>>>>
> > >>>>>>> Best,
> > >>>>>>> Chia-Ping
> > >>>>>>>
> > >>>>>>>> Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> > >>>>>>>>
> > >>>>>>>> I don't think we have for 4.5... I believe that there was a slight,
> > >>>>> high level agreement for the long term, that we want to follow the
> > >>>>> "calendar week proposal"?
> > >>>>>>>>
> > >>>>>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with
> > >>>>> feature freeze
> > >>>>>>>>>>>>> +3 weeks, and code freeze +2 weeks.
> > >>>>>>>>
> > >>>>>>>> AK 4.4 KIP freeze was Jul/8.
> > >>>>>>>>
> > >>>>>>>> For 2026, week 41 would be Oct/7 which is way too close IMHO.
> > >>>>>>>> For 2027, week 9 would be Mar/3 which seems to be too far out?
> > >>>>>>>>
> > >>>>>>>> Second release (week 25) for 2027 would be Jun/23, which would be
> > >>>>> 4.6?
> > >>>>>>>>
> > >>>>>>>> So the question remains, where to put 4.5 so we can start to follow
> > >>>>> the new schedule starting with 4.6. It seems KIP freeze between Jan 13
> > >>>>> might be a good compromise putting it roughly between 4.4 and 4.6 KIP
> > >>>>> freeze dates. Of course, the release would overlap with Chinese New
> > >>> Year,
> > >>>>> but it would be the last time we have this overlap.
> > >>>>>>>>
> > >>>>>>>> Just my 2tc.
> > >>>>>>>>
> > >>>>>>>>
> > >>>>>>>> -Matthias
> > >>>>>>>>
> > >>>>>>>>> On 9/15/26 8:17 AM, David Jacot wrote:
> > >>>>>>>>> Hi all,
> > >>>>>>>>> Have we converged on the schedule that we want to use yet? Based on
> > >>>>>>>>> the last emails, it is not so clear to me. For your context, I am
> > >>>>>>>>> trying to figure out the timeline for 4.5.
> > >>>>>>>>> Best,
> > >>>>>>>>> David
> > >>>>>>>>>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <
> > >>>>> christololov@gmail.com> wrote:
> > >>>>>>>>>>
> > >>>>>>>>>> I think we could be a bit more lax with the KIP freeze of 6.7.0
> > >>>>> when
> > >>>>>>>>>> 2034 comes around to account for LNY :)
> > >>>>>>>>>>
> > >>>>>>>>>> Joking aside, my preference between "KIP freeze would be
> > >>>>> Wednesday's,
> > >>>>>>>>>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> > >>>>>>>>>> weeks" and "start Q1 release with X buffer after LYN and complete
> > >>>>> Q3
> > >>>>>>>>>> release with Y buffer before Christmas" would be to start with the
> > >>>>>>>>>> first and adjust by shifting by 1 week on individual basis the ones
> > >>>>>>>>>> which are too close. I am okay with too close being defined as 2
> > >>>>>>>>>> weeks. Based on your very detailed tables we seem to have only a
> > >>>>> few
> > >>>>>>>>>> such outliers in the next decade.
> > >>>>>>>>>>
> > >>>>>>>>>> Best,
> > >>>>>>>>>> Christo
> > >>>>>>>>>>
> > >>>>>>>>>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org>
> > >>>>> wrote:
> > >>>>>>>>>>>
> > >>>>>>>>>>> Matthias, you are definitely a master of scheduling! I'm +100 to
> > >>>>> your plans.
> > >>>>>>>>>>>
> > >>>>>>>>>>> Also, it is absolutely helpful to write down the exact release
> > >>>>> dates for the next few years at once. It will ease our anxiety a lot!
> > >>> I'd
> > >>>>> be more than happy to update the wiki with these dates once we all
> > >>> agree on
> > >>>>> this new schedule. For example:
> > >>>>>>>>>>>
> > >>>>>>>>>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target
> > >>>>> Release
> > >>>>>>>>>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19,
> > >>>>> 2026
> > >>>>>>>>>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9,
> > >>>>> 2026
> > >>>>>>>>>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5,
> > >>>>> 2027
> > >>>>>>>>>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25,
> > >>>>> 2027
> > >>>>>>>>>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15,
> > >>>>> 2027
> > >>>>>>>>>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3,
> > >>>>> 2028
> > >>>>>>>>>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23,
> > >>>>> 2028
> > >>>>>>>>>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13,
> > >>>>> 2028
> > >>>>>>>>>>>
> > >>>>>>>>>>> Furthermore, let's do our best to avoid fixing release blockers
> > >>>>> under the Christmas tree :)
> > >>>>>>>>>>>
> > >>>>>>>>>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> > >>>>>>>>>>>> Thanks for the input.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
> > >>>>>>>>>>>> windows gets a lot of support.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> However, the idea of dividing the release windows not into 3
> > >>>>> equally
> > >>>>>>>>>>>> long junks, but be more flexible to account for holiday seasons,
> > >>>>> seems
> > >>>>>>>>>>>> not to get a lot of support? Not sure why... I personally think
> > >>>>> it would
> > >>>>>>>>>>>> be good to take it into account, and I also think that Chia-Ping
> > >>>>> raises
> > >>>>>>>>>>>> a very good point.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising
> > >>>>> it. I
> > >>>>>>>>>>>> did actually briefly think about it, but it's difficult, and
> > >>>>> thus I just
> > >>>>>>>>>>>> started with something simple as proposal to kick off a
> > >>>>> discussion.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> I think, Diwali is short enough to not make it a concern
> > >>>>> (similar to
> > >>>>>>>>>>>> Thanksgiving)?
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> For Lunar New Year, I agree that it would be good to plan for it.
> > >>>>>>>>>>>> However, it seems it's shifting quite a lot year-over-year what
> > >>>>> makes it
> > >>>>>>>>>>>> very difficult (w/o a huge window) to include?
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced
> > >>>>> (this
> > >>>>>>>>>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5
> > >>>>> month
> > >>>>>>>>>>>> releases does not sound very appealing?
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> I was playing with it a little bit more. Overall we have 52
> > >>>>> weeks, so
> > >>>>>>>>>>>> roughly 17.3 weeks per release. So we can naturally split it into
> > >>>>>>>>>>>> 17/17/18 using the additional week for the holidays. But that's
> > >>>>> not good
> > >>>>>>>>>>>> enough to take both Christmas and Lunar New Year into account.
> > >>>>> However,
> > >>>>>>>>>>>> with a 16/16/20, we could make it work I believe, even if it
> > >>>>> does cut it
> > >>>>>>>>>>>> tight sometimes... To be able to phrase a "simple rule", we can
> > >>>>> use ISO
> > >>>>>>>>>>>> weeks:
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature
> > >>>>> freeze
> > >>>>>>>>>>>> +3 weeks, and code freeze +2 weeks.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer
> > >>>>> for most
> > >>>>>>>>>>>> years:
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> > >>>>>>>>>>>>>     │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │
> > >>>>> Buffer                 │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2027 │ Feb 6  │ Mar 3                      │ 25 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2028 │ Jan 26 │ Mar 1                      │ 34 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2029 │ Feb 13 │ Feb 28                     │ 15 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2030 │ Feb 3  │ Feb 27                     │ 24 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2031 │ Jan 23 │ Feb 26                     │ 34 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2032 │ Feb 11 │ Feb 25                     │ 14 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2033 │ Jan 31 │ Mar 2                      │ 30 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ←
> > >>>>> tightest with realistic LNY │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>>>>     │ 2035 │ Feb 8  │ Feb 28                     │ 20 days
> > >>>>>                          │
> > >>>>>>>>>>>>>
> > >>>>>
> > >>>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> Guess, we could also make it week-8 instead of 9, and accept
> > >>>>> that 2034
> > >>>>>>>>>>>> would cut it too short (or be even more flexible and keep week-9
> > >>>>> for
> > >>>>>>>>>>>> 2034 as an exception to the rule), but that's a single year...
> > >>>>> for 2029
> > >>>>>>>>>>>> and 2032 we still have one week buffer.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> For Christmas, we would still ensure that the release can go out
> > >>>>> on
> > >>>>>>>>>>>> time, too. The below Q3 target release date is 4 weeks after code
> > >>>>>>>>>>>> freeze. This is a little bit more tight, as we often need more
> > >>>>> than 4
> > >>>>>>>>>>>> weeks to get it done:
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>>     ┌──────┬──────────────────────────┬────────────────┐
> > >>>>>>>>>>>>>     │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2026 │ Dec 9                    │ 16             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2027 │ Dec 15                   │ 10             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2028 │ Dec 13                   │ 12             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2029 │ Dec 12                   │ 13             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2030 │ Dec 11                   │ 14             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2031 │ Dec 10                   │ 15             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2032 │ Dec 8                    │ 17             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2033 │ Dec 14                   │ 11             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2034 │ Dec 13                   │ 12             │
> > >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>>>>     │ 2035 │ Dec 12                   │ 13             │
> > >>>>>>>>>>>>>     └──────┴──────────────────────────┴────────────────┘
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> For some years, we would still cut it a little bit tight, but we
> > >>>>> could
> > >>>>>>>>>>>> also apply an exception to the rule, to compensate for it. If we
> > >>>>> have
> > >>>>>>>>>>>> enough buffer room from LYN, we can start the Q1 release train a
> > >>>>> week
> > >>>>>>>>>>>> earlier, reducing the Q3->Q1 timeline by one week.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> Curious to hear what people think about it. In the end, if we
> > >>>>> can agree
> > >>>>>>>>>>>> on high level rules like "start Q1 release with X buffer after
> > >>>>> LYN", and
> > >>>>>>>>>>>> "complete Q3 release with Y buffer before Christmas", and still
> > >>>>> aim for
> > >>>>>>>>>>>> an as balance week-allocation per release, we can just put
> > >>>>> together a
> > >>>>>>>>>>>> release plan for the next 10 years and be done with it.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> -Matthias
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
> > >>>>>>>>>>>>> Hello,
> > >>>>>>>>>>>>>
> > >>>>>>>>>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's
> > >>>>> release
> > >>>>>>>>>>>>> coincided with a reasonable amount of holidays, so I would
> > >>>>> favour a
> > >>>>>>>>>>>>> 3+2 schedule with releases in March, July and November.
> > >>>>>>>>>>>>>
> > >>>>>>>>>>>>> Best,
> > >>>>>>>>>>>>> Christo
> > >>>>>>>>>>>>>
> > >>>>>>>>>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com>
> > >>>>> wrote:
> > >>>>>>>>>>>>>>
> > >>>>>>>>>>>>>> Hi,
> > >>>>>>>>>>>>>>
> > >>>>>>>>>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <
> > >>>>> schofielaj@apache.org>
> > >>>>>>>>>>>>>> wrote:
> > >>>>>>>>>>>>>>
> > >>>>>>>>>>>>>>> I would go for 3 releases of equal lengths in March, July and
> > >>>>> November. I
> > >>>>>>>>>>>>>>> favour the 3+2 schedule.
> > >>>>>>>>>>>>>>>
> > >>>>>>>>>>>>>>
> > >>>>>>>>>>>>>> +1
> > >>>>>>>>>>>>>>
> > >>>>>>>>>>>>>> Thanks,
> > >>>>>>>>>>>>>> Ismael
> > >>>>>>>>>>>>
> > >>>>>>>>>>>>
> > >>>>>>>>
> > >>>>>>>
> > >>>>>>
> > >>>>>
> > >>>
> > >>>
> > >>
> > 
> > 
> 

Re: [DISCUSS] Tightening time-based releases

Posted by Andrew Schofield via dev <de...@kafka.apache.org>.
Hi Matthias,
Option 2 works for me and seems like a good way to transition to
the more regular release cadence in the future.

Thanks,
Andrew

On 2026/10/29 22:32:18 Matthias J. Sax wrote:
> Thanks Andrew.
> 
> I see that you did keep Jan/13 for KIP freeze, and did not pull it 
> forward. LGTM.
> 
> 
> > but I think it's very hard to be concrete about 4.6.0 at this point.
> 
> Is it? In the end, if we say we want to follow the long-term proposed 
> timeline in general, we are looking into the following dates:
> 
> Proposed KIP-freeze dates for 2027
> 
> Release-1: March/3 (week 9)
> Relasee-2: June/23 (week 25)
> Release-3: Oct/13 (week 41)
> 
> So there is two options:
> 
>   (1) We keep March/3, and do 4 release in 2027. Personally, I don't 
> think this make a lot of sense. With the Jan/13 KIP freeze for AK 4.5 
> the cycle for AK 4.6 would be way too short IMHO.
> 
>   (2) We treat the AK 4.5 release as release-1, ie, pull forward the 
> March/13 date to Jan/13, and follow with release-2 in June/23 as 
> planned. -- Here we would still have some wiggle room: release-1 is 7 
> weeks earlier than the proposed schedule, so it would make sense to pull 
> forward release-2, too. If we keep June/23 for release-2, we would have 
> 23 weeks for AK 4.6 release, and only 16 weeks for release-3. It might 
> be better to even it out and move release-2 forward by 3 week to have 20 
> weeks for AK 4.6, and 19 weeks for AK 4.7 with Oct/13 KIP freeze date as 
> originally proposed.
> 
> 
> I would propose to go with option-2 and doing AK 4.6 with KIP freeze 
> Jun/2 (ie, 3 weeks earlier than the proposed long-term schedule).
> 
> 
> -Matthias
> 
> 
> On 10/12/26 3:17 AM, Andrew Schofield wrote:
> > Hi,
> > I have gone ahead with the 4.5.0 schedule as listed, but I think it's very
> > hard to be concrete about 4.6.0 at this point.
> > 
> > Thanks,
> > Andrew
> > 
> > On 2026/10/10 20:33:35 Muralidhar Basani via dev wrote:
> >> Hi all,
> >>
> >> I'm not deeply involved in release management, so please take this as just
> >> some data in case it's useful for the 4.5/4.6 discussion.
> >>
> >> I compared the planned dates on the wiki release plans with what actually
> >> happened in git for the last five releases:
> >>
> >> Version  Planned Feat Freeze  Branch cut  Planned target   Released
> >>   Late   RCs
> >> 4.0.0      2024-12-11                  2024-12-16  2025-01-29
> >>   2025-03-18     48d    5
> >> 4.1.0      2025-06-04                  2025-06-10  2025-07-09
> >>   2025-09-02     55d    4
> >> 4.2.0      2025-11-19                  2025-11-21  2026-01-07
> >>   2026-02-16     40d    5
> >> 4.3.0      2026-03-18                 2026-03-18  2026-05-06
> >>   2026-05-20     14d    3
> >> 4.4.0      2026-07-29                 2026-07-27  2026-09-09
> >>   2026-10-12*    33d    5
> >> (* expected)
> >>
> >> Feature freeze is usually on time; the time between code freeze and the
> >> release is where we tend to need more than planned (median ~40 days after
> >> the target). With the current 4.5.0 plan, that would put the 4.5.0 release
> >> around late April 2027, which may be worth keeping in mind when choosing
> >> the 4.6.0 dates.
> >>
> >> I put the small script in this PR, in case it's helpful for future
> >> planning: https://github.com/apache/kafka/pull/23769
> >> Added sample runs in the PR.
> >> If this is not the direction we want, happy to close it.
> >>
> >>
> >> Here is a short snippet from the scripts run
> >>
> >> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >>
> >> % python3 release/schedule.py
> >> ....
> >> ....
> >> version  KIP freeze    derived from          feature freeze  code freeze
> >>   first RC        target release       realistic release
> >> -------     ----------         -----------------         --------------
> >>       -----------       ----------        --------------
> >> -----------------
> >> 4.5.0     2027-01-13   wiki release plan   2027-02-03      2027-02-17
> >> 2027-03-16  2027-03-17          2027-04-28
> >> 4.6.0     2027-05-19   history                   2027-06-09      2027-06-23
> >>    2027-07-20  2027-07-21          2027-09-01
> >> 4.7.0     2027-09-22   history                   2027-10-13      2027-10-27
> >>    2027-11-23  2027-11-24           2028-01-05
> >> ...
> >>
> >> %python3 release/schedule.py --release-date 4.4.0=2026-10-12 status 4.6.0
> >> ....
> >> ....
> >> Release 4.6.0 as of 2026-10-10 (dates: history)
> >> KIP freeze         2027-05-19  (in 221 days)
> >> feature freeze     2027-06-09  (in 242 days)
> >> code freeze        2027-06-23  (in 256 days)
> >> first RC           2027-07-20  (in 283 days)
> >> target release     2027-07-21  (in 284 days)
> >> realistic release  2027-09-01  (in 326 days)
> >> ...
> >> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >>
> >>
> >> Thanks,
> >> Murali
> >>
> >> On Sat, Oct 10, 2026 at 3:33 PM Matthias J. Sax <mj...@apache.org> wrote:
> >>
> >>> What's you rational for doing 4.6 in March already? If we so 4.5
> >>> early/mid January, it would be a very small release. Given the overhead
> >>> releases have, I am wondering if it would be worth doing?
> >>>
> >>> For 4.5, I am also ok to start a Jan/6 instead of Jan/13 with KIP
> >>> freeze. If we do 4.6 in July (instead of March) I don't see a need to
> >>> pull it forward by one a week though. But I am fine in any case.
> >>>
> >>> -Matthias
> >>>
> >>> On 9/28/26 11:49 PM, Dejan Stojadinović wrote:
> >>>>> "Anyway, all I want to say is that the original schedule was hard to
> >>>> follow. The codebase is large and CVEs are common these days."
> >>>>
> >>>> As an occasional contributor I can confirm this.
> >>>>   From my POV committers/maintainers are having a challenging task to keep
> >>>> up, especially while trying to juggle around with multiple release
> >>>> branches/versions (trunk/4.5.0, 4.4.0, 4.3.2, 4.2.2).
> >>>>
> >>>> Best regards,
> >>>> Dejan S.
> >>>> https://github.com/dejan2609
> >>>>
> >>>>
> >>>> On Tue, Sep 29, 2026 at 7:55 AM Chia-Ping Tsai <ch...@apache.org>
> >>> wrote:
> >>>>
> >>>>> Hi Andrew,
> >>>>>
> >>>>> +1 to the revised 4.5.0 schedule. I would not pull it a week earlier,
> >>>>> since that puts the KIP freeze right after the holidays.
> >>>>>
> >>>>> Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0
> >>>>> reaches code freeze. We had too many must-have changes in 4.4.0, which
> >>>>> caused a serious delay. I try to review all backports but my tokens are
> >>>>> limited :)
> >>>>>
> >>>>> Anyway, all I want to say is that the original schedule was hard to
> >>>>> follow. The codebase is large and CVEs are common these days.
> >>>>>
> >>>>> Best,
> >>>>> Chia-Ping
> >>>>>
> >>>>>
> >>>>> On 2026/09/25 08:36:23 Andrew Schofield wrote:
> >>>>>> Hi,
> >>>>>> I'm starting to look at the release schedule for 4.5.0.
> >>>>>>
> >>>>>> In Chia-Ping's table from 29 May 2026, we have:
> >>>>>>
> >>>>>> Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
> >>>>>> 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19,
> >>> 2026
> >>>>>> 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
> >>>>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> >>>>>>
> >>>>>> In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP
> >>>>> freeze of Jan 13, 2027 which gives this.
> >>>>>>
> >>>>>> 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon
> >>> now
> >>>>>> 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17,
> >>> 2027
> >>>>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> >>>>>>
> >>>>>> I think we could even pull 4.5.0 a week earlier than that, but I don't
> >>>>> think there's anything fundamentally wrong with a short 4.6.0 provided
> >>>>> people know in advance.
> >>>>>>
> >>>>>> Are people happy with the 4.5.0 (revised) dates above? Would a week
> >>>>> earlier be better given that it gives an extra week to 4.6.0 which is
> >>>>> looking very squeezed?
> >>>>>>
> >>>>>> Thanks,
> >>>>>> Andrew
> >>>>>>
> >>>>>> On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
> >>>>>>> Hi all,
> >>>>>>>
> >>>>>>> +1 to the 4.5 schedule, given the delay of 4.4.0.
> >>>>>>>
> >>>>>>> I think having 4.5 land right on Lunar New Year should be fine this
> >>>>> time, since "too close" is the real issue here.
> >>>>>>>
> >>>>>>> Best,
> >>>>>>> Chia-Ping
> >>>>>>>
> >>>>>>>> Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> >>>>>>>>
> >>>>>>>> I don't think we have for 4.5... I believe that there was a slight,
> >>>>> high level agreement for the long term, that we want to follow the
> >>>>> "calendar week proposal"?
> >>>>>>>>
> >>>>>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with
> >>>>> feature freeze
> >>>>>>>>>>>>> +3 weeks, and code freeze +2 weeks.
> >>>>>>>>
> >>>>>>>> AK 4.4 KIP freeze was Jul/8.
> >>>>>>>>
> >>>>>>>> For 2026, week 41 would be Oct/7 which is way too close IMHO.
> >>>>>>>> For 2027, week 9 would be Mar/3 which seems to be too far out?
> >>>>>>>>
> >>>>>>>> Second release (week 25) for 2027 would be Jun/23, which would be
> >>>>> 4.6?
> >>>>>>>>
> >>>>>>>> So the question remains, where to put 4.5 so we can start to follow
> >>>>> the new schedule starting with 4.6. It seems KIP freeze between Jan 13
> >>>>> might be a good compromise putting it roughly between 4.4 and 4.6 KIP
> >>>>> freeze dates. Of course, the release would overlap with Chinese New
> >>> Year,
> >>>>> but it would be the last time we have this overlap.
> >>>>>>>>
> >>>>>>>> Just my 2tc.
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> -Matthias
> >>>>>>>>
> >>>>>>>>> On 9/15/26 8:17 AM, David Jacot wrote:
> >>>>>>>>> Hi all,
> >>>>>>>>> Have we converged on the schedule that we want to use yet? Based on
> >>>>>>>>> the last emails, it is not so clear to me. For your context, I am
> >>>>>>>>> trying to figure out the timeline for 4.5.
> >>>>>>>>> Best,
> >>>>>>>>> David
> >>>>>>>>>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <
> >>>>> christololov@gmail.com> wrote:
> >>>>>>>>>>
> >>>>>>>>>> I think we could be a bit more lax with the KIP freeze of 6.7.0
> >>>>> when
> >>>>>>>>>> 2034 comes around to account for LNY :)
> >>>>>>>>>>
> >>>>>>>>>> Joking aside, my preference between "KIP freeze would be
> >>>>> Wednesday's,
> >>>>>>>>>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> >>>>>>>>>> weeks" and "start Q1 release with X buffer after LYN and complete
> >>>>> Q3
> >>>>>>>>>> release with Y buffer before Christmas" would be to start with the
> >>>>>>>>>> first and adjust by shifting by 1 week on individual basis the ones
> >>>>>>>>>> which are too close. I am okay with too close being defined as 2
> >>>>>>>>>> weeks. Based on your very detailed tables we seem to have only a
> >>>>> few
> >>>>>>>>>> such outliers in the next decade.
> >>>>>>>>>>
> >>>>>>>>>> Best,
> >>>>>>>>>> Christo
> >>>>>>>>>>
> >>>>>>>>>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org>
> >>>>> wrote:
> >>>>>>>>>>>
> >>>>>>>>>>> Matthias, you are definitely a master of scheduling! I'm +100 to
> >>>>> your plans.
> >>>>>>>>>>>
> >>>>>>>>>>> Also, it is absolutely helpful to write down the exact release
> >>>>> dates for the next few years at once. It will ease our anxiety a lot!
> >>> I'd
> >>>>> be more than happy to update the wiki with these dates once we all
> >>> agree on
> >>>>> this new schedule. For example:
> >>>>>>>>>>>
> >>>>>>>>>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target
> >>>>> Release
> >>>>>>>>>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19,
> >>>>> 2026
> >>>>>>>>>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9,
> >>>>> 2026
> >>>>>>>>>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5,
> >>>>> 2027
> >>>>>>>>>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25,
> >>>>> 2027
> >>>>>>>>>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15,
> >>>>> 2027
> >>>>>>>>>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3,
> >>>>> 2028
> >>>>>>>>>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23,
> >>>>> 2028
> >>>>>>>>>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13,
> >>>>> 2028
> >>>>>>>>>>>
> >>>>>>>>>>> Furthermore, let's do our best to avoid fixing release blockers
> >>>>> under the Christmas tree :)
> >>>>>>>>>>>
> >>>>>>>>>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> >>>>>>>>>>>> Thanks for the input.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
> >>>>>>>>>>>> windows gets a lot of support.
> >>>>>>>>>>>>
> >>>>>>>>>>>> However, the idea of dividing the release windows not into 3
> >>>>> equally
> >>>>>>>>>>>> long junks, but be more flexible to account for holiday seasons,
> >>>>> seems
> >>>>>>>>>>>> not to get a lot of support? Not sure why... I personally think
> >>>>> it would
> >>>>>>>>>>>> be good to take it into account, and I also think that Chia-Ping
> >>>>> raises
> >>>>>>>>>>>> a very good point.
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising
> >>>>> it. I
> >>>>>>>>>>>> did actually briefly think about it, but it's difficult, and
> >>>>> thus I just
> >>>>>>>>>>>> started with something simple as proposal to kick off a
> >>>>> discussion.
> >>>>>>>>>>>>
> >>>>>>>>>>>> I think, Diwali is short enough to not make it a concern
> >>>>> (similar to
> >>>>>>>>>>>> Thanksgiving)?
> >>>>>>>>>>>>
> >>>>>>>>>>>> For Lunar New Year, I agree that it would be good to plan for it.
> >>>>>>>>>>>> However, it seems it's shifting quite a lot year-over-year what
> >>>>> makes it
> >>>>>>>>>>>> very difficult (w/o a huge window) to include?
> >>>>>>>>>>>>
> >>>>>>>>>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced
> >>>>> (this
> >>>>>>>>>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5
> >>>>> month
> >>>>>>>>>>>> releases does not sound very appealing?
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> I was playing with it a little bit more. Overall we have 52
> >>>>> weeks, so
> >>>>>>>>>>>> roughly 17.3 weeks per release. So we can naturally split it into
> >>>>>>>>>>>> 17/17/18 using the additional week for the holidays. But that's
> >>>>> not good
> >>>>>>>>>>>> enough to take both Christmas and Lunar New Year into account.
> >>>>> However,
> >>>>>>>>>>>> with a 16/16/20, we could make it work I believe, even if it
> >>>>> does cut it
> >>>>>>>>>>>> tight sometimes... To be able to phrase a "simple rule", we can
> >>>>> use ISO
> >>>>>>>>>>>> weeks:
> >>>>>>>>>>>>
> >>>>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature
> >>>>> freeze
> >>>>>>>>>>>> +3 weeks, and code freeze +2 weeks.
> >>>>>>>>>>>>
> >>>>>>>>>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer
> >>>>> for most
> >>>>>>>>>>>> years:
> >>>>>>>>>>>>
> >>>>>>>>>>>>>
> >>>>>
> >>>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> >>>>>>>>>>>>>     │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │
> >>>>> Buffer                 │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2027 │ Feb 6  │ Mar 3                      │ 25 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2028 │ Jan 26 │ Mar 1                      │ 34 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2029 │ Feb 13 │ Feb 28                     │ 15 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2030 │ Feb 3  │ Feb 27                     │ 24 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2031 │ Jan 23 │ Feb 26                     │ 34 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2032 │ Feb 11 │ Feb 25                     │ 14 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2033 │ Jan 31 │ Mar 2                      │ 30 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ←
> >>>>> tightest with realistic LNY │
> >>>>>>>>>>>>>
> >>>>>
> >>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>>>>     │ 2035 │ Feb 8  │ Feb 28                     │ 20 days
> >>>>>                          │
> >>>>>>>>>>>>>
> >>>>>
> >>>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> >>>>>>>>>>>>
> >>>>>>>>>>>> Guess, we could also make it week-8 instead of 9, and accept
> >>>>> that 2034
> >>>>>>>>>>>> would cut it too short (or be even more flexible and keep week-9
> >>>>> for
> >>>>>>>>>>>> 2034 as an exception to the rule), but that's a single year...
> >>>>> for 2029
> >>>>>>>>>>>> and 2032 we still have one week buffer.
> >>>>>>>>>>>>
> >>>>>>>>>>>> For Christmas, we would still ensure that the release can go out
> >>>>> on
> >>>>>>>>>>>> time, too. The below Q3 target release date is 4 weeks after code
> >>>>>>>>>>>> freeze. This is a little bit more tight, as we often need more
> >>>>> than 4
> >>>>>>>>>>>> weeks to get it done:
> >>>>>>>>>>>>
> >>>>>>>>>>>>>     ┌──────┬──────────────────────────┬────────────────┐
> >>>>>>>>>>>>>     │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2026 │ Dec 9                    │ 16             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2027 │ Dec 15                   │ 10             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2028 │ Dec 13                   │ 12             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2029 │ Dec 12                   │ 13             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2030 │ Dec 11                   │ 14             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2031 │ Dec 10                   │ 15             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2032 │ Dec 8                    │ 17             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2033 │ Dec 14                   │ 11             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2034 │ Dec 13                   │ 12             │
> >>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>>>>     │ 2035 │ Dec 12                   │ 13             │
> >>>>>>>>>>>>>     └──────┴──────────────────────────┴────────────────┘
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> For some years, we would still cut it a little bit tight, but we
> >>>>> could
> >>>>>>>>>>>> also apply an exception to the rule, to compensate for it. If we
> >>>>> have
> >>>>>>>>>>>> enough buffer room from LYN, we can start the Q1 release train a
> >>>>> week
> >>>>>>>>>>>> earlier, reducing the Q3->Q1 timeline by one week.
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> Curious to hear what people think about it. In the end, if we
> >>>>> can agree
> >>>>>>>>>>>> on high level rules like "start Q1 release with X buffer after
> >>>>> LYN", and
> >>>>>>>>>>>> "complete Q3 release with Y buffer before Christmas", and still
> >>>>> aim for
> >>>>>>>>>>>> an as balance week-allocation per release, we can just put
> >>>>> together a
> >>>>>>>>>>>> release plan for the next 10 years and be done with it.
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> -Matthias
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
> >>>>>>>>>>>>> Hello,
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's
> >>>>> release
> >>>>>>>>>>>>> coincided with a reasonable amount of holidays, so I would
> >>>>> favour a
> >>>>>>>>>>>>> 3+2 schedule with releases in March, July and November.
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> Best,
> >>>>>>>>>>>>> Christo
> >>>>>>>>>>>>>
> >>>>>>>>>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com>
> >>>>> wrote:
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Hi,
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <
> >>>>> schofielaj@apache.org>
> >>>>>>>>>>>>>> wrote:
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>>> I would go for 3 releases of equal lengths in March, July and
> >>>>> November. I
> >>>>>>>>>>>>>>> favour the 3+2 schedule.
> >>>>>>>>>>>>>>>
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> +1
> >>>>>>>>>>>>>>
> >>>>>>>>>>>>>> Thanks,
> >>>>>>>>>>>>>> Ismael
> >>>>>>>>>>>>
> >>>>>>>>>>>>
> >>>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>
> >>>
> >>>
> >>
> 
> 

Re: [DISCUSS] Tightening time-based releases

Posted by "Matthias J. Sax" <mj...@apache.org>.
Thanks Andrew.

I see that you did keep Jan/13 for KIP freeze, and did not pull it 
forward. LGTM.


> but I think it's very hard to be concrete about 4.6.0 at this point.

Is it? In the end, if we say we want to follow the long-term proposed 
timeline in general, we are looking into the following dates:

Proposed KIP-freeze dates for 2027

Release-1: March/3 (week 9)
Relasee-2: June/23 (week 25)
Release-3: Oct/13 (week 41)

So there is two options:

  (1) We keep March/3, and do 4 release in 2027. Personally, I don't 
think this make a lot of sense. With the Jan/13 KIP freeze for AK 4.5 
the cycle for AK 4.6 would be way too short IMHO.

  (2) We treat the AK 4.5 release as release-1, ie, pull forward the 
March/13 date to Jan/13, and follow with release-2 in June/23 as 
planned. -- Here we would still have some wiggle room: release-1 is 7 
weeks earlier than the proposed schedule, so it would make sense to pull 
forward release-2, too. If we keep June/23 for release-2, we would have 
23 weeks for AK 4.6 release, and only 16 weeks for release-3. It might 
be better to even it out and move release-2 forward by 3 week to have 20 
weeks for AK 4.6, and 19 weeks for AK 4.7 with Oct/13 KIP freeze date as 
originally proposed.


I would propose to go with option-2 and doing AK 4.6 with KIP freeze 
Jun/2 (ie, 3 weeks earlier than the proposed long-term schedule).


-Matthias


On 10/12/26 3:17 AM, Andrew Schofield wrote:
> Hi,
> I have gone ahead with the 4.5.0 schedule as listed, but I think it's very
> hard to be concrete about 4.6.0 at this point.
> 
> Thanks,
> Andrew
> 
> On 2026/10/10 20:33:35 Muralidhar Basani via dev wrote:
>> Hi all,
>>
>> I'm not deeply involved in release management, so please take this as just
>> some data in case it's useful for the 4.5/4.6 discussion.
>>
>> I compared the planned dates on the wiki release plans with what actually
>> happened in git for the last five releases:
>>
>> Version  Planned Feat Freeze  Branch cut  Planned target   Released
>>   Late   RCs
>> 4.0.0      2024-12-11                  2024-12-16  2025-01-29
>>   2025-03-18     48d    5
>> 4.1.0      2025-06-04                  2025-06-10  2025-07-09
>>   2025-09-02     55d    4
>> 4.2.0      2025-11-19                  2025-11-21  2026-01-07
>>   2026-02-16     40d    5
>> 4.3.0      2026-03-18                 2026-03-18  2026-05-06
>>   2026-05-20     14d    3
>> 4.4.0      2026-07-29                 2026-07-27  2026-09-09
>>   2026-10-12*    33d    5
>> (* expected)
>>
>> Feature freeze is usually on time; the time between code freeze and the
>> release is where we tend to need more than planned (median ~40 days after
>> the target). With the current 4.5.0 plan, that would put the 4.5.0 release
>> around late April 2027, which may be worth keeping in mind when choosing
>> the 4.6.0 dates.
>>
>> I put the small script in this PR, in case it's helpful for future
>> planning: https://github.com/apache/kafka/pull/23769
>> Added sample runs in the PR.
>> If this is not the direction we want, happy to close it.
>>
>>
>> Here is a short snippet from the scripts run
>>
>> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>
>> % python3 release/schedule.py
>> ....
>> ....
>> version  KIP freeze    derived from          feature freeze  code freeze
>>   first RC        target release       realistic release
>> -------     ----------         -----------------         --------------
>>       -----------       ----------        --------------
>> -----------------
>> 4.5.0     2027-01-13   wiki release plan   2027-02-03      2027-02-17
>> 2027-03-16  2027-03-17          2027-04-28
>> 4.6.0     2027-05-19   history                   2027-06-09      2027-06-23
>>    2027-07-20  2027-07-21          2027-09-01
>> 4.7.0     2027-09-22   history                   2027-10-13      2027-10-27
>>    2027-11-23  2027-11-24           2028-01-05
>> ...
>>
>> %python3 release/schedule.py --release-date 4.4.0=2026-10-12 status 4.6.0
>> ....
>> ....
>> Release 4.6.0 as of 2026-10-10 (dates: history)
>> KIP freeze         2027-05-19  (in 221 days)
>> feature freeze     2027-06-09  (in 242 days)
>> code freeze        2027-06-23  (in 256 days)
>> first RC           2027-07-20  (in 283 days)
>> target release     2027-07-21  (in 284 days)
>> realistic release  2027-09-01  (in 326 days)
>> ...
>> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>>
>>
>> Thanks,
>> Murali
>>
>> On Sat, Oct 10, 2026 at 3:33 PM Matthias J. Sax <mj...@apache.org> wrote:
>>
>>> What's you rational for doing 4.6 in March already? If we so 4.5
>>> early/mid January, it would be a very small release. Given the overhead
>>> releases have, I am wondering if it would be worth doing?
>>>
>>> For 4.5, I am also ok to start a Jan/6 instead of Jan/13 with KIP
>>> freeze. If we do 4.6 in July (instead of March) I don't see a need to
>>> pull it forward by one a week though. But I am fine in any case.
>>>
>>> -Matthias
>>>
>>> On 9/28/26 11:49 PM, Dejan Stojadinović wrote:
>>>>> "Anyway, all I want to say is that the original schedule was hard to
>>>> follow. The codebase is large and CVEs are common these days."
>>>>
>>>> As an occasional contributor I can confirm this.
>>>>   From my POV committers/maintainers are having a challenging task to keep
>>>> up, especially while trying to juggle around with multiple release
>>>> branches/versions (trunk/4.5.0, 4.4.0, 4.3.2, 4.2.2).
>>>>
>>>> Best regards,
>>>> Dejan S.
>>>> https://github.com/dejan2609
>>>>
>>>>
>>>> On Tue, Sep 29, 2026 at 7:55 AM Chia-Ping Tsai <ch...@apache.org>
>>> wrote:
>>>>
>>>>> Hi Andrew,
>>>>>
>>>>> +1 to the revised 4.5.0 schedule. I would not pull it a week earlier,
>>>>> since that puts the KIP freeze right after the holidays.
>>>>>
>>>>> Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0
>>>>> reaches code freeze. We had too many must-have changes in 4.4.0, which
>>>>> caused a serious delay. I try to review all backports but my tokens are
>>>>> limited :)
>>>>>
>>>>> Anyway, all I want to say is that the original schedule was hard to
>>>>> follow. The codebase is large and CVEs are common these days.
>>>>>
>>>>> Best,
>>>>> Chia-Ping
>>>>>
>>>>>
>>>>> On 2026/09/25 08:36:23 Andrew Schofield wrote:
>>>>>> Hi,
>>>>>> I'm starting to look at the release schedule for 4.5.0.
>>>>>>
>>>>>> In Chia-Ping's table from 29 May 2026, we have:
>>>>>>
>>>>>> Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
>>>>>> 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19,
>>> 2026
>>>>>> 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
>>>>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
>>>>>>
>>>>>> In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP
>>>>> freeze of Jan 13, 2027 which gives this.
>>>>>>
>>>>>> 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon
>>> now
>>>>>> 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17,
>>> 2027
>>>>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
>>>>>>
>>>>>> I think we could even pull 4.5.0 a week earlier than that, but I don't
>>>>> think there's anything fundamentally wrong with a short 4.6.0 provided
>>>>> people know in advance.
>>>>>>
>>>>>> Are people happy with the 4.5.0 (revised) dates above? Would a week
>>>>> earlier be better given that it gives an extra week to 4.6.0 which is
>>>>> looking very squeezed?
>>>>>>
>>>>>> Thanks,
>>>>>> Andrew
>>>>>>
>>>>>> On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
>>>>>>> Hi all,
>>>>>>>
>>>>>>> +1 to the 4.5 schedule, given the delay of 4.4.0.
>>>>>>>
>>>>>>> I think having 4.5 land right on Lunar New Year should be fine this
>>>>> time, since "too close" is the real issue here.
>>>>>>>
>>>>>>> Best,
>>>>>>> Chia-Ping
>>>>>>>
>>>>>>>> Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
>>>>>>>>
>>>>>>>> I don't think we have for 4.5... I believe that there was a slight,
>>>>> high level agreement for the long term, that we want to follow the
>>>>> "calendar week proposal"?
>>>>>>>>
>>>>>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with
>>>>> feature freeze
>>>>>>>>>>>>> +3 weeks, and code freeze +2 weeks.
>>>>>>>>
>>>>>>>> AK 4.4 KIP freeze was Jul/8.
>>>>>>>>
>>>>>>>> For 2026, week 41 would be Oct/7 which is way too close IMHO.
>>>>>>>> For 2027, week 9 would be Mar/3 which seems to be too far out?
>>>>>>>>
>>>>>>>> Second release (week 25) for 2027 would be Jun/23, which would be
>>>>> 4.6?
>>>>>>>>
>>>>>>>> So the question remains, where to put 4.5 so we can start to follow
>>>>> the new schedule starting with 4.6. It seems KIP freeze between Jan 13
>>>>> might be a good compromise putting it roughly between 4.4 and 4.6 KIP
>>>>> freeze dates. Of course, the release would overlap with Chinese New
>>> Year,
>>>>> but it would be the last time we have this overlap.
>>>>>>>>
>>>>>>>> Just my 2tc.
>>>>>>>>
>>>>>>>>
>>>>>>>> -Matthias
>>>>>>>>
>>>>>>>>> On 9/15/26 8:17 AM, David Jacot wrote:
>>>>>>>>> Hi all,
>>>>>>>>> Have we converged on the schedule that we want to use yet? Based on
>>>>>>>>> the last emails, it is not so clear to me. For your context, I am
>>>>>>>>> trying to figure out the timeline for 4.5.
>>>>>>>>> Best,
>>>>>>>>> David
>>>>>>>>>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <
>>>>> christololov@gmail.com> wrote:
>>>>>>>>>>
>>>>>>>>>> I think we could be a bit more lax with the KIP freeze of 6.7.0
>>>>> when
>>>>>>>>>> 2034 comes around to account for LNY :)
>>>>>>>>>>
>>>>>>>>>> Joking aside, my preference between "KIP freeze would be
>>>>> Wednesday's,
>>>>>>>>>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
>>>>>>>>>> weeks" and "start Q1 release with X buffer after LYN and complete
>>>>> Q3
>>>>>>>>>> release with Y buffer before Christmas" would be to start with the
>>>>>>>>>> first and adjust by shifting by 1 week on individual basis the ones
>>>>>>>>>> which are too close. I am okay with too close being defined as 2
>>>>>>>>>> weeks. Based on your very detailed tables we seem to have only a
>>>>> few
>>>>>>>>>> such outliers in the next decade.
>>>>>>>>>>
>>>>>>>>>> Best,
>>>>>>>>>> Christo
>>>>>>>>>>
>>>>>>>>>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org>
>>>>> wrote:
>>>>>>>>>>>
>>>>>>>>>>> Matthias, you are definitely a master of scheduling! I'm +100 to
>>>>> your plans.
>>>>>>>>>>>
>>>>>>>>>>> Also, it is absolutely helpful to write down the exact release
>>>>> dates for the next few years at once. It will ease our anxiety a lot!
>>> I'd
>>>>> be more than happy to update the wiki with these dates once we all
>>> agree on
>>>>> this new schedule. For example:
>>>>>>>>>>>
>>>>>>>>>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target
>>>>> Release
>>>>>>>>>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19,
>>>>> 2026
>>>>>>>>>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9,
>>>>> 2026
>>>>>>>>>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5,
>>>>> 2027
>>>>>>>>>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25,
>>>>> 2027
>>>>>>>>>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15,
>>>>> 2027
>>>>>>>>>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3,
>>>>> 2028
>>>>>>>>>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23,
>>>>> 2028
>>>>>>>>>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13,
>>>>> 2028
>>>>>>>>>>>
>>>>>>>>>>> Furthermore, let's do our best to avoid fixing release blockers
>>>>> under the Christmas tree :)
>>>>>>>>>>>
>>>>>>>>>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
>>>>>>>>>>>> Thanks for the input.
>>>>>>>>>>>>
>>>>>>>>>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
>>>>>>>>>>>> windows gets a lot of support.
>>>>>>>>>>>>
>>>>>>>>>>>> However, the idea of dividing the release windows not into 3
>>>>> equally
>>>>>>>>>>>> long junks, but be more flexible to account for holiday seasons,
>>>>> seems
>>>>>>>>>>>> not to get a lot of support? Not sure why... I personally think
>>>>> it would
>>>>>>>>>>>> be good to take it into account, and I also think that Chia-Ping
>>>>> raises
>>>>>>>>>>>> a very good point.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising
>>>>> it. I
>>>>>>>>>>>> did actually briefly think about it, but it's difficult, and
>>>>> thus I just
>>>>>>>>>>>> started with something simple as proposal to kick off a
>>>>> discussion.
>>>>>>>>>>>>
>>>>>>>>>>>> I think, Diwali is short enough to not make it a concern
>>>>> (similar to
>>>>>>>>>>>> Thanksgiving)?
>>>>>>>>>>>>
>>>>>>>>>>>> For Lunar New Year, I agree that it would be good to plan for it.
>>>>>>>>>>>> However, it seems it's shifting quite a lot year-over-year what
>>>>> makes it
>>>>>>>>>>>> very difficult (w/o a huge window) to include?
>>>>>>>>>>>>
>>>>>>>>>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced
>>>>> (this
>>>>>>>>>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5
>>>>> month
>>>>>>>>>>>> releases does not sound very appealing?
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> I was playing with it a little bit more. Overall we have 52
>>>>> weeks, so
>>>>>>>>>>>> roughly 17.3 weeks per release. So we can naturally split it into
>>>>>>>>>>>> 17/17/18 using the additional week for the holidays. But that's
>>>>> not good
>>>>>>>>>>>> enough to take both Christmas and Lunar New Year into account.
>>>>> However,
>>>>>>>>>>>> with a 16/16/20, we could make it work I believe, even if it
>>>>> does cut it
>>>>>>>>>>>> tight sometimes... To be able to phrase a "simple rule", we can
>>>>> use ISO
>>>>>>>>>>>> weeks:
>>>>>>>>>>>>
>>>>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature
>>>>> freeze
>>>>>>>>>>>> +3 weeks, and code freeze +2 weeks.
>>>>>>>>>>>>
>>>>>>>>>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer
>>>>> for most
>>>>>>>>>>>> years:
>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>
>>>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
>>>>>>>>>>>>>     │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │
>>>>> Buffer                 │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2027 │ Feb 6  │ Mar 3                      │ 25 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2028 │ Jan 26 │ Mar 1                      │ 34 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2029 │ Feb 13 │ Feb 28                     │ 15 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2030 │ Feb 3  │ Feb 27                     │ 24 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2031 │ Jan 23 │ Feb 26                     │ 34 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2032 │ Feb 11 │ Feb 25                     │ 14 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2033 │ Jan 31 │ Mar 2                      │ 30 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ←
>>>>> tightest with realistic LNY │
>>>>>>>>>>>>>
>>>>>
>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>>>>     │ 2035 │ Feb 8  │ Feb 28                     │ 20 days
>>>>>                          │
>>>>>>>>>>>>>
>>>>>
>>>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
>>>>>>>>>>>>
>>>>>>>>>>>> Guess, we could also make it week-8 instead of 9, and accept
>>>>> that 2034
>>>>>>>>>>>> would cut it too short (or be even more flexible and keep week-9
>>>>> for
>>>>>>>>>>>> 2034 as an exception to the rule), but that's a single year...
>>>>> for 2029
>>>>>>>>>>>> and 2032 we still have one week buffer.
>>>>>>>>>>>>
>>>>>>>>>>>> For Christmas, we would still ensure that the release can go out
>>>>> on
>>>>>>>>>>>> time, too. The below Q3 target release date is 4 weeks after code
>>>>>>>>>>>> freeze. This is a little bit more tight, as we often need more
>>>>> than 4
>>>>>>>>>>>> weeks to get it done:
>>>>>>>>>>>>
>>>>>>>>>>>>>     ┌──────┬──────────────────────────┬────────────────┐
>>>>>>>>>>>>>     │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2026 │ Dec 9                    │ 16             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2027 │ Dec 15                   │ 10             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2028 │ Dec 13                   │ 12             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2029 │ Dec 12                   │ 13             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2030 │ Dec 11                   │ 14             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2031 │ Dec 10                   │ 15             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2032 │ Dec 8                    │ 17             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2033 │ Dec 14                   │ 11             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2034 │ Dec 13                   │ 12             │
>>>>>>>>>>>>>     ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>>>>     │ 2035 │ Dec 12                   │ 13             │
>>>>>>>>>>>>>     └──────┴──────────────────────────┴────────────────┘
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> For some years, we would still cut it a little bit tight, but we
>>>>> could
>>>>>>>>>>>> also apply an exception to the rule, to compensate for it. If we
>>>>> have
>>>>>>>>>>>> enough buffer room from LYN, we can start the Q1 release train a
>>>>> week
>>>>>>>>>>>> earlier, reducing the Q3->Q1 timeline by one week.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> Curious to hear what people think about it. In the end, if we
>>>>> can agree
>>>>>>>>>>>> on high level rules like "start Q1 release with X buffer after
>>>>> LYN", and
>>>>>>>>>>>> "complete Q3 release with Y buffer before Christmas", and still
>>>>> aim for
>>>>>>>>>>>> an as balance week-allocation per release, we can just put
>>>>> together a
>>>>>>>>>>>> release plan for the next 10 years and be done with it.
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> -Matthias
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
>>>>>>>>>>>>> Hello,
>>>>>>>>>>>>>
>>>>>>>>>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's
>>>>> release
>>>>>>>>>>>>> coincided with a reasonable amount of holidays, so I would
>>>>> favour a
>>>>>>>>>>>>> 3+2 schedule with releases in March, July and November.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Best,
>>>>>>>>>>>>> Christo
>>>>>>>>>>>>>
>>>>>>>>>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com>
>>>>> wrote:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Hi,
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <
>>>>> schofielaj@apache.org>
>>>>>>>>>>>>>> wrote:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> I would go for 3 releases of equal lengths in March, July and
>>>>> November. I
>>>>>>>>>>>>>>> favour the 3+2 schedule.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> +1
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Thanks,
>>>>>>>>>>>>>> Ismael
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>
>>>
>>


Re: [DISCUSS] Tightening time-based releases

Posted by Andrew Schofield <sc...@apache.org>.
Hi,
I have gone ahead with the 4.5.0 schedule as listed, but I think it's very
hard to be concrete about 4.6.0 at this point.

Thanks,
Andrew

On 2026/10/10 20:33:35 Muralidhar Basani via dev wrote:
> Hi all,
> 
> I'm not deeply involved in release management, so please take this as just
> some data in case it's useful for the 4.5/4.6 discussion.
> 
> I compared the planned dates on the wiki release plans with what actually
> happened in git for the last five releases:
> 
> Version  Planned Feat Freeze  Branch cut  Planned target   Released
>  Late   RCs
> 4.0.0      2024-12-11                  2024-12-16  2025-01-29
>  2025-03-18     48d    5
> 4.1.0      2025-06-04                  2025-06-10  2025-07-09
>  2025-09-02     55d    4
> 4.2.0      2025-11-19                  2025-11-21  2026-01-07
>  2026-02-16     40d    5
> 4.3.0      2026-03-18                 2026-03-18  2026-05-06
>  2026-05-20     14d    3
> 4.4.0      2026-07-29                 2026-07-27  2026-09-09
>  2026-10-12*    33d    5
> (* expected)
> 
> Feature freeze is usually on time; the time between code freeze and the
> release is where we tend to need more than planned (median ~40 days after
> the target). With the current 4.5.0 plan, that would put the 4.5.0 release
> around late April 2027, which may be worth keeping in mind when choosing
> the 4.6.0 dates.
> 
> I put the small script in this PR, in case it's helpful for future
> planning: https://github.com/apache/kafka/pull/23769
> Added sample runs in the PR.
> If this is not the direction we want, happy to close it.
> 
> 
> Here is a short snippet from the scripts run
> 
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> % python3 release/schedule.py
> ....
> ....
> version  KIP freeze    derived from          feature freeze  code freeze
>  first RC        target release       realistic release
> -------     ----------         -----------------         --------------
>      -----------       ----------        --------------
> -----------------
> 4.5.0     2027-01-13   wiki release plan   2027-02-03      2027-02-17
> 2027-03-16  2027-03-17          2027-04-28
> 4.6.0     2027-05-19   history                   2027-06-09      2027-06-23
>   2027-07-20  2027-07-21          2027-09-01
> 4.7.0     2027-09-22   history                   2027-10-13      2027-10-27
>   2027-11-23  2027-11-24           2028-01-05
> ...
> 
> %python3 release/schedule.py --release-date 4.4.0=2026-10-12 status 4.6.0
> ....
> ....
> Release 4.6.0 as of 2026-10-10 (dates: history)
> KIP freeze         2027-05-19  (in 221 days)
> feature freeze     2027-06-09  (in 242 days)
> code freeze        2027-06-23  (in 256 days)
> first RC           2027-07-20  (in 283 days)
> target release     2027-07-21  (in 284 days)
> realistic release  2027-09-01  (in 326 days)
> ...
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> 
> Thanks,
> Murali
> 
> On Sat, Oct 10, 2026 at 3:33 PM Matthias J. Sax <mj...@apache.org> wrote:
> 
> > What's you rational for doing 4.6 in March already? If we so 4.5
> > early/mid January, it would be a very small release. Given the overhead
> > releases have, I am wondering if it would be worth doing?
> >
> > For 4.5, I am also ok to start a Jan/6 instead of Jan/13 with KIP
> > freeze. If we do 4.6 in July (instead of March) I don't see a need to
> > pull it forward by one a week though. But I am fine in any case.
> >
> > -Matthias
> >
> > On 9/28/26 11:49 PM, Dejan Stojadinović wrote:
> > >> "Anyway, all I want to say is that the original schedule was hard to
> > > follow. The codebase is large and CVEs are common these days."
> > >
> > > As an occasional contributor I can confirm this.
> > >  From my POV committers/maintainers are having a challenging task to keep
> > > up, especially while trying to juggle around with multiple release
> > > branches/versions (trunk/4.5.0, 4.4.0, 4.3.2, 4.2.2).
> > >
> > > Best regards,
> > > Dejan S.
> > > https://github.com/dejan2609
> > >
> > >
> > > On Tue, Sep 29, 2026 at 7:55 AM Chia-Ping Tsai <ch...@apache.org>
> > wrote:
> > >
> > >> Hi Andrew,
> > >>
> > >> +1 to the revised 4.5.0 schedule. I would not pull it a week earlier,
> > >> since that puts the KIP freeze right after the holidays.
> > >>
> > >> Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0
> > >> reaches code freeze. We had too many must-have changes in 4.4.0, which
> > >> caused a serious delay. I try to review all backports but my tokens are
> > >> limited :)
> > >>
> > >> Anyway, all I want to say is that the original schedule was hard to
> > >> follow. The codebase is large and CVEs are common these days.
> > >>
> > >> Best,
> > >> Chia-Ping
> > >>
> > >>
> > >> On 2026/09/25 08:36:23 Andrew Schofield wrote:
> > >>> Hi,
> > >>> I'm starting to look at the release schedule for 4.5.0.
> > >>>
> > >>> In Chia-Ping's table from 29 May 2026, we have:
> > >>>
> > >>> Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
> > >>> 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19,
> > 2026
> > >>> 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
> > >>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> > >>>
> > >>> In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP
> > >> freeze of Jan 13, 2027 which gives this.
> > >>>
> > >>> 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon
> > now
> > >>> 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17,
> > 2027
> > >>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> > >>>
> > >>> I think we could even pull 4.5.0 a week earlier than that, but I don't
> > >> think there's anything fundamentally wrong with a short 4.6.0 provided
> > >> people know in advance.
> > >>>
> > >>> Are people happy with the 4.5.0 (revised) dates above? Would a week
> > >> earlier be better given that it gives an extra week to 4.6.0 which is
> > >> looking very squeezed?
> > >>>
> > >>> Thanks,
> > >>> Andrew
> > >>>
> > >>> On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
> > >>>> Hi all,
> > >>>>
> > >>>> +1 to the 4.5 schedule, given the delay of 4.4.0.
> > >>>>
> > >>>> I think having 4.5 land right on Lunar New Year should be fine this
> > >> time, since "too close" is the real issue here.
> > >>>>
> > >>>> Best,
> > >>>> Chia-Ping
> > >>>>
> > >>>>> Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> > >>>>>
> > >>>>> I don't think we have for 4.5... I believe that there was a slight,
> > >> high level agreement for the long term, that we want to follow the
> > >> "calendar week proposal"?
> > >>>>>
> > >>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with
> > >> feature freeze
> > >>>>>>>>>> +3 weeks, and code freeze +2 weeks.
> > >>>>>
> > >>>>> AK 4.4 KIP freeze was Jul/8.
> > >>>>>
> > >>>>> For 2026, week 41 would be Oct/7 which is way too close IMHO.
> > >>>>> For 2027, week 9 would be Mar/3 which seems to be too far out?
> > >>>>>
> > >>>>> Second release (week 25) for 2027 would be Jun/23, which would be
> > >> 4.6?
> > >>>>>
> > >>>>> So the question remains, where to put 4.5 so we can start to follow
> > >> the new schedule starting with 4.6. It seems KIP freeze between Jan 13
> > >> might be a good compromise putting it roughly between 4.4 and 4.6 KIP
> > >> freeze dates. Of course, the release would overlap with Chinese New
> > Year,
> > >> but it would be the last time we have this overlap.
> > >>>>>
> > >>>>> Just my 2tc.
> > >>>>>
> > >>>>>
> > >>>>> -Matthias
> > >>>>>
> > >>>>>> On 9/15/26 8:17 AM, David Jacot wrote:
> > >>>>>> Hi all,
> > >>>>>> Have we converged on the schedule that we want to use yet? Based on
> > >>>>>> the last emails, it is not so clear to me. For your context, I am
> > >>>>>> trying to figure out the timeline for 4.5.
> > >>>>>> Best,
> > >>>>>> David
> > >>>>>>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <
> > >> christololov@gmail.com> wrote:
> > >>>>>>>
> > >>>>>>> I think we could be a bit more lax with the KIP freeze of 6.7.0
> > >> when
> > >>>>>>> 2034 comes around to account for LNY :)
> > >>>>>>>
> > >>>>>>> Joking aside, my preference between "KIP freeze would be
> > >> Wednesday's,
> > >>>>>>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> > >>>>>>> weeks" and "start Q1 release with X buffer after LYN and complete
> > >> Q3
> > >>>>>>> release with Y buffer before Christmas" would be to start with the
> > >>>>>>> first and adjust by shifting by 1 week on individual basis the ones
> > >>>>>>> which are too close. I am okay with too close being defined as 2
> > >>>>>>> weeks. Based on your very detailed tables we seem to have only a
> > >> few
> > >>>>>>> such outliers in the next decade.
> > >>>>>>>
> > >>>>>>> Best,
> > >>>>>>> Christo
> > >>>>>>>
> > >>>>>>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org>
> > >> wrote:
> > >>>>>>>>
> > >>>>>>>> Matthias, you are definitely a master of scheduling! I'm +100 to
> > >> your plans.
> > >>>>>>>>
> > >>>>>>>> Also, it is absolutely helpful to write down the exact release
> > >> dates for the next few years at once. It will ease our anxiety a lot!
> > I'd
> > >> be more than happy to update the wiki with these dates once we all
> > agree on
> > >> this new schedule. For example:
> > >>>>>>>>
> > >>>>>>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target
> > >> Release
> > >>>>>>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19,
> > >> 2026
> > >>>>>>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9,
> > >> 2026
> > >>>>>>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5,
> > >> 2027
> > >>>>>>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25,
> > >> 2027
> > >>>>>>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15,
> > >> 2027
> > >>>>>>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3,
> > >> 2028
> > >>>>>>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23,
> > >> 2028
> > >>>>>>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13,
> > >> 2028
> > >>>>>>>>
> > >>>>>>>> Furthermore, let's do our best to avoid fixing release blockers
> > >> under the Christmas tree :)
> > >>>>>>>>
> > >>>>>>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> > >>>>>>>>> Thanks for the input.
> > >>>>>>>>>
> > >>>>>>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
> > >>>>>>>>> windows gets a lot of support.
> > >>>>>>>>>
> > >>>>>>>>> However, the idea of dividing the release windows not into 3
> > >> equally
> > >>>>>>>>> long junks, but be more flexible to account for holiday seasons,
> > >> seems
> > >>>>>>>>> not to get a lot of support? Not sure why... I personally think
> > >> it would
> > >>>>>>>>> be good to take it into account, and I also think that Chia-Ping
> > >> raises
> > >>>>>>>>> a very good point.
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising
> > >> it. I
> > >>>>>>>>> did actually briefly think about it, but it's difficult, and
> > >> thus I just
> > >>>>>>>>> started with something simple as proposal to kick off a
> > >> discussion.
> > >>>>>>>>>
> > >>>>>>>>> I think, Diwali is short enough to not make it a concern
> > >> (similar to
> > >>>>>>>>> Thanksgiving)?
> > >>>>>>>>>
> > >>>>>>>>> For Lunar New Year, I agree that it would be good to plan for it.
> > >>>>>>>>> However, it seems it's shifting quite a lot year-over-year what
> > >> makes it
> > >>>>>>>>> very difficult (w/o a huge window) to include?
> > >>>>>>>>>
> > >>>>>>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced
> > >> (this
> > >>>>>>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5
> > >> month
> > >>>>>>>>> releases does not sound very appealing?
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>> I was playing with it a little bit more. Overall we have 52
> > >> weeks, so
> > >>>>>>>>> roughly 17.3 weeks per release. So we can naturally split it into
> > >>>>>>>>> 17/17/18 using the additional week for the holidays. But that's
> > >> not good
> > >>>>>>>>> enough to take both Christmas and Lunar New Year into account.
> > >> However,
> > >>>>>>>>> with a 16/16/20, we could make it work I believe, even if it
> > >> does cut it
> > >>>>>>>>> tight sometimes... To be able to phrase a "simple rule", we can
> > >> use ISO
> > >>>>>>>>> weeks:
> > >>>>>>>>>
> > >>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature
> > >> freeze
> > >>>>>>>>> +3 weeks, and code freeze +2 weeks.
> > >>>>>>>>>
> > >>>>>>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer
> > >> for most
> > >>>>>>>>> years:
> > >>>>>>>>>
> > >>>>>>>>>>
> > >>
> >  ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> > >>>>>>>>>>    │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │
> > >> Buffer                 │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2027 │ Feb 6  │ Mar 3                      │ 25 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2028 │ Jan 26 │ Mar 1                      │ 34 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2029 │ Feb 13 │ Feb 28                     │ 15 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2030 │ Feb 3  │ Feb 27                     │ 24 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2031 │ Jan 23 │ Feb 26                     │ 34 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2032 │ Feb 11 │ Feb 25                     │ 14 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2033 │ Jan 31 │ Mar 2                      │ 30 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ←
> > >> tightest with realistic LNY │
> > >>>>>>>>>>
> > >>
> >  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>>>>>    │ 2035 │ Feb 8  │ Feb 28                     │ 20 days
> > >>                         │
> > >>>>>>>>>>
> > >>
> >  └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> > >>>>>>>>>
> > >>>>>>>>> Guess, we could also make it week-8 instead of 9, and accept
> > >> that 2034
> > >>>>>>>>> would cut it too short (or be even more flexible and keep week-9
> > >> for
> > >>>>>>>>> 2034 as an exception to the rule), but that's a single year...
> > >> for 2029
> > >>>>>>>>> and 2032 we still have one week buffer.
> > >>>>>>>>>
> > >>>>>>>>> For Christmas, we would still ensure that the release can go out
> > >> on
> > >>>>>>>>> time, too. The below Q3 target release date is 4 weeks after code
> > >>>>>>>>> freeze. This is a little bit more tight, as we often need more
> > >> than 4
> > >>>>>>>>> weeks to get it done:
> > >>>>>>>>>
> > >>>>>>>>>>    ┌──────┬──────────────────────────┬────────────────┐
> > >>>>>>>>>>    │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2026 │ Dec 9                    │ 16             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2027 │ Dec 15                   │ 10             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2028 │ Dec 13                   │ 12             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2029 │ Dec 12                   │ 13             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2030 │ Dec 11                   │ 14             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2031 │ Dec 10                   │ 15             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2032 │ Dec 8                    │ 17             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2033 │ Dec 14                   │ 11             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2034 │ Dec 13                   │ 12             │
> > >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>>>>>    │ 2035 │ Dec 12                   │ 13             │
> > >>>>>>>>>>    └──────┴──────────────────────────┴────────────────┘
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>> For some years, we would still cut it a little bit tight, but we
> > >> could
> > >>>>>>>>> also apply an exception to the rule, to compensate for it. If we
> > >> have
> > >>>>>>>>> enough buffer room from LYN, we can start the Q1 release train a
> > >> week
> > >>>>>>>>> earlier, reducing the Q3->Q1 timeline by one week.
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>> Curious to hear what people think about it. In the end, if we
> > >> can agree
> > >>>>>>>>> on high level rules like "start Q1 release with X buffer after
> > >> LYN", and
> > >>>>>>>>> "complete Q3 release with Y buffer before Christmas", and still
> > >> aim for
> > >>>>>>>>> an as balance week-allocation per release, we can just put
> > >> together a
> > >>>>>>>>> release plan for the next 10 years and be done with it.
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>> -Matthias
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
> > >>>>>>>>>> Hello,
> > >>>>>>>>>>
> > >>>>>>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's
> > >> release
> > >>>>>>>>>> coincided with a reasonable amount of holidays, so I would
> > >> favour a
> > >>>>>>>>>> 3+2 schedule with releases in March, July and November.
> > >>>>>>>>>>
> > >>>>>>>>>> Best,
> > >>>>>>>>>> Christo
> > >>>>>>>>>>
> > >>>>>>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com>
> > >> wrote:
> > >>>>>>>>>>>
> > >>>>>>>>>>> Hi,
> > >>>>>>>>>>>
> > >>>>>>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <
> > >> schofielaj@apache.org>
> > >>>>>>>>>>> wrote:
> > >>>>>>>>>>>
> > >>>>>>>>>>>> I would go for 3 releases of equal lengths in March, July and
> > >> November. I
> > >>>>>>>>>>>> favour the 3+2 schedule.
> > >>>>>>>>>>>>
> > >>>>>>>>>>>
> > >>>>>>>>>>> +1
> > >>>>>>>>>>>
> > >>>>>>>>>>> Thanks,
> > >>>>>>>>>>> Ismael
> > >>>>>>>>>
> > >>>>>>>>>
> > >>>>>
> > >>>>
> > >>>
> > >>
> >
> >
> 

Re: [DISCUSS] Tightening time-based releases

Posted by Muralidhar Basani via dev <de...@kafka.apache.org>.
Hi all,

I'm not deeply involved in release management, so please take this as just
some data in case it's useful for the 4.5/4.6 discussion.

I compared the planned dates on the wiki release plans with what actually
happened in git for the last five releases:

Version  Planned Feat Freeze  Branch cut  Planned target   Released
 Late   RCs
4.0.0      2024-12-11                  2024-12-16  2025-01-29
 2025-03-18     48d    5
4.1.0      2025-06-04                  2025-06-10  2025-07-09
 2025-09-02     55d    4
4.2.0      2025-11-19                  2025-11-21  2026-01-07
 2026-02-16     40d    5
4.3.0      2026-03-18                 2026-03-18  2026-05-06
 2026-05-20     14d    3
4.4.0      2026-07-29                 2026-07-27  2026-09-09
 2026-10-12*    33d    5
(* expected)

Feature freeze is usually on time; the time between code freeze and the
release is where we tend to need more than planned (median ~40 days after
the target). With the current 4.5.0 plan, that would put the 4.5.0 release
around late April 2027, which may be worth keeping in mind when choosing
the 4.6.0 dates.

I put the small script in this PR, in case it's helpful for future
planning: https://github.com/apache/kafka/pull/23769
Added sample runs in the PR.
If this is not the direction we want, happy to close it.


Here is a short snippet from the scripts run

++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

% python3 release/schedule.py
....
....
version  KIP freeze    derived from          feature freeze  code freeze
 first RC        target release       realistic release
-------     ----------         -----------------         --------------
     -----------       ----------        --------------
-----------------
4.5.0     2027-01-13   wiki release plan   2027-02-03      2027-02-17
2027-03-16  2027-03-17          2027-04-28
4.6.0     2027-05-19   history                   2027-06-09      2027-06-23
  2027-07-20  2027-07-21          2027-09-01
4.7.0     2027-09-22   history                   2027-10-13      2027-10-27
  2027-11-23  2027-11-24           2028-01-05
...

%python3 release/schedule.py --release-date 4.4.0=2026-10-12 status 4.6.0
....
....
Release 4.6.0 as of 2026-10-10 (dates: history)
KIP freeze         2027-05-19  (in 221 days)
feature freeze     2027-06-09  (in 242 days)
code freeze        2027-06-23  (in 256 days)
first RC           2027-07-20  (in 283 days)
target release     2027-07-21  (in 284 days)
realistic release  2027-09-01  (in 326 days)
...
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++


Thanks,
Murali

On Sat, Oct 10, 2026 at 3:33 PM Matthias J. Sax <mj...@apache.org> wrote:

> What's you rational for doing 4.6 in March already? If we so 4.5
> early/mid January, it would be a very small release. Given the overhead
> releases have, I am wondering if it would be worth doing?
>
> For 4.5, I am also ok to start a Jan/6 instead of Jan/13 with KIP
> freeze. If we do 4.6 in July (instead of March) I don't see a need to
> pull it forward by one a week though. But I am fine in any case.
>
> -Matthias
>
> On 9/28/26 11:49 PM, Dejan Stojadinović wrote:
> >> "Anyway, all I want to say is that the original schedule was hard to
> > follow. The codebase is large and CVEs are common these days."
> >
> > As an occasional contributor I can confirm this.
> >  From my POV committers/maintainers are having a challenging task to keep
> > up, especially while trying to juggle around with multiple release
> > branches/versions (trunk/4.5.0, 4.4.0, 4.3.2, 4.2.2).
> >
> > Best regards,
> > Dejan S.
> > https://github.com/dejan2609
> >
> >
> > On Tue, Sep 29, 2026 at 7:55 AM Chia-Ping Tsai <ch...@apache.org>
> wrote:
> >
> >> Hi Andrew,
> >>
> >> +1 to the revised 4.5.0 schedule. I would not pull it a week earlier,
> >> since that puts the KIP freeze right after the holidays.
> >>
> >> Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0
> >> reaches code freeze. We had too many must-have changes in 4.4.0, which
> >> caused a serious delay. I try to review all backports but my tokens are
> >> limited :)
> >>
> >> Anyway, all I want to say is that the original schedule was hard to
> >> follow. The codebase is large and CVEs are common these days.
> >>
> >> Best,
> >> Chia-Ping
> >>
> >>
> >> On 2026/09/25 08:36:23 Andrew Schofield wrote:
> >>> Hi,
> >>> I'm starting to look at the release schedule for 4.5.0.
> >>>
> >>> In Chia-Ping's table from 29 May 2026, we have:
> >>>
> >>> Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
> >>> 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19,
> 2026
> >>> 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
> >>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> >>>
> >>> In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP
> >> freeze of Jan 13, 2027 which gives this.
> >>>
> >>> 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon
> now
> >>> 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17,
> 2027
> >>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> >>>
> >>> I think we could even pull 4.5.0 a week earlier than that, but I don't
> >> think there's anything fundamentally wrong with a short 4.6.0 provided
> >> people know in advance.
> >>>
> >>> Are people happy with the 4.5.0 (revised) dates above? Would a week
> >> earlier be better given that it gives an extra week to 4.6.0 which is
> >> looking very squeezed?
> >>>
> >>> Thanks,
> >>> Andrew
> >>>
> >>> On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
> >>>> Hi all,
> >>>>
> >>>> +1 to the 4.5 schedule, given the delay of 4.4.0.
> >>>>
> >>>> I think having 4.5 land right on Lunar New Year should be fine this
> >> time, since "too close" is the real issue here.
> >>>>
> >>>> Best,
> >>>> Chia-Ping
> >>>>
> >>>>> Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> >>>>>
> >>>>> I don't think we have for 4.5... I believe that there was a slight,
> >> high level agreement for the long term, that we want to follow the
> >> "calendar week proposal"?
> >>>>>
> >>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with
> >> feature freeze
> >>>>>>>>>> +3 weeks, and code freeze +2 weeks.
> >>>>>
> >>>>> AK 4.4 KIP freeze was Jul/8.
> >>>>>
> >>>>> For 2026, week 41 would be Oct/7 which is way too close IMHO.
> >>>>> For 2027, week 9 would be Mar/3 which seems to be too far out?
> >>>>>
> >>>>> Second release (week 25) for 2027 would be Jun/23, which would be
> >> 4.6?
> >>>>>
> >>>>> So the question remains, where to put 4.5 so we can start to follow
> >> the new schedule starting with 4.6. It seems KIP freeze between Jan 13
> >> might be a good compromise putting it roughly between 4.4 and 4.6 KIP
> >> freeze dates. Of course, the release would overlap with Chinese New
> Year,
> >> but it would be the last time we have this overlap.
> >>>>>
> >>>>> Just my 2tc.
> >>>>>
> >>>>>
> >>>>> -Matthias
> >>>>>
> >>>>>> On 9/15/26 8:17 AM, David Jacot wrote:
> >>>>>> Hi all,
> >>>>>> Have we converged on the schedule that we want to use yet? Based on
> >>>>>> the last emails, it is not so clear to me. For your context, I am
> >>>>>> trying to figure out the timeline for 4.5.
> >>>>>> Best,
> >>>>>> David
> >>>>>>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <
> >> christololov@gmail.com> wrote:
> >>>>>>>
> >>>>>>> I think we could be a bit more lax with the KIP freeze of 6.7.0
> >> when
> >>>>>>> 2034 comes around to account for LNY :)
> >>>>>>>
> >>>>>>> Joking aside, my preference between "KIP freeze would be
> >> Wednesday's,
> >>>>>>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> >>>>>>> weeks" and "start Q1 release with X buffer after LYN and complete
> >> Q3
> >>>>>>> release with Y buffer before Christmas" would be to start with the
> >>>>>>> first and adjust by shifting by 1 week on individual basis the ones
> >>>>>>> which are too close. I am okay with too close being defined as 2
> >>>>>>> weeks. Based on your very detailed tables we seem to have only a
> >> few
> >>>>>>> such outliers in the next decade.
> >>>>>>>
> >>>>>>> Best,
> >>>>>>> Christo
> >>>>>>>
> >>>>>>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org>
> >> wrote:
> >>>>>>>>
> >>>>>>>> Matthias, you are definitely a master of scheduling! I'm +100 to
> >> your plans.
> >>>>>>>>
> >>>>>>>> Also, it is absolutely helpful to write down the exact release
> >> dates for the next few years at once. It will ease our anxiety a lot!
> I'd
> >> be more than happy to update the wiki with these dates once we all
> agree on
> >> this new schedule. For example:
> >>>>>>>>
> >>>>>>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target
> >> Release
> >>>>>>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19,
> >> 2026
> >>>>>>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9,
> >> 2026
> >>>>>>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5,
> >> 2027
> >>>>>>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25,
> >> 2027
> >>>>>>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15,
> >> 2027
> >>>>>>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3,
> >> 2028
> >>>>>>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23,
> >> 2028
> >>>>>>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13,
> >> 2028
> >>>>>>>>
> >>>>>>>> Furthermore, let's do our best to avoid fixing release blockers
> >> under the Christmas tree :)
> >>>>>>>>
> >>>>>>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> >>>>>>>>> Thanks for the input.
> >>>>>>>>>
> >>>>>>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
> >>>>>>>>> windows gets a lot of support.
> >>>>>>>>>
> >>>>>>>>> However, the idea of dividing the release windows not into 3
> >> equally
> >>>>>>>>> long junks, but be more flexible to account for holiday seasons,
> >> seems
> >>>>>>>>> not to get a lot of support? Not sure why... I personally think
> >> it would
> >>>>>>>>> be good to take it into account, and I also think that Chia-Ping
> >> raises
> >>>>>>>>> a very good point.
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising
> >> it. I
> >>>>>>>>> did actually briefly think about it, but it's difficult, and
> >> thus I just
> >>>>>>>>> started with something simple as proposal to kick off a
> >> discussion.
> >>>>>>>>>
> >>>>>>>>> I think, Diwali is short enough to not make it a concern
> >> (similar to
> >>>>>>>>> Thanksgiving)?
> >>>>>>>>>
> >>>>>>>>> For Lunar New Year, I agree that it would be good to plan for it.
> >>>>>>>>> However, it seems it's shifting quite a lot year-over-year what
> >> makes it
> >>>>>>>>> very difficult (w/o a huge window) to include?
> >>>>>>>>>
> >>>>>>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced
> >> (this
> >>>>>>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5
> >> month
> >>>>>>>>> releases does not sound very appealing?
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>> I was playing with it a little bit more. Overall we have 52
> >> weeks, so
> >>>>>>>>> roughly 17.3 weeks per release. So we can naturally split it into
> >>>>>>>>> 17/17/18 using the additional week for the holidays. But that's
> >> not good
> >>>>>>>>> enough to take both Christmas and Lunar New Year into account.
> >> However,
> >>>>>>>>> with a 16/16/20, we could make it work I believe, even if it
> >> does cut it
> >>>>>>>>> tight sometimes... To be able to phrase a "simple rule", we can
> >> use ISO
> >>>>>>>>> weeks:
> >>>>>>>>>
> >>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature
> >> freeze
> >>>>>>>>> +3 weeks, and code freeze +2 weeks.
> >>>>>>>>>
> >>>>>>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer
> >> for most
> >>>>>>>>> years:
> >>>>>>>>>
> >>>>>>>>>>
> >>
>  ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> >>>>>>>>>>    │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │
> >> Buffer                 │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2027 │ Feb 6  │ Mar 3                      │ 25 days
> >>                         │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2028 │ Jan 26 │ Mar 1                      │ 34 days
> >>                         │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2029 │ Feb 13 │ Feb 28                     │ 15 days
> >>                         │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2030 │ Feb 3  │ Feb 27                     │ 24 days
> >>                         │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2031 │ Jan 23 │ Feb 26                     │ 34 days
> >>                         │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2032 │ Feb 11 │ Feb 25                     │ 14 days
> >>                         │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2033 │ Jan 31 │ Mar 2                      │ 30 days
> >>                         │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ←
> >> tightest with realistic LNY │
> >>>>>>>>>>
> >>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>>>>>    │ 2035 │ Feb 8  │ Feb 28                     │ 20 days
> >>                         │
> >>>>>>>>>>
> >>
>  └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> >>>>>>>>>
> >>>>>>>>> Guess, we could also make it week-8 instead of 9, and accept
> >> that 2034
> >>>>>>>>> would cut it too short (or be even more flexible and keep week-9
> >> for
> >>>>>>>>> 2034 as an exception to the rule), but that's a single year...
> >> for 2029
> >>>>>>>>> and 2032 we still have one week buffer.
> >>>>>>>>>
> >>>>>>>>> For Christmas, we would still ensure that the release can go out
> >> on
> >>>>>>>>> time, too. The below Q3 target release date is 4 weeks after code
> >>>>>>>>> freeze. This is a little bit more tight, as we often need more
> >> than 4
> >>>>>>>>> weeks to get it done:
> >>>>>>>>>
> >>>>>>>>>>    ┌──────┬──────────────────────────┬────────────────┐
> >>>>>>>>>>    │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2026 │ Dec 9                    │ 16             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2027 │ Dec 15                   │ 10             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2028 │ Dec 13                   │ 12             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2029 │ Dec 12                   │ 13             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2030 │ Dec 11                   │ 14             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2031 │ Dec 10                   │ 15             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2032 │ Dec 8                    │ 17             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2033 │ Dec 14                   │ 11             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2034 │ Dec 13                   │ 12             │
> >>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
> >>>>>>>>>>    │ 2035 │ Dec 12                   │ 13             │
> >>>>>>>>>>    └──────┴──────────────────────────┴────────────────┘
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>> For some years, we would still cut it a little bit tight, but we
> >> could
> >>>>>>>>> also apply an exception to the rule, to compensate for it. If we
> >> have
> >>>>>>>>> enough buffer room from LYN, we can start the Q1 release train a
> >> week
> >>>>>>>>> earlier, reducing the Q3->Q1 timeline by one week.
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>> Curious to hear what people think about it. In the end, if we
> >> can agree
> >>>>>>>>> on high level rules like "start Q1 release with X buffer after
> >> LYN", and
> >>>>>>>>> "complete Q3 release with Y buffer before Christmas", and still
> >> aim for
> >>>>>>>>> an as balance week-allocation per release, we can just put
> >> together a
> >>>>>>>>> release plan for the next 10 years and be done with it.
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>> -Matthias
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>>
> >>>>>>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
> >>>>>>>>>> Hello,
> >>>>>>>>>>
> >>>>>>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's
> >> release
> >>>>>>>>>> coincided with a reasonable amount of holidays, so I would
> >> favour a
> >>>>>>>>>> 3+2 schedule with releases in March, July and November.
> >>>>>>>>>>
> >>>>>>>>>> Best,
> >>>>>>>>>> Christo
> >>>>>>>>>>
> >>>>>>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com>
> >> wrote:
> >>>>>>>>>>>
> >>>>>>>>>>> Hi,
> >>>>>>>>>>>
> >>>>>>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <
> >> schofielaj@apache.org>
> >>>>>>>>>>> wrote:
> >>>>>>>>>>>
> >>>>>>>>>>>> I would go for 3 releases of equal lengths in March, July and
> >> November. I
> >>>>>>>>>>>> favour the 3+2 schedule.
> >>>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> +1
> >>>>>>>>>>>
> >>>>>>>>>>> Thanks,
> >>>>>>>>>>> Ismael
> >>>>>>>>>
> >>>>>>>>>
> >>>>>
> >>>>
> >>>
> >>
>
>

Re: [DISCUSS] Tightening time-based releases

Posted by "Matthias J. Sax" <mj...@apache.org>.
What's you rational for doing 4.6 in March already? If we so 4.5 
early/mid January, it would be a very small release. Given the overhead 
releases have, I am wondering if it would be worth doing?

For 4.5, I am also ok to start a Jan/6 instead of Jan/13 with KIP 
freeze. If we do 4.6 in July (instead of March) I don't see a need to 
pull it forward by one a week though. But I am fine in any case.

-Matthias

On 9/28/26 11:49 PM, Dejan Stojadinović wrote:
>> "Anyway, all I want to say is that the original schedule was hard to
> follow. The codebase is large and CVEs are common these days."
> 
> As an occasional contributor I can confirm this.
>  From my POV committers/maintainers are having a challenging task to keep
> up, especially while trying to juggle around with multiple release
> branches/versions (trunk/4.5.0, 4.4.0, 4.3.2, 4.2.2).
> 
> Best regards,
> Dejan S.
> https://github.com/dejan2609
> 
> 
> On Tue, Sep 29, 2026 at 7:55 AM Chia-Ping Tsai <ch...@apache.org> wrote:
> 
>> Hi Andrew,
>>
>> +1 to the revised 4.5.0 schedule. I would not pull it a week earlier,
>> since that puts the KIP freeze right after the holidays.
>>
>> Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0
>> reaches code freeze. We had too many must-have changes in 4.4.0, which
>> caused a serious delay. I try to review all backports but my tokens are
>> limited :)
>>
>> Anyway, all I want to say is that the original schedule was hard to
>> follow. The codebase is large and CVEs are common these days.
>>
>> Best,
>> Chia-Ping
>>
>>
>> On 2026/09/25 08:36:23 Andrew Schofield wrote:
>>> Hi,
>>> I'm starting to look at the release schedule for 4.5.0.
>>>
>>> In Chia-Ping's table from 29 May 2026, we have:
>>>
>>> Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
>>> 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19, 2026
>>> 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
>>>
>>> In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP
>> freeze of Jan 13, 2027 which gives this.
>>>
>>> 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon now
>>> 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17, 2027
>>> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
>>>
>>> I think we could even pull 4.5.0 a week earlier than that, but I don't
>> think there's anything fundamentally wrong with a short 4.6.0 provided
>> people know in advance.
>>>
>>> Are people happy with the 4.5.0 (revised) dates above? Would a week
>> earlier be better given that it gives an extra week to 4.6.0 which is
>> looking very squeezed?
>>>
>>> Thanks,
>>> Andrew
>>>
>>> On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
>>>> Hi all,
>>>>
>>>> +1 to the 4.5 schedule, given the delay of 4.4.0.
>>>>
>>>> I think having 4.5 land right on Lunar New Year should be fine this
>> time, since "too close" is the real issue here.
>>>>
>>>> Best,
>>>> Chia-Ping
>>>>
>>>>> Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
>>>>>
>>>>> I don't think we have for 4.5... I believe that there was a slight,
>> high level agreement for the long term, that we want to follow the
>> "calendar week proposal"?
>>>>>
>>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with
>> feature freeze
>>>>>>>>>> +3 weeks, and code freeze +2 weeks.
>>>>>
>>>>> AK 4.4 KIP freeze was Jul/8.
>>>>>
>>>>> For 2026, week 41 would be Oct/7 which is way too close IMHO.
>>>>> For 2027, week 9 would be Mar/3 which seems to be too far out?
>>>>>
>>>>> Second release (week 25) for 2027 would be Jun/23, which would be
>> 4.6?
>>>>>
>>>>> So the question remains, where to put 4.5 so we can start to follow
>> the new schedule starting with 4.6. It seems KIP freeze between Jan 13
>> might be a good compromise putting it roughly between 4.4 and 4.6 KIP
>> freeze dates. Of course, the release would overlap with Chinese New Year,
>> but it would be the last time we have this overlap.
>>>>>
>>>>> Just my 2tc.
>>>>>
>>>>>
>>>>> -Matthias
>>>>>
>>>>>> On 9/15/26 8:17 AM, David Jacot wrote:
>>>>>> Hi all,
>>>>>> Have we converged on the schedule that we want to use yet? Based on
>>>>>> the last emails, it is not so clear to me. For your context, I am
>>>>>> trying to figure out the timeline for 4.5.
>>>>>> Best,
>>>>>> David
>>>>>>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <
>> christololov@gmail.com> wrote:
>>>>>>>
>>>>>>> I think we could be a bit more lax with the KIP freeze of 6.7.0
>> when
>>>>>>> 2034 comes around to account for LNY :)
>>>>>>>
>>>>>>> Joking aside, my preference between "KIP freeze would be
>> Wednesday's,
>>>>>>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
>>>>>>> weeks" and "start Q1 release with X buffer after LYN and complete
>> Q3
>>>>>>> release with Y buffer before Christmas" would be to start with the
>>>>>>> first and adjust by shifting by 1 week on individual basis the ones
>>>>>>> which are too close. I am okay with too close being defined as 2
>>>>>>> weeks. Based on your very detailed tables we seem to have only a
>> few
>>>>>>> such outliers in the next decade.
>>>>>>>
>>>>>>> Best,
>>>>>>> Christo
>>>>>>>
>>>>>>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org>
>> wrote:
>>>>>>>>
>>>>>>>> Matthias, you are definitely a master of scheduling! I'm +100 to
>> your plans.
>>>>>>>>
>>>>>>>> Also, it is absolutely helpful to write down the exact release
>> dates for the next few years at once. It will ease our anxiety a lot! I'd
>> be more than happy to update the wiki with these dates once we all agree on
>> this new schedule. For example:
>>>>>>>>
>>>>>>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target
>> Release
>>>>>>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19,
>> 2026
>>>>>>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9,
>> 2026
>>>>>>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5,
>> 2027
>>>>>>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25,
>> 2027
>>>>>>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15,
>> 2027
>>>>>>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3,
>> 2028
>>>>>>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23,
>> 2028
>>>>>>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13,
>> 2028
>>>>>>>>
>>>>>>>> Furthermore, let's do our best to avoid fixing release blockers
>> under the Christmas tree :)
>>>>>>>>
>>>>>>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
>>>>>>>>> Thanks for the input.
>>>>>>>>>
>>>>>>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
>>>>>>>>> windows gets a lot of support.
>>>>>>>>>
>>>>>>>>> However, the idea of dividing the release windows not into 3
>> equally
>>>>>>>>> long junks, but be more flexible to account for holiday seasons,
>> seems
>>>>>>>>> not to get a lot of support? Not sure why... I personally think
>> it would
>>>>>>>>> be good to take it into account, and I also think that Chia-Ping
>> raises
>>>>>>>>> a very good point.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising
>> it. I
>>>>>>>>> did actually briefly think about it, but it's difficult, and
>> thus I just
>>>>>>>>> started with something simple as proposal to kick off a
>> discussion.
>>>>>>>>>
>>>>>>>>> I think, Diwali is short enough to not make it a concern
>> (similar to
>>>>>>>>> Thanksgiving)?
>>>>>>>>>
>>>>>>>>> For Lunar New Year, I agree that it would be good to plan for it.
>>>>>>>>> However, it seems it's shifting quite a lot year-over-year what
>> makes it
>>>>>>>>> very difficult (w/o a huge window) to include?
>>>>>>>>>
>>>>>>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced
>> (this
>>>>>>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5
>> month
>>>>>>>>> releases does not sound very appealing?
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> I was playing with it a little bit more. Overall we have 52
>> weeks, so
>>>>>>>>> roughly 17.3 weeks per release. So we can naturally split it into
>>>>>>>>> 17/17/18 using the additional week for the holidays. But that's
>> not good
>>>>>>>>> enough to take both Christmas and Lunar New Year into account.
>> However,
>>>>>>>>> with a 16/16/20, we could make it work I believe, even if it
>> does cut it
>>>>>>>>> tight sometimes... To be able to phrase a "simple rule", we can
>> use ISO
>>>>>>>>> weeks:
>>>>>>>>>
>>>>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature
>> freeze
>>>>>>>>> +3 weeks, and code freeze +2 weeks.
>>>>>>>>>
>>>>>>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer
>> for most
>>>>>>>>> years:
>>>>>>>>>
>>>>>>>>>>
>>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
>>>>>>>>>>    │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │
>> Buffer                 │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2027 │ Feb 6  │ Mar 3                      │ 25 days
>>                         │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2028 │ Jan 26 │ Mar 1                      │ 34 days
>>                         │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2029 │ Feb 13 │ Feb 28                     │ 15 days
>>                         │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2030 │ Feb 3  │ Feb 27                     │ 24 days
>>                         │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2031 │ Jan 23 │ Feb 26                     │ 34 days
>>                         │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2032 │ Feb 11 │ Feb 25                     │ 14 days
>>                         │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2033 │ Jan 31 │ Mar 2                      │ 30 days
>>                         │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ←
>> tightest with realistic LNY │
>>>>>>>>>>
>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>>>>>    │ 2035 │ Feb 8  │ Feb 28                     │ 20 days
>>                         │
>>>>>>>>>>
>>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
>>>>>>>>>
>>>>>>>>> Guess, we could also make it week-8 instead of 9, and accept
>> that 2034
>>>>>>>>> would cut it too short (or be even more flexible and keep week-9
>> for
>>>>>>>>> 2034 as an exception to the rule), but that's a single year...
>> for 2029
>>>>>>>>> and 2032 we still have one week buffer.
>>>>>>>>>
>>>>>>>>> For Christmas, we would still ensure that the release can go out
>> on
>>>>>>>>> time, too. The below Q3 target release date is 4 weeks after code
>>>>>>>>> freeze. This is a little bit more tight, as we often need more
>> than 4
>>>>>>>>> weeks to get it done:
>>>>>>>>>
>>>>>>>>>>    ┌──────┬──────────────────────────┬────────────────┐
>>>>>>>>>>    │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2026 │ Dec 9                    │ 16             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2027 │ Dec 15                   │ 10             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2028 │ Dec 13                   │ 12             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2029 │ Dec 12                   │ 13             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2030 │ Dec 11                   │ 14             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2031 │ Dec 10                   │ 15             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2032 │ Dec 8                    │ 17             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2033 │ Dec 14                   │ 11             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2034 │ Dec 13                   │ 12             │
>>>>>>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>>>>>>    │ 2035 │ Dec 12                   │ 13             │
>>>>>>>>>>    └──────┴──────────────────────────┴────────────────┘
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> For some years, we would still cut it a little bit tight, but we
>> could
>>>>>>>>> also apply an exception to the rule, to compensate for it. If we
>> have
>>>>>>>>> enough buffer room from LYN, we can start the Q1 release train a
>> week
>>>>>>>>> earlier, reducing the Q3->Q1 timeline by one week.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> Curious to hear what people think about it. In the end, if we
>> can agree
>>>>>>>>> on high level rules like "start Q1 release with X buffer after
>> LYN", and
>>>>>>>>> "complete Q3 release with Y buffer before Christmas", and still
>> aim for
>>>>>>>>> an as balance week-allocation per release, we can just put
>> together a
>>>>>>>>> release plan for the next 10 years and be done with it.
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> -Matthias
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
>>>>>>>>>> Hello,
>>>>>>>>>>
>>>>>>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's
>> release
>>>>>>>>>> coincided with a reasonable amount of holidays, so I would
>> favour a
>>>>>>>>>> 3+2 schedule with releases in March, July and November.
>>>>>>>>>>
>>>>>>>>>> Best,
>>>>>>>>>> Christo
>>>>>>>>>>
>>>>>>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com>
>> wrote:
>>>>>>>>>>>
>>>>>>>>>>> Hi,
>>>>>>>>>>>
>>>>>>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <
>> schofielaj@apache.org>
>>>>>>>>>>> wrote:
>>>>>>>>>>>
>>>>>>>>>>>> I would go for 3 releases of equal lengths in March, July and
>> November. I
>>>>>>>>>>>> favour the 3+2 schedule.
>>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> +1
>>>>>>>>>>>
>>>>>>>>>>> Thanks,
>>>>>>>>>>> Ismael
>>>>>>>>>
>>>>>>>>>
>>>>>
>>>>
>>>
>>


Re: [DISCUSS] Tightening time-based releases

Posted by Dejan Stojadinović <de...@gmail.com>.
> "Anyway, all I want to say is that the original schedule was hard to
follow. The codebase is large and CVEs are common these days."

As an occasional contributor I can confirm this.
From my POV committers/maintainers are having a challenging task to keep
up, especially while trying to juggle around with multiple release
branches/versions (trunk/4.5.0, 4.4.0, 4.3.2, 4.2.2).

Best regards,
Dejan S.
https://github.com/dejan2609


On Tue, Sep 29, 2026 at 7:55 AM Chia-Ping Tsai <ch...@apache.org> wrote:

> Hi Andrew,
>
> +1 to the revised 4.5.0 schedule. I would not pull it a week earlier,
> since that puts the KIP freeze right after the holidays.
>
> Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0
> reaches code freeze. We had too many must-have changes in 4.4.0, which
> caused a serious delay. I try to review all backports but my tokens are
> limited :)
>
> Anyway, all I want to say is that the original schedule was hard to
> follow. The codebase is large and CVEs are common these days.
>
> Best,
> Chia-Ping
>
>
> On 2026/09/25 08:36:23 Andrew Schofield wrote:
> > Hi,
> > I'm starting to look at the release schedule for 4.5.0.
> >
> > In Chia-Ping's table from 29 May 2026, we have:
> >
> > Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
> > 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19, 2026
> > 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
> > 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> >
> > In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP
> freeze of Jan 13, 2027 which gives this.
> >
> > 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon now
> > 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17, 2027
> > 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> >
> > I think we could even pull 4.5.0 a week earlier than that, but I don't
> think there's anything fundamentally wrong with a short 4.6.0 provided
> people know in advance.
> >
> > Are people happy with the 4.5.0 (revised) dates above? Would a week
> earlier be better given that it gives an extra week to 4.6.0 which is
> looking very squeezed?
> >
> > Thanks,
> > Andrew
> >
> > On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
> > > Hi all,
> > >
> > > +1 to the 4.5 schedule, given the delay of 4.4.0.
> > >
> > > I think having 4.5 land right on Lunar New Year should be fine this
> time, since "too close" is the real issue here.
> > >
> > > Best,
> > > Chia-Ping
> > >
> > > > Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> > > >
> > > > I don't think we have for 4.5... I believe that there was a slight,
> high level agreement for the long term, that we want to follow the
> "calendar week proposal"?
> > > >
> > > >>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with
> feature freeze
> > > >>>>>> +3 weeks, and code freeze +2 weeks.
> > > >
> > > > AK 4.4 KIP freeze was Jul/8.
> > > >
> > > > For 2026, week 41 would be Oct/7 which is way too close IMHO.
> > > > For 2027, week 9 would be Mar/3 which seems to be too far out?
> > > >
> > > > Second release (week 25) for 2027 would be Jun/23, which would be
> 4.6?
> > > >
> > > > So the question remains, where to put 4.5 so we can start to follow
> the new schedule starting with 4.6. It seems KIP freeze between Jan 13
> might be a good compromise putting it roughly between 4.4 and 4.6 KIP
> freeze dates. Of course, the release would overlap with Chinese New Year,
> but it would be the last time we have this overlap.
> > > >
> > > > Just my 2tc.
> > > >
> > > >
> > > > -Matthias
> > > >
> > > >> On 9/15/26 8:17 AM, David Jacot wrote:
> > > >> Hi all,
> > > >> Have we converged on the schedule that we want to use yet? Based on
> > > >> the last emails, it is not so clear to me. For your context, I am
> > > >> trying to figure out the timeline for 4.5.
> > > >> Best,
> > > >> David
> > > >>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <
> christololov@gmail.com> wrote:
> > > >>>
> > > >>> I think we could be a bit more lax with the KIP freeze of 6.7.0
> when
> > > >>> 2034 comes around to account for LNY :)
> > > >>>
> > > >>> Joking aside, my preference between "KIP freeze would be
> Wednesday's,
> > > >>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> > > >>> weeks" and "start Q1 release with X buffer after LYN and complete
> Q3
> > > >>> release with Y buffer before Christmas" would be to start with the
> > > >>> first and adjust by shifting by 1 week on individual basis the ones
> > > >>> which are too close. I am okay with too close being defined as 2
> > > >>> weeks. Based on your very detailed tables we seem to have only a
> few
> > > >>> such outliers in the next decade.
> > > >>>
> > > >>> Best,
> > > >>> Christo
> > > >>>
> > > >>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org>
> wrote:
> > > >>>>
> > > >>>> Matthias, you are definitely a master of scheduling! I'm +100 to
> your plans.
> > > >>>>
> > > >>>> Also, it is absolutely helpful to write down the exact release
> dates for the next few years at once. It will ease our anxiety a lot! I'd
> be more than happy to update the wiki with these dates once we all agree on
> this new schedule. For example:
> > > >>>>
> > > >>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target
> Release
> > > >>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19,
> 2026
> > > >>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9,
> 2026
> > > >>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5,
> 2027
> > > >>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25,
> 2027
> > > >>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15,
> 2027
> > > >>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3,
> 2028
> > > >>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23,
> 2028
> > > >>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13,
> 2028
> > > >>>>
> > > >>>> Furthermore, let's do our best to avoid fixing release blockers
> under the Christmas tree :)
> > > >>>>
> > > >>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> > > >>>>> Thanks for the input.
> > > >>>>>
> > > >>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
> > > >>>>> windows gets a lot of support.
> > > >>>>>
> > > >>>>> However, the idea of dividing the release windows not into 3
> equally
> > > >>>>> long junks, but be more flexible to account for holiday seasons,
> seems
> > > >>>>> not to get a lot of support? Not sure why... I personally think
> it would
> > > >>>>> be good to take it into account, and I also think that Chia-Ping
> raises
> > > >>>>> a very good point.
> > > >>>>>
> > > >>>>>
> > > >>>>>
> > > >>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising
> it. I
> > > >>>>> did actually briefly think about it, but it's difficult, and
> thus I just
> > > >>>>> started with something simple as proposal to kick off a
> discussion.
> > > >>>>>
> > > >>>>> I think, Diwali is short enough to not make it a concern
> (similar to
> > > >>>>> Thanksgiving)?
> > > >>>>>
> > > >>>>> For Lunar New Year, I agree that it would be good to plan for it.
> > > >>>>> However, it seems it's shifting quite a lot year-over-year what
> makes it
> > > >>>>> very difficult (w/o a huge window) to include?
> > > >>>>>
> > > >>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced
> (this
> > > >>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5
> month
> > > >>>>> releases does not sound very appealing?
> > > >>>>>
> > > >>>>>
> > > >>>>> I was playing with it a little bit more. Overall we have 52
> weeks, so
> > > >>>>> roughly 17.3 weeks per release. So we can naturally split it into
> > > >>>>> 17/17/18 using the additional week for the holidays. But that's
> not good
> > > >>>>> enough to take both Christmas and Lunar New Year into account.
> However,
> > > >>>>> with a 16/16/20, we could make it work I believe, even if it
> does cut it
> > > >>>>> tight sometimes... To be able to phrase a "simple rule", we can
> use ISO
> > > >>>>> weeks:
> > > >>>>>
> > > >>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature
> freeze
> > > >>>>> +3 weeks, and code freeze +2 weeks.
> > > >>>>>
> > > >>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer
> for most
> > > >>>>> years:
> > > >>>>>
> > > >>>>>>
>  ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> > > >>>>>>   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │
> Buffer                 │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days
>                        │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days
>                        │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days
>                        │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days
>                        │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days
>                        │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days
>                        │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days
>                        │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ←
> tightest with realistic LNY │
> > > >>>>>>
>  ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >>>>>>   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days
>                        │
> > > >>>>>>
>  └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> > > >>>>>
> > > >>>>> Guess, we could also make it week-8 instead of 9, and accept
> that 2034
> > > >>>>> would cut it too short (or be even more flexible and keep week-9
> for
> > > >>>>> 2034 as an exception to the rule), but that's a single year...
> for 2029
> > > >>>>> and 2032 we still have one week buffer.
> > > >>>>>
> > > >>>>> For Christmas, we would still ensure that the release can go out
> on
> > > >>>>> time, too. The below Q3 target release date is 4 weeks after code
> > > >>>>> freeze. This is a little bit more tight, as we often need more
> than 4
> > > >>>>> weeks to get it done:
> > > >>>>>
> > > >>>>>>   ┌──────┬──────────────────────────┬────────────────┐
> > > >>>>>>   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2026 │ Dec 9                    │ 16             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2027 │ Dec 15                   │ 10             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2028 │ Dec 13                   │ 12             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2029 │ Dec 12                   │ 13             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2030 │ Dec 11                   │ 14             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2031 │ Dec 10                   │ 15             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2032 │ Dec 8                    │ 17             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2033 │ Dec 14                   │ 11             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2034 │ Dec 13                   │ 12             │
> > > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > > >>>>>>   │ 2035 │ Dec 12                   │ 13             │
> > > >>>>>>   └──────┴──────────────────────────┴────────────────┘
> > > >>>>>
> > > >>>>>
> > > >>>>>
> > > >>>>> For some years, we would still cut it a little bit tight, but we
> could
> > > >>>>> also apply an exception to the rule, to compensate for it. If we
> have
> > > >>>>> enough buffer room from LYN, we can start the Q1 release train a
> week
> > > >>>>> earlier, reducing the Q3->Q1 timeline by one week.
> > > >>>>>
> > > >>>>>
> > > >>>>> Curious to hear what people think about it. In the end, if we
> can agree
> > > >>>>> on high level rules like "start Q1 release with X buffer after
> LYN", and
> > > >>>>> "complete Q3 release with Y buffer before Christmas", and still
> aim for
> > > >>>>> an as balance week-allocation per release, we can just put
> together a
> > > >>>>> release plan for the next 10 years and be done with it.
> > > >>>>>
> > > >>>>>
> > > >>>>> -Matthias
> > > >>>>>
> > > >>>>>
> > > >>>>>
> > > >>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
> > > >>>>>> Hello,
> > > >>>>>>
> > > >>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's
> release
> > > >>>>>> coincided with a reasonable amount of holidays, so I would
> favour a
> > > >>>>>> 3+2 schedule with releases in March, July and November.
> > > >>>>>>
> > > >>>>>> Best,
> > > >>>>>> Christo
> > > >>>>>>
> > > >>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com>
> wrote:
> > > >>>>>>>
> > > >>>>>>> Hi,
> > > >>>>>>>
> > > >>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <
> schofielaj@apache.org>
> > > >>>>>>> wrote:
> > > >>>>>>>
> > > >>>>>>>> I would go for 3 releases of equal lengths in March, July and
> November. I
> > > >>>>>>>> favour the 3+2 schedule.
> > > >>>>>>>>
> > > >>>>>>>
> > > >>>>>>> +1
> > > >>>>>>>
> > > >>>>>>> Thanks,
> > > >>>>>>> Ismael
> > > >>>>>
> > > >>>>>
> > > >
> > >
> >
>

Re: [DISCUSS] Tightening time-based releases

Posted by Chia-Ping Tsai <ch...@apache.org>.
Hi Andrew,

+1 to the revised 4.5.0 schedule. I would not pull it a week earlier, since that puts the KIP freeze right after the holidays.

Regarding the 4.6.0 schedule, I think we could discuss it once 4.5.0 reaches code freeze. We had too many must-have changes in 4.4.0, which caused a serious delay. I try to review all backports but my tokens are limited :)

Anyway, all I want to say is that the original schedule was hard to follow. The codebase is large and CVEs are common these days.

Best,
Chia-Ping


On 2026/09/25 08:36:23 Andrew Schofield wrote:
> Hi,
> I'm starting to look at the release schedule for 4.5.0.
> 
> In Chia-Ping's table from 29 May 2026, we have:
> 
> Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
> 4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19, 2026
> 4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> 
> In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP freeze of Jan 13, 2027 which gives this.
> 
> 4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon now
> 4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17, 2027
> 4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027
> 
> I think we could even pull 4.5.0 a week earlier than that, but I don't think there's anything fundamentally wrong with a short 4.6.0 provided people know in advance.
> 
> Are people happy with the 4.5.0 (revised) dates above? Would a week earlier be better given that it gives an extra week to 4.6.0 which is looking very squeezed?
> 
> Thanks,
> Andrew
> 
> On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
> > Hi all,
> > 
> > +1 to the 4.5 schedule, given the delay of 4.4.0.
> > 
> > I think having 4.5 land right on Lunar New Year should be fine this time, since "too close" is the real issue here.
> > 
> > Best,
> > Chia-Ping
> > 
> > > Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> > > 
> > > I don't think we have for 4.5... I believe that there was a slight, high level agreement for the long term, that we want to follow the "calendar week proposal"?
> > > 
> > >>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
> > >>>>>> +3 weeks, and code freeze +2 weeks.
> > > 
> > > AK 4.4 KIP freeze was Jul/8.
> > > 
> > > For 2026, week 41 would be Oct/7 which is way too close IMHO.
> > > For 2027, week 9 would be Mar/3 which seems to be too far out?
> > > 
> > > Second release (week 25) for 2027 would be Jun/23, which would be 4.6?
> > > 
> > > So the question remains, where to put 4.5 so we can start to follow the new schedule starting with 4.6. It seems KIP freeze between Jan 13 might be a good compromise putting it roughly between 4.4 and 4.6 KIP freeze dates. Of course, the release would overlap with Chinese New Year, but it would be the last time we have this overlap.
> > > 
> > > Just my 2tc.
> > > 
> > > 
> > > -Matthias
> > > 
> > >> On 9/15/26 8:17 AM, David Jacot wrote:
> > >> Hi all,
> > >> Have we converged on the schedule that we want to use yet? Based on
> > >> the last emails, it is not so clear to me. For your context, I am
> > >> trying to figure out the timeline for 4.5.
> > >> Best,
> > >> David
> > >>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <ch...@gmail.com> wrote:
> > >>> 
> > >>> I think we could be a bit more lax with the KIP freeze of 6.7.0 when
> > >>> 2034 comes around to account for LNY :)
> > >>> 
> > >>> Joking aside, my preference between "KIP freeze would be Wednesday's,
> > >>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> > >>> weeks" and "start Q1 release with X buffer after LYN and complete Q3
> > >>> release with Y buffer before Christmas" would be to start with the
> > >>> first and adjust by shifting by 1 week on individual basis the ones
> > >>> which are too close. I am okay with too close being defined as 2
> > >>> weeks. Based on your very detailed tables we seem to have only a few
> > >>> such outliers in the next decade.
> > >>> 
> > >>> Best,
> > >>> Christo
> > >>> 
> > >>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org> wrote:
> > >>>> 
> > >>>> Matthias, you are definitely a master of scheduling! I'm +100 to your plans.
> > >>>> 
> > >>>> Also, it is absolutely helpful to write down the exact release dates for the next few years at once. It will ease our anxiety a lot! I'd be more than happy to update the wiki with these dates once we all agree on this new schedule. For example:
> > >>>> 
> > >>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target Release
> > >>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19, 2026
> > >>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9, 2026
> > >>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5, 2027
> > >>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25, 2027
> > >>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15, 2027
> > >>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3, 2028
> > >>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23, 2028
> > >>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13, 2028
> > >>>> 
> > >>>> Furthermore, let's do our best to avoid fixing release blockers under the Christmas tree :)
> > >>>> 
> > >>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> > >>>>> Thanks for the input.
> > >>>>> 
> > >>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
> > >>>>> windows gets a lot of support.
> > >>>>> 
> > >>>>> However, the idea of dividing the release windows not into 3 equally
> > >>>>> long junks, but be more flexible to account for holiday seasons, seems
> > >>>>> not to get a lot of support? Not sure why... I personally think it would
> > >>>>> be good to take it into account, and I also think that Chia-Ping raises
> > >>>>> a very good point.
> > >>>>> 
> > >>>>> 
> > >>>>> 
> > >>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising it. I
> > >>>>> did actually briefly think about it, but it's difficult, and thus I just
> > >>>>> started with something simple as proposal to kick off a discussion.
> > >>>>> 
> > >>>>> I think, Diwali is short enough to not make it a concern (similar to
> > >>>>> Thanksgiving)?
> > >>>>> 
> > >>>>> For Lunar New Year, I agree that it would be good to plan for it.
> > >>>>> However, it seems it's shifting quite a lot year-over-year what makes it
> > >>>>> very difficult (w/o a huge window) to include?
> > >>>>> 
> > >>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this
> > >>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5 month
> > >>>>> releases does not sound very appealing?
> > >>>>> 
> > >>>>> 
> > >>>>> I was playing with it a little bit more. Overall we have 52 weeks, so
> > >>>>> roughly 17.3 weeks per release. So we can naturally split it into
> > >>>>> 17/17/18 using the additional week for the holidays. But that's not good
> > >>>>> enough to take both Christmas and Lunar New Year into account. However,
> > >>>>> with a 16/16/20, we could make it work I believe, even if it does cut it
> > >>>>> tight sometimes... To be able to phrase a "simple rule", we can use ISO
> > >>>>> weeks:
> > >>>>> 
> > >>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
> > >>>>> +3 weeks, and code freeze +2 weeks.
> > >>>>> 
> > >>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer for most
> > >>>>> years:
> > >>>>> 
> > >>>>>>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> > >>>>>>   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
> > >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >>>>>>   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
> > >>>>>>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> > >>>>> 
> > >>>>> Guess, we could also make it week-8 instead of 9, and accept that 2034
> > >>>>> would cut it too short (or be even more flexible and keep week-9 for
> > >>>>> 2034 as an exception to the rule), but that's a single year... for 2029
> > >>>>> and 2032 we still have one week buffer.
> > >>>>> 
> > >>>>> For Christmas, we would still ensure that the release can go out on
> > >>>>> time, too. The below Q3 target release date is 4 weeks after code
> > >>>>> freeze. This is a little bit more tight, as we often need more than 4
> > >>>>> weeks to get it done:
> > >>>>> 
> > >>>>>>   ┌──────┬──────────────────────────┬────────────────┐
> > >>>>>>   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2026 │ Dec 9                    │ 16             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2027 │ Dec 15                   │ 10             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2028 │ Dec 13                   │ 12             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2029 │ Dec 12                   │ 13             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2030 │ Dec 11                   │ 14             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2031 │ Dec 10                   │ 15             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2032 │ Dec 8                    │ 17             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2033 │ Dec 14                   │ 11             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2034 │ Dec 13                   │ 12             │
> > >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> > >>>>>>   │ 2035 │ Dec 12                   │ 13             │
> > >>>>>>   └──────┴──────────────────────────┴────────────────┘
> > >>>>> 
> > >>>>> 
> > >>>>> 
> > >>>>> For some years, we would still cut it a little bit tight, but we could
> > >>>>> also apply an exception to the rule, to compensate for it. If we have
> > >>>>> enough buffer room from LYN, we can start the Q1 release train a week
> > >>>>> earlier, reducing the Q3->Q1 timeline by one week.
> > >>>>> 
> > >>>>> 
> > >>>>> Curious to hear what people think about it. In the end, if we can agree
> > >>>>> on high level rules like "start Q1 release with X buffer after LYN", and
> > >>>>> "complete Q3 release with Y buffer before Christmas", and still aim for
> > >>>>> an as balance week-allocation per release, we can just put together a
> > >>>>> release plan for the next 10 years and be done with it.
> > >>>>> 
> > >>>>> 
> > >>>>> -Matthias
> > >>>>> 
> > >>>>> 
> > >>>>> 
> > >>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
> > >>>>>> Hello,
> > >>>>>> 
> > >>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's release
> > >>>>>> coincided with a reasonable amount of holidays, so I would favour a
> > >>>>>> 3+2 schedule with releases in March, July and November.
> > >>>>>> 
> > >>>>>> Best,
> > >>>>>> Christo
> > >>>>>> 
> > >>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
> > >>>>>>> 
> > >>>>>>> Hi,
> > >>>>>>> 
> > >>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
> > >>>>>>> wrote:
> > >>>>>>> 
> > >>>>>>>> I would go for 3 releases of equal lengths in March, July and November. I
> > >>>>>>>> favour the 3+2 schedule.
> > >>>>>>>> 
> > >>>>>>> 
> > >>>>>>> +1
> > >>>>>>> 
> > >>>>>>> Thanks,
> > >>>>>>> Ismael
> > >>>>> 
> > >>>>> 
> > > 
> > 
> 

Re: [DISCUSS] Tightening time-based releases

Posted by Andrew Schofield <sc...@apache.org>.
Hi,
I'm starting to look at the release schedule for 4.5.0.

In Chia-Ping's table from 29 May 2026, we have:

Version : KIP Freeze : Feature Freeze : Code Freeze : Target Release
4.4.0 (planned) : Jun 17, 2026: Jul 8, 2026 : Jul 22, 2026 : Aug 19, 2026
4.5.0 : Oct 7, 2026 : Oct 28, 2026 : Nov 11, 2026 : Dec 9, 2026
4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027

In reality, 4.4.0 ran 3 weeks later than this. Matthias suggested a KIP freeze of Jan 13, 2027 which gives this.

4.4.0 (actual) : Jul 8, 2026: Jul 29, 2026 : Aug 12, 2026 : Real soon now
4.5.0 (revised): Jan 13, 2027 : Feb 3, 2027 : Feb 17, 2027 : Mar 17, 2027
4.6.0 : Mar 3, 2027 : Mar 24, 2027 : Apr 7, 2027 : May 5, 2027

I think we could even pull 4.5.0 a week earlier than that, but I don't think there's anything fundamentally wrong with a short 4.6.0 provided people know in advance.

Are people happy with the 4.5.0 (revised) dates above? Would a week earlier be better given that it gives an extra week to 4.6.0 which is looking very squeezed?

Thanks,
Andrew

On 2026/09/16 04:01:41 Chia-Ping Tsai wrote:
> Hi all,
> 
> +1 to the 4.5 schedule, given the delay of 4.4.0.
> 
> I think having 4.5 land right on Lunar New Year should be fine this time, since "too close" is the real issue here.
> 
> Best,
> Chia-Ping
> 
> > Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> > 
> > I don't think we have for 4.5... I believe that there was a slight, high level agreement for the long term, that we want to follow the "calendar week proposal"?
> > 
> >>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
> >>>>>> +3 weeks, and code freeze +2 weeks.
> > 
> > AK 4.4 KIP freeze was Jul/8.
> > 
> > For 2026, week 41 would be Oct/7 which is way too close IMHO.
> > For 2027, week 9 would be Mar/3 which seems to be too far out?
> > 
> > Second release (week 25) for 2027 would be Jun/23, which would be 4.6?
> > 
> > So the question remains, where to put 4.5 so we can start to follow the new schedule starting with 4.6. It seems KIP freeze between Jan 13 might be a good compromise putting it roughly between 4.4 and 4.6 KIP freeze dates. Of course, the release would overlap with Chinese New Year, but it would be the last time we have this overlap.
> > 
> > Just my 2tc.
> > 
> > 
> > -Matthias
> > 
> >> On 9/15/26 8:17 AM, David Jacot wrote:
> >> Hi all,
> >> Have we converged on the schedule that we want to use yet? Based on
> >> the last emails, it is not so clear to me. For your context, I am
> >> trying to figure out the timeline for 4.5.
> >> Best,
> >> David
> >>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <ch...@gmail.com> wrote:
> >>> 
> >>> I think we could be a bit more lax with the KIP freeze of 6.7.0 when
> >>> 2034 comes around to account for LNY :)
> >>> 
> >>> Joking aside, my preference between "KIP freeze would be Wednesday's,
> >>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> >>> weeks" and "start Q1 release with X buffer after LYN and complete Q3
> >>> release with Y buffer before Christmas" would be to start with the
> >>> first and adjust by shifting by 1 week on individual basis the ones
> >>> which are too close. I am okay with too close being defined as 2
> >>> weeks. Based on your very detailed tables we seem to have only a few
> >>> such outliers in the next decade.
> >>> 
> >>> Best,
> >>> Christo
> >>> 
> >>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org> wrote:
> >>>> 
> >>>> Matthias, you are definitely a master of scheduling! I'm +100 to your plans.
> >>>> 
> >>>> Also, it is absolutely helpful to write down the exact release dates for the next few years at once. It will ease our anxiety a lot! I'd be more than happy to update the wiki with these dates once we all agree on this new schedule. For example:
> >>>> 
> >>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target Release
> >>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19, 2026
> >>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9, 2026
> >>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5, 2027
> >>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25, 2027
> >>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15, 2027
> >>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3, 2028
> >>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23, 2028
> >>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13, 2028
> >>>> 
> >>>> Furthermore, let's do our best to avoid fixing release blockers under the Christmas tree :)
> >>>> 
> >>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> >>>>> Thanks for the input.
> >>>>> 
> >>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
> >>>>> windows gets a lot of support.
> >>>>> 
> >>>>> However, the idea of dividing the release windows not into 3 equally
> >>>>> long junks, but be more flexible to account for holiday seasons, seems
> >>>>> not to get a lot of support? Not sure why... I personally think it would
> >>>>> be good to take it into account, and I also think that Chia-Ping raises
> >>>>> a very good point.
> >>>>> 
> >>>>> 
> >>>>> 
> >>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising it. I
> >>>>> did actually briefly think about it, but it's difficult, and thus I just
> >>>>> started with something simple as proposal to kick off a discussion.
> >>>>> 
> >>>>> I think, Diwali is short enough to not make it a concern (similar to
> >>>>> Thanksgiving)?
> >>>>> 
> >>>>> For Lunar New Year, I agree that it would be good to plan for it.
> >>>>> However, it seems it's shifting quite a lot year-over-year what makes it
> >>>>> very difficult (w/o a huge window) to include?
> >>>>> 
> >>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this
> >>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5 month
> >>>>> releases does not sound very appealing?
> >>>>> 
> >>>>> 
> >>>>> I was playing with it a little bit more. Overall we have 52 weeks, so
> >>>>> roughly 17.3 weeks per release. So we can naturally split it into
> >>>>> 17/17/18 using the additional week for the holidays. But that's not good
> >>>>> enough to take both Christmas and Lunar New Year into account. However,
> >>>>> with a 16/16/20, we could make it work I believe, even if it does cut it
> >>>>> tight sometimes... To be able to phrase a "simple rule", we can use ISO
> >>>>> weeks:
> >>>>> 
> >>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
> >>>>> +3 weeks, and code freeze +2 weeks.
> >>>>> 
> >>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer for most
> >>>>> years:
> >>>>> 
> >>>>>>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> >>>>>>   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
> >>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >>>>>>   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
> >>>>>>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> >>>>> 
> >>>>> Guess, we could also make it week-8 instead of 9, and accept that 2034
> >>>>> would cut it too short (or be even more flexible and keep week-9 for
> >>>>> 2034 as an exception to the rule), but that's a single year... for 2029
> >>>>> and 2032 we still have one week buffer.
> >>>>> 
> >>>>> For Christmas, we would still ensure that the release can go out on
> >>>>> time, too. The below Q3 target release date is 4 weeks after code
> >>>>> freeze. This is a little bit more tight, as we often need more than 4
> >>>>> weeks to get it done:
> >>>>> 
> >>>>>>   ┌──────┬──────────────────────────┬────────────────┐
> >>>>>>   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2026 │ Dec 9                    │ 16             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2027 │ Dec 15                   │ 10             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2028 │ Dec 13                   │ 12             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2029 │ Dec 12                   │ 13             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2030 │ Dec 11                   │ 14             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2031 │ Dec 10                   │ 15             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2032 │ Dec 8                    │ 17             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2033 │ Dec 14                   │ 11             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2034 │ Dec 13                   │ 12             │
> >>>>>>   ├──────┼──────────────────────────┼────────────────┤
> >>>>>>   │ 2035 │ Dec 12                   │ 13             │
> >>>>>>   └──────┴──────────────────────────┴────────────────┘
> >>>>> 
> >>>>> 
> >>>>> 
> >>>>> For some years, we would still cut it a little bit tight, but we could
> >>>>> also apply an exception to the rule, to compensate for it. If we have
> >>>>> enough buffer room from LYN, we can start the Q1 release train a week
> >>>>> earlier, reducing the Q3->Q1 timeline by one week.
> >>>>> 
> >>>>> 
> >>>>> Curious to hear what people think about it. In the end, if we can agree
> >>>>> on high level rules like "start Q1 release with X buffer after LYN", and
> >>>>> "complete Q3 release with Y buffer before Christmas", and still aim for
> >>>>> an as balance week-allocation per release, we can just put together a
> >>>>> release plan for the next 10 years and be done with it.
> >>>>> 
> >>>>> 
> >>>>> -Matthias
> >>>>> 
> >>>>> 
> >>>>> 
> >>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
> >>>>>> Hello,
> >>>>>> 
> >>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's release
> >>>>>> coincided with a reasonable amount of holidays, so I would favour a
> >>>>>> 3+2 schedule with releases in March, July and November.
> >>>>>> 
> >>>>>> Best,
> >>>>>> Christo
> >>>>>> 
> >>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
> >>>>>>> 
> >>>>>>> Hi,
> >>>>>>> 
> >>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
> >>>>>>> wrote:
> >>>>>>> 
> >>>>>>>> I would go for 3 releases of equal lengths in March, July and November. I
> >>>>>>>> favour the 3+2 schedule.
> >>>>>>>> 
> >>>>>>> 
> >>>>>>> +1
> >>>>>>> 
> >>>>>>> Thanks,
> >>>>>>> Ismael
> >>>>> 
> >>>>> 
> > 
> 

Re: [DISCUSS] Tightening time-based releases

Posted by Chia-Ping Tsai <ch...@gmail.com>.
Hi all,

+1 to the 4.5 schedule, given the delay of 4.4.0.

I think having 4.5 land right on Lunar New Year should be fine this time, since "too close" is the real issue here.

Best,
Chia-Ping

> Matthias J. Sax <mj...@apache.org> 於 2026年9月16日 清晨7:05 寫道:
> 
> I don't think we have for 4.5... I believe that there was a slight, high level agreement for the long term, that we want to follow the "calendar week proposal"?
> 
>>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
>>>>>> +3 weeks, and code freeze +2 weeks.
> 
> AK 4.4 KIP freeze was Jul/8.
> 
> For 2026, week 41 would be Oct/7 which is way too close IMHO.
> For 2027, week 9 would be Mar/3 which seems to be too far out?
> 
> Second release (week 25) for 2027 would be Jun/23, which would be 4.6?
> 
> So the question remains, where to put 4.5 so we can start to follow the new schedule starting with 4.6. It seems KIP freeze between Jan 13 might be a good compromise putting it roughly between 4.4 and 4.6 KIP freeze dates. Of course, the release would overlap with Chinese New Year, but it would be the last time we have this overlap.
> 
> Just my 2tc.
> 
> 
> -Matthias
> 
>> On 9/15/26 8:17 AM, David Jacot wrote:
>> Hi all,
>> Have we converged on the schedule that we want to use yet? Based on
>> the last emails, it is not so clear to me. For your context, I am
>> trying to figure out the timeline for 4.5.
>> Best,
>> David
>>> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <ch...@gmail.com> wrote:
>>> 
>>> I think we could be a bit more lax with the KIP freeze of 6.7.0 when
>>> 2034 comes around to account for LNY :)
>>> 
>>> Joking aside, my preference between "KIP freeze would be Wednesday's,
>>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
>>> weeks" and "start Q1 release with X buffer after LYN and complete Q3
>>> release with Y buffer before Christmas" would be to start with the
>>> first and adjust by shifting by 1 week on individual basis the ones
>>> which are too close. I am okay with too close being defined as 2
>>> weeks. Based on your very detailed tables we seem to have only a few
>>> such outliers in the next decade.
>>> 
>>> Best,
>>> Christo
>>> 
>>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org> wrote:
>>>> 
>>>> Matthias, you are definitely a master of scheduling! I'm +100 to your plans.
>>>> 
>>>> Also, it is absolutely helpful to write down the exact release dates for the next few years at once. It will ease our anxiety a lot! I'd be more than happy to update the wiki with these dates once we all agree on this new schedule. For example:
>>>> 
>>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target Release
>>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19, 2026
>>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9, 2026
>>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5, 2027
>>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25, 2027
>>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15, 2027
>>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3, 2028
>>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23, 2028
>>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13, 2028
>>>> 
>>>> Furthermore, let's do our best to avoid fixing release blockers under the Christmas tree :)
>>>> 
>>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
>>>>> Thanks for the input.
>>>>> 
>>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
>>>>> windows gets a lot of support.
>>>>> 
>>>>> However, the idea of dividing the release windows not into 3 equally
>>>>> long junks, but be more flexible to account for holiday seasons, seems
>>>>> not to get a lot of support? Not sure why... I personally think it would
>>>>> be good to take it into account, and I also think that Chia-Ping raises
>>>>> a very good point.
>>>>> 
>>>>> 
>>>>> 
>>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising it. I
>>>>> did actually briefly think about it, but it's difficult, and thus I just
>>>>> started with something simple as proposal to kick off a discussion.
>>>>> 
>>>>> I think, Diwali is short enough to not make it a concern (similar to
>>>>> Thanksgiving)?
>>>>> 
>>>>> For Lunar New Year, I agree that it would be good to plan for it.
>>>>> However, it seems it's shifting quite a lot year-over-year what makes it
>>>>> very difficult (w/o a huge window) to include?
>>>>> 
>>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this
>>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5 month
>>>>> releases does not sound very appealing?
>>>>> 
>>>>> 
>>>>> I was playing with it a little bit more. Overall we have 52 weeks, so
>>>>> roughly 17.3 weeks per release. So we can naturally split it into
>>>>> 17/17/18 using the additional week for the holidays. But that's not good
>>>>> enough to take both Christmas and Lunar New Year into account. However,
>>>>> with a 16/16/20, we could make it work I believe, even if it does cut it
>>>>> tight sometimes... To be able to phrase a "simple rule", we can use ISO
>>>>> weeks:
>>>>> 
>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
>>>>> +3 weeks, and code freeze +2 weeks.
>>>>> 
>>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer for most
>>>>> years:
>>>>> 
>>>>>>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
>>>>>>   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
>>>>>>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>>   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
>>>>>>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
>>>>> 
>>>>> Guess, we could also make it week-8 instead of 9, and accept that 2034
>>>>> would cut it too short (or be even more flexible and keep week-9 for
>>>>> 2034 as an exception to the rule), but that's a single year... for 2029
>>>>> and 2032 we still have one week buffer.
>>>>> 
>>>>> For Christmas, we would still ensure that the release can go out on
>>>>> time, too. The below Q3 target release date is 4 weeks after code
>>>>> freeze. This is a little bit more tight, as we often need more than 4
>>>>> weeks to get it done:
>>>>> 
>>>>>>   ┌──────┬──────────────────────────┬────────────────┐
>>>>>>   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2026 │ Dec 9                    │ 16             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2027 │ Dec 15                   │ 10             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2028 │ Dec 13                   │ 12             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2029 │ Dec 12                   │ 13             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2030 │ Dec 11                   │ 14             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2031 │ Dec 10                   │ 15             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2032 │ Dec 8                    │ 17             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2033 │ Dec 14                   │ 11             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2034 │ Dec 13                   │ 12             │
>>>>>>   ├──────┼──────────────────────────┼────────────────┤
>>>>>>   │ 2035 │ Dec 12                   │ 13             │
>>>>>>   └──────┴──────────────────────────┴────────────────┘
>>>>> 
>>>>> 
>>>>> 
>>>>> For some years, we would still cut it a little bit tight, but we could
>>>>> also apply an exception to the rule, to compensate for it. If we have
>>>>> enough buffer room from LYN, we can start the Q1 release train a week
>>>>> earlier, reducing the Q3->Q1 timeline by one week.
>>>>> 
>>>>> 
>>>>> Curious to hear what people think about it. In the end, if we can agree
>>>>> on high level rules like "start Q1 release with X buffer after LYN", and
>>>>> "complete Q3 release with Y buffer before Christmas", and still aim for
>>>>> an as balance week-allocation per release, we can just put together a
>>>>> release plan for the next 10 years and be done with it.
>>>>> 
>>>>> 
>>>>> -Matthias
>>>>> 
>>>>> 
>>>>> 
>>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
>>>>>> Hello,
>>>>>> 
>>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's release
>>>>>> coincided with a reasonable amount of holidays, so I would favour a
>>>>>> 3+2 schedule with releases in March, July and November.
>>>>>> 
>>>>>> Best,
>>>>>> Christo
>>>>>> 
>>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
>>>>>>> 
>>>>>>> Hi,
>>>>>>> 
>>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
>>>>>>> wrote:
>>>>>>> 
>>>>>>>> I would go for 3 releases of equal lengths in March, July and November. I
>>>>>>>> favour the 3+2 schedule.
>>>>>>>> 
>>>>>>> 
>>>>>>> +1
>>>>>>> 
>>>>>>> Thanks,
>>>>>>> Ismael
>>>>> 
>>>>> 
> 

Re: [DISCUSS] Tightening time-based releases

Posted by "Matthias J. Sax" <mj...@apache.org>.
I don't think we have for 4.5... I believe that there was a slight, high 
level agreement for the long term, that we want to follow the "calendar 
week proposal"?

>>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
>>>>> +3 weeks, and code freeze +2 weeks.

AK 4.4 KIP freeze was Jul/8.

For 2026, week 41 would be Oct/7 which is way too close IMHO.
For 2027, week 9 would be Mar/3 which seems to be too far out?

Second release (week 25) for 2027 would be Jun/23, which would be 4.6?

So the question remains, where to put 4.5 so we can start to follow the 
new schedule starting with 4.6. It seems KIP freeze between Jan 13 might 
be a good compromise putting it roughly between 4.4 and 4.6 KIP freeze 
dates. Of course, the release would overlap with Chinese New Year, but 
it would be the last time we have this overlap.

Just my 2tc.


-Matthias

On 9/15/26 8:17 AM, David Jacot wrote:
> Hi all,
> 
> Have we converged on the schedule that we want to use yet? Based on
> the last emails, it is not so clear to me. For your context, I am
> trying to figure out the timeline for 4.5.
> 
> Best,
> David
> 
> On Fri, May 29, 2026 at 1:03 PM Christo Lolov <ch...@gmail.com> wrote:
>>
>> I think we could be a bit more lax with the KIP freeze of 6.7.0 when
>> 2034 comes around to account for LNY :)
>>
>> Joking aside, my preference between "KIP freeze would be Wednesday's,
>> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
>> weeks" and "start Q1 release with X buffer after LYN and complete Q3
>> release with Y buffer before Christmas" would be to start with the
>> first and adjust by shifting by 1 week on individual basis the ones
>> which are too close. I am okay with too close being defined as 2
>> weeks. Based on your very detailed tables we seem to have only a few
>> such outliers in the next decade.
>>
>> Best,
>> Christo
>>
>> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org> wrote:
>>>
>>> Matthias, you are definitely a master of scheduling! I'm +100 to your plans.
>>>
>>> Also, it is absolutely helpful to write down the exact release dates for the next few years at once. It will ease our anxiety a lot! I'd be more than happy to update the wiki with these dates once we all agree on this new schedule. For example:
>>>
>>> Version KIP Freeze      Feature Freeze  Code Freeze     Target Release
>>> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19, 2026
>>> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9, 2026
>>> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5, 2027
>>> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25, 2027
>>> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15, 2027
>>> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3, 2028
>>> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23, 2028
>>> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13, 2028
>>>
>>> Furthermore, let's do our best to avoid fixing release blockers under the Christmas tree :)
>>>
>>> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
>>>> Thanks for the input.
>>>>
>>>> Overall, it seems that 3+2 weeks for feature freeze / code freeze
>>>> windows gets a lot of support.
>>>>
>>>> However, the idea of dividing the release windows not into 3 equally
>>>> long junks, but be more flexible to account for holiday seasons, seems
>>>> not to get a lot of support? Not sure why... I personally think it would
>>>> be good to take it into account, and I also think that Chia-Ping raises
>>>> a very good point.
>>>>
>>>>
>>>>
>>>> @Chia-Ping: Other holidays are a great point. Thanks for raising it. I
>>>> did actually briefly think about it, but it's difficult, and thus I just
>>>> started with something simple as proposal to kick off a discussion.
>>>>
>>>> I think, Diwali is short enough to not make it a concern (similar to
>>>> Thanksgiving)?
>>>>
>>>> For Lunar New Year, I agree that it would be good to plan for it.
>>>> However, it seems it's shifting quite a lot year-over-year what makes it
>>>> very difficult (w/o a huge window) to include?
>>>>
>>>> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this
>>>> would be 22 / 15 / 15 weeks per release? And two very short 3.5 month
>>>> releases does not sound very appealing?
>>>>
>>>>
>>>> I was playing with it a little bit more. Overall we have 52 weeks, so
>>>> roughly 17.3 weeks per release. So we can naturally split it into
>>>> 17/17/18 using the additional week for the holidays. But that's not good
>>>> enough to take both Christmas and Lunar New Year into account. However,
>>>> with a 16/16/20, we could make it work I believe, even if it does cut it
>>>> tight sometimes... To be able to phrase a "simple rule", we can use ISO
>>>> weeks:
>>>>
>>>> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
>>>> +3 weeks, and code freeze +2 weeks.
>>>>
>>>> Ie, we put the first KIP freeze after LNY with plenty of buffer for most
>>>> years:
>>>>
>>>>>    ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
>>>>>    │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
>>>>>    ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>>>>>    │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
>>>>>    └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
>>>>
>>>> Guess, we could also make it week-8 instead of 9, and accept that 2034
>>>> would cut it too short (or be even more flexible and keep week-9 for
>>>> 2034 as an exception to the rule), but that's a single year... for 2029
>>>> and 2032 we still have one week buffer.
>>>>
>>>> For Christmas, we would still ensure that the release can go out on
>>>> time, too. The below Q3 target release date is 4 weeks after code
>>>> freeze. This is a little bit more tight, as we often need more than 4
>>>> weeks to get it done:
>>>>
>>>>>    ┌──────┬──────────────────────────┬────────────────┐
>>>>>    │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2026 │ Dec 9                    │ 16             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2027 │ Dec 15                   │ 10             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2028 │ Dec 13                   │ 12             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2029 │ Dec 12                   │ 13             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2030 │ Dec 11                   │ 14             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2031 │ Dec 10                   │ 15             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2032 │ Dec 8                    │ 17             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2033 │ Dec 14                   │ 11             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2034 │ Dec 13                   │ 12             │
>>>>>    ├──────┼──────────────────────────┼────────────────┤
>>>>>    │ 2035 │ Dec 12                   │ 13             │
>>>>>    └──────┴──────────────────────────┴────────────────┘
>>>>
>>>>
>>>>
>>>> For some years, we would still cut it a little bit tight, but we could
>>>> also apply an exception to the rule, to compensate for it. If we have
>>>> enough buffer room from LYN, we can start the Q1 release train a week
>>>> earlier, reducing the Q3->Q1 timeline by one week.
>>>>
>>>>
>>>> Curious to hear what people think about it. In the end, if we can agree
>>>> on high level rules like "start Q1 release with X buffer after LYN", and
>>>> "complete Q3 release with Y buffer before Christmas", and still aim for
>>>> an as balance week-allocation per release, we can just put together a
>>>> release plan for the next 10 years and be done with it.
>>>>
>>>>
>>>> -Matthias
>>>>
>>>>
>>>>
>>>> On 5/28/26 3:51 AM, Christo Lolov wrote:
>>>>> Hello,
>>>>>
>>>>> Thanks for pushing for this! I can confirm that the 4.2.0's release
>>>>> coincided with a reasonable amount of holidays, so I would favour a
>>>>> 3+2 schedule with releases in March, July and November.
>>>>>
>>>>> Best,
>>>>> Christo
>>>>>
>>>>> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
>>>>>>
>>>>>> Hi,
>>>>>>
>>>>>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
>>>>>> wrote:
>>>>>>
>>>>>>> I would go for 3 releases of equal lengths in March, July and November. I
>>>>>>> favour the 3+2 schedule.
>>>>>>>
>>>>>>
>>>>>> +1
>>>>>>
>>>>>> Thanks,
>>>>>> Ismael
>>>>
>>>>


Re: [DISCUSS] Tightening time-based releases

Posted by David Jacot <da...@apache.org>.
Hi all,

Have we converged on the schedule that we want to use yet? Based on
the last emails, it is not so clear to me. For your context, I am
trying to figure out the timeline for 4.5.

Best,
David

On Fri, May 29, 2026 at 1:03 PM Christo Lolov <ch...@gmail.com> wrote:
>
> I think we could be a bit more lax with the KIP freeze of 6.7.0 when
> 2034 comes around to account for LNY :)
>
> Joking aside, my preference between "KIP freeze would be Wednesday's,
> weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
> weeks" and "start Q1 release with X buffer after LYN and complete Q3
> release with Y buffer before Christmas" would be to start with the
> first and adjust by shifting by 1 week on individual basis the ones
> which are too close. I am okay with too close being defined as 2
> weeks. Based on your very detailed tables we seem to have only a few
> such outliers in the next decade.
>
> Best,
> Christo
>
> On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org> wrote:
> >
> > Matthias, you are definitely a master of scheduling! I'm +100 to your plans.
> >
> > Also, it is absolutely helpful to write down the exact release dates for the next few years at once. It will ease our anxiety a lot! I'd be more than happy to update the wiki with these dates once we all agree on this new schedule. For example:
> >
> > Version KIP Freeze      Feature Freeze  Code Freeze     Target Release
> > 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19, 2026
> > 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9, 2026
> > 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5, 2027
> > 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25, 2027
> > 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15, 2027
> > 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3, 2028
> > 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23, 2028
> > 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13, 2028
> >
> > Furthermore, let's do our best to avoid fixing release blockers under the Christmas tree :)
> >
> > On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> > > Thanks for the input.
> > >
> > > Overall, it seems that 3+2 weeks for feature freeze / code freeze
> > > windows gets a lot of support.
> > >
> > > However, the idea of dividing the release windows not into 3 equally
> > > long junks, but be more flexible to account for holiday seasons, seems
> > > not to get a lot of support? Not sure why... I personally think it would
> > > be good to take it into account, and I also think that Chia-Ping raises
> > > a very good point.
> > >
> > >
> > >
> > > @Chia-Ping: Other holidays are a great point. Thanks for raising it. I
> > > did actually briefly think about it, but it's difficult, and thus I just
> > > started with something simple as proposal to kick off a discussion.
> > >
> > > I think, Diwali is short enough to not make it a concern (similar to
> > > Thanksgiving)?
> > >
> > > For Lunar New Year, I agree that it would be good to plan for it.
> > > However, it seems it's shifting quite a lot year-over-year what makes it
> > > very difficult (w/o a huge window) to include?
> > >
> > > Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this
> > > would be 22 / 15 / 15 weeks per release? And two very short 3.5 month
> > > releases does not sound very appealing?
> > >
> > >
> > > I was playing with it a little bit more. Overall we have 52 weeks, so
> > > roughly 17.3 weeks per release. So we can naturally split it into
> > > 17/17/18 using the additional week for the holidays. But that's not good
> > > enough to take both Christmas and Lunar New Year into account. However,
> > > with a 16/16/20, we could make it work I believe, even if it does cut it
> > > tight sometimes... To be able to phrase a "simple rule", we can use ISO
> > > weeks:
> > >
> > > KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
> > > +3 weeks, and code freeze +2 weeks.
> > >
> > > Ie, we put the first KIP freeze after LNY with plenty of buffer for most
> > > years:
> > >
> > > >   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> > > >   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
> > > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > > >   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
> > > >   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> > >
> > > Guess, we could also make it week-8 instead of 9, and accept that 2034
> > > would cut it too short (or be even more flexible and keep week-9 for
> > > 2034 as an exception to the rule), but that's a single year... for 2029
> > > and 2032 we still have one week buffer.
> > >
> > > For Christmas, we would still ensure that the release can go out on
> > > time, too. The below Q3 target release date is 4 weeks after code
> > > freeze. This is a little bit more tight, as we often need more than 4
> > > weeks to get it done:
> > >
> > > >   ┌──────┬──────────────────────────┬────────────────┐
> > > >   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2026 │ Dec 9                    │ 16             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2027 │ Dec 15                   │ 10             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2028 │ Dec 13                   │ 12             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2029 │ Dec 12                   │ 13             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2030 │ Dec 11                   │ 14             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2031 │ Dec 10                   │ 15             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2032 │ Dec 8                    │ 17             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2033 │ Dec 14                   │ 11             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2034 │ Dec 13                   │ 12             │
> > > >   ├──────┼──────────────────────────┼────────────────┤
> > > >   │ 2035 │ Dec 12                   │ 13             │
> > > >   └──────┴──────────────────────────┴────────────────┘
> > >
> > >
> > >
> > > For some years, we would still cut it a little bit tight, but we could
> > > also apply an exception to the rule, to compensate for it. If we have
> > > enough buffer room from LYN, we can start the Q1 release train a week
> > > earlier, reducing the Q3->Q1 timeline by one week.
> > >
> > >
> > > Curious to hear what people think about it. In the end, if we can agree
> > > on high level rules like "start Q1 release with X buffer after LYN", and
> > > "complete Q3 release with Y buffer before Christmas", and still aim for
> > > an as balance week-allocation per release, we can just put together a
> > > release plan for the next 10 years and be done with it.
> > >
> > >
> > > -Matthias
> > >
> > >
> > >
> > > On 5/28/26 3:51 AM, Christo Lolov wrote:
> > > > Hello,
> > > >
> > > > Thanks for pushing for this! I can confirm that the 4.2.0's release
> > > > coincided with a reasonable amount of holidays, so I would favour a
> > > > 3+2 schedule with releases in March, July and November.
> > > >
> > > > Best,
> > > > Christo
> > > >
> > > > On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
> > > >>
> > > >> Hi,
> > > >>
> > > >> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
> > > >> wrote:
> > > >>
> > > >>> I would go for 3 releases of equal lengths in March, July and November. I
> > > >>> favour the 3+2 schedule.
> > > >>>
> > > >>
> > > >> +1
> > > >>
> > > >> Thanks,
> > > >> Ismael
> > >
> > >

Re: [DISCUSS] Tightening time-based releases

Posted by Christo Lolov <ch...@gmail.com>.
I think we could be a bit more lax with the KIP freeze of 6.7.0 when
2034 comes around to account for LNY :)

Joking aside, my preference between "KIP freeze would be Wednesday's,
weeks 9 / 25 / 41, with feature freeze +3 weeks, and code freeze +2
weeks" and "start Q1 release with X buffer after LYN and complete Q3
release with Y buffer before Christmas" would be to start with the
first and adjust by shifting by 1 week on individual basis the ones
which are too close. I am okay with too close being defined as 2
weeks. Based on your very detailed tables we seem to have only a few
such outliers in the next decade.

Best,
Christo

On Fri, 29 May 2026 at 08:35, Chia-Ping Tsai <ch...@apache.org> wrote:
>
> Matthias, you are definitely a master of scheduling! I'm +100 to your plans.
>
> Also, it is absolutely helpful to write down the exact release dates for the next few years at once. It will ease our anxiety a lot! I'd be more than happy to update the wiki with these dates once we all agree on this new schedule. For example:
>
> Version KIP Freeze      Feature Freeze  Code Freeze     Target Release
> 4.4.0   Jun 17, 2026    Jul 8, 2026     Jul 22, 2026    Aug 19, 2026
> 4.5.0   Oct 7, 2026     Oct 28, 2026    Nov 11, 2026    Dec 9, 2026
> 4.6.0   Mar 3, 2027     Mar 24, 2027    Apr 7, 2027     May 5, 2027
> 4.7.0   Jun 23, 2027    Jul 14, 2027    Jul 28, 2027    Aug 25, 2027
> 4.8.0   Oct 13, 2027    Nov 3, 2027     Nov 17, 2027    Dec 15, 2027
> 4.9.0   Mar 1, 2028     Mar 22, 2028    Apr 5, 2028     May 3, 2028
> 5.0.0   Jun 21, 2028    Jul 12, 2028    Jul 26, 2028    Aug 23, 2028
> 5.1.0   Oct 11, 2028    Nov 1, 2028     Nov 15, 2028    Dec 13, 2028
>
> Furthermore, let's do our best to avoid fixing release blockers under the Christmas tree :)
>
> On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> > Thanks for the input.
> >
> > Overall, it seems that 3+2 weeks for feature freeze / code freeze
> > windows gets a lot of support.
> >
> > However, the idea of dividing the release windows not into 3 equally
> > long junks, but be more flexible to account for holiday seasons, seems
> > not to get a lot of support? Not sure why... I personally think it would
> > be good to take it into account, and I also think that Chia-Ping raises
> > a very good point.
> >
> >
> >
> > @Chia-Ping: Other holidays are a great point. Thanks for raising it. I
> > did actually briefly think about it, but it's difficult, and thus I just
> > started with something simple as proposal to kick off a discussion.
> >
> > I think, Diwali is short enough to not make it a concern (similar to
> > Thanksgiving)?
> >
> > For Lunar New Year, I agree that it would be good to plan for it.
> > However, it seems it's shifting quite a lot year-over-year what makes it
> > very difficult (w/o a huge window) to include?
> >
> > Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this
> > would be 22 / 15 / 15 weeks per release? And two very short 3.5 month
> > releases does not sound very appealing?
> >
> >
> > I was playing with it a little bit more. Overall we have 52 weeks, so
> > roughly 17.3 weeks per release. So we can naturally split it into
> > 17/17/18 using the additional week for the holidays. But that's not good
> > enough to take both Christmas and Lunar New Year into account. However,
> > with a 16/16/20, we could make it work I believe, even if it does cut it
> > tight sometimes... To be able to phrase a "simple rule", we can use ISO
> > weeks:
> >
> > KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze
> > +3 weeks, and code freeze +2 weeks.
> >
> > Ie, we put the first KIP freeze after LNY with plenty of buffer for most
> > years:
> >
> > >   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> > >   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
> > >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> > >   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
> > >   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> >
> > Guess, we could also make it week-8 instead of 9, and accept that 2034
> > would cut it too short (or be even more flexible and keep week-9 for
> > 2034 as an exception to the rule), but that's a single year... for 2029
> > and 2032 we still have one week buffer.
> >
> > For Christmas, we would still ensure that the release can go out on
> > time, too. The below Q3 target release date is 4 weeks after code
> > freeze. This is a little bit more tight, as we often need more than 4
> > weeks to get it done:
> >
> > >   ┌──────┬──────────────────────────┬────────────────┐
> > >   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2026 │ Dec 9                    │ 16             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2027 │ Dec 15                   │ 10             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2028 │ Dec 13                   │ 12             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2029 │ Dec 12                   │ 13             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2030 │ Dec 11                   │ 14             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2031 │ Dec 10                   │ 15             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2032 │ Dec 8                    │ 17             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2033 │ Dec 14                   │ 11             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2034 │ Dec 13                   │ 12             │
> > >   ├──────┼──────────────────────────┼────────────────┤
> > >   │ 2035 │ Dec 12                   │ 13             │
> > >   └──────┴──────────────────────────┴────────────────┘
> >
> >
> >
> > For some years, we would still cut it a little bit tight, but we could
> > also apply an exception to the rule, to compensate for it. If we have
> > enough buffer room from LYN, we can start the Q1 release train a week
> > earlier, reducing the Q3->Q1 timeline by one week.
> >
> >
> > Curious to hear what people think about it. In the end, if we can agree
> > on high level rules like "start Q1 release with X buffer after LYN", and
> > "complete Q3 release with Y buffer before Christmas", and still aim for
> > an as balance week-allocation per release, we can just put together a
> > release plan for the next 10 years and be done with it.
> >
> >
> > -Matthias
> >
> >
> >
> > On 5/28/26 3:51 AM, Christo Lolov wrote:
> > > Hello,
> > >
> > > Thanks for pushing for this! I can confirm that the 4.2.0's release
> > > coincided with a reasonable amount of holidays, so I would favour a
> > > 3+2 schedule with releases in March, July and November.
> > >
> > > Best,
> > > Christo
> > >
> > > On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
> > >>
> > >> Hi,
> > >>
> > >> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
> > >> wrote:
> > >>
> > >>> I would go for 3 releases of equal lengths in March, July and November. I
> > >>> favour the 3+2 schedule.
> > >>>
> > >>
> > >> +1
> > >>
> > >> Thanks,
> > >> Ismael
> >
> >

Re: [DISCUSS] Tightening time-based releases

Posted by Chia-Ping Tsai <ch...@apache.org>.
Matthias, you are definitely a master of scheduling! I'm +100 to your plans.

Also, it is absolutely helpful to write down the exact release dates for the next few years at once. It will ease our anxiety a lot! I'd be more than happy to update the wiki with these dates once we all agree on this new schedule. For example:

Version	KIP Freeze	Feature Freeze	Code Freeze	Target Release
4.4.0	Jun 17, 2026	Jul 8, 2026	Jul 22, 2026	Aug 19, 2026
4.5.0	Oct 7, 2026	Oct 28, 2026	Nov 11, 2026	Dec 9, 2026
4.6.0	Mar 3, 2027	Mar 24, 2027	Apr 7, 2027	May 5, 2027
4.7.0	Jun 23, 2027	Jul 14, 2027	Jul 28, 2027	Aug 25, 2027
4.8.0	Oct 13, 2027	Nov 3, 2027	Nov 17, 2027	Dec 15, 2027
4.9.0	Mar 1, 2028	Mar 22, 2028	Apr 5, 2028	May 3, 2028
5.0.0	Jun 21, 2028	Jul 12, 2028	Jul 26, 2028	Aug 23, 2028
5.1.0	Oct 11, 2028	Nov 1, 2028	Nov 15, 2028	Dec 13, 2028

Furthermore, let's do our best to avoid fixing release blockers under the Christmas tree :)

On 2026/05/28 21:22:51 "Matthias J. Sax" wrote:
> Thanks for the input.
> 
> Overall, it seems that 3+2 weeks for feature freeze / code freeze 
> windows gets a lot of support.
> 
> However, the idea of dividing the release windows not into 3 equally 
> long junks, but be more flexible to account for holiday seasons, seems 
> not to get a lot of support? Not sure why... I personally think it would 
> be good to take it into account, and I also think that Chia-Ping raises 
> a very good point.
> 
> 
> 
> @Chia-Ping: Other holidays are a great point. Thanks for raising it. I 
> did actually briefly think about it, but it's difficult, and thus I just 
> started with something simple as proposal to kick off a discussion.
> 
> I think, Diwali is short enough to not make it a concern (similar to 
> Thanksgiving)?
> 
> For Lunar New Year, I agree that it would be good to plan for it. 
> However, it seems it's shifting quite a lot year-over-year what makes it 
> very difficult (w/o a huge window) to include?
> 
> Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this 
> would be 22 / 15 / 15 weeks per release? And two very short 3.5 month 
> releases does not sound very appealing?
> 
> 
> I was playing with it a little bit more. Overall we have 52 weeks, so 
> roughly 17.3 weeks per release. So we can naturally split it into 
> 17/17/18 using the additional week for the holidays. But that's not good 
> enough to take both Christmas and Lunar New Year into account. However, 
> with a 16/16/20, we could make it work I believe, even if it does cut it 
> tight sometimes... To be able to phrase a "simple rule", we can use ISO 
> weeks:
> 
> KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze 
> +3 weeks, and code freeze +2 weeks.
> 
> Ie, we put the first KIP freeze after LNY with plenty of buffer for most 
> years:
> 
> >   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
> >   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │                                                              
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
> >   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
> >   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
> >   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘
> 
> Guess, we could also make it week-8 instead of 9, and accept that 2034 
> would cut it too short (or be even more flexible and keep week-9 for 
> 2034 as an exception to the rule), but that's a single year... for 2029 
> and 2032 we still have one week buffer.
> 
> For Christmas, we would still ensure that the release can go out on 
> time, too. The below Q3 target release date is 4 weeks after code 
> freeze. This is a little bit more tight, as we often need more than 4 
> weeks to get it done:
> 
> >   ┌──────┬──────────────────────────┬────────────────┐
> >   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2026 │ Dec 9                    │ 16             │                                                                              
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2027 │ Dec 15                   │ 10             │
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2028 │ Dec 13                   │ 12             │
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2029 │ Dec 12                   │ 13             │
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2030 │ Dec 11                   │ 14             │                                                                              
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2031 │ Dec 10                   │ 15             │
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2032 │ Dec 8                    │ 17             │
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2033 │ Dec 14                   │ 11             │                                                                              
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2034 │ Dec 13                   │ 12             │
> >   ├──────┼──────────────────────────┼────────────────┤
> >   │ 2035 │ Dec 12                   │ 13             │
> >   └──────┴──────────────────────────┴────────────────┘
> 
> 
> 
> For some years, we would still cut it a little bit tight, but we could 
> also apply an exception to the rule, to compensate for it. If we have 
> enough buffer room from LYN, we can start the Q1 release train a week 
> earlier, reducing the Q3->Q1 timeline by one week.
> 
> 
> Curious to hear what people think about it. In the end, if we can agree 
> on high level rules like "start Q1 release with X buffer after LYN", and 
> "complete Q3 release with Y buffer before Christmas", and still aim for 
> an as balance week-allocation per release, we can just put together a 
> release plan for the next 10 years and be done with it.
> 
> 
> -Matthias
> 
> 
> 
> On 5/28/26 3:51 AM, Christo Lolov wrote:
> > Hello,
> > 
> > Thanks for pushing for this! I can confirm that the 4.2.0's release
> > coincided with a reasonable amount of holidays, so I would favour a
> > 3+2 schedule with releases in March, July and November.
> > 
> > Best,
> > Christo
> > 
> > On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
> >>
> >> Hi,
> >>
> >> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
> >> wrote:
> >>
> >>> I would go for 3 releases of equal lengths in March, July and November. I
> >>> favour the 3+2 schedule.
> >>>
> >>
> >> +1
> >>
> >> Thanks,
> >> Ismael
> 
> 

Re: [DISCUSS] Tightening time-based releases

Posted by "Matthias J. Sax" <mj...@apache.org>.
Thanks for the input.

Overall, it seems that 3+2 weeks for feature freeze / code freeze 
windows gets a lot of support.

However, the idea of dividing the release windows not into 3 equally 
long junks, but be more flexible to account for holiday seasons, seems 
not to get a lot of support? Not sure why... I personally think it would 
be good to take it into account, and I also think that Chia-Ping raises 
a very good point.



@Chia-Ping: Other holidays are a great point. Thanks for raising it. I 
did actually briefly think about it, but it's difficult, and thus I just 
started with something simple as proposal to kick off a discussion.

I think, Diwali is short enough to not make it a concern (similar to 
Thanksgiving)?

For Lunar New Year, I agree that it would be good to plan for it. 
However, it seems it's shifting quite a lot year-over-year what makes it 
very difficult (w/o a huge window) to include?

Overall, a 5 / 3.5 / 3.5 schedule seems to be quite unbalanced (this 
would be 22 / 15 / 15 weeks per release? And two very short 3.5 month 
releases does not sound very appealing?


I was playing with it a little bit more. Overall we have 52 weeks, so 
roughly 17.3 weeks per release. So we can naturally split it into 
17/17/18 using the additional week for the holidays. But that's not good 
enough to take both Christmas and Lunar New Year into account. However, 
with a 16/16/20, we could make it work I believe, even if it does cut it 
tight sometimes... To be able to phrase a "simple rule", we can use ISO 
weeks:

KIP freeze would be Wednesday's, weeks 9 / 25 / 41, with feature freeze 
+3 weeks, and code freeze +2 weeks.

Ie, we put the first KIP freeze after LNY with plenty of buffer for most 
years:

>   ┌──────┬────────┬────────────────────────────┬───────────────────────────────────────┐
>   │ Year │  LNY   │ Q1 KIP freeze (week 9 Wed) │                Buffer                 │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2027 │ Feb 6  │ Mar 3                      │ 25 days                               │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2028 │ Jan 26 │ Mar 1                      │ 34 days                               │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2029 │ Feb 13 │ Feb 28                     │ 15 days                               │                                                              
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2030 │ Feb 3  │ Feb 27                     │ 24 days                               │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2031 │ Jan 23 │ Feb 26                     │ 34 days                               │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2032 │ Feb 11 │ Feb 25                     │ 14 days                               │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2033 │ Jan 31 │ Mar 2                      │ 30 days                               │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2034 │ Feb 19 │ Mar 1                      │ 10 days ← tightest with realistic LNY │
>   ├──────┼────────┼────────────────────────────┼───────────────────────────────────────┤
>   │ 2035 │ Feb 8  │ Feb 28                     │ 20 days                               │
>   └──────┴────────┴────────────────────────────┴───────────────────────────────────────┘

Guess, we could also make it week-8 instead of 9, and accept that 2034 
would cut it too short (or be even more flexible and keep week-9 for 
2034 as an exception to the rule), but that's a single year... for 2029 
and 2032 we still have one week buffer.

For Christmas, we would still ensure that the release can go out on 
time, too. The below Q3 target release date is 4 weeks after code 
freeze. This is a little bit more tight, as we often need more than 4 
weeks to get it done:

>   ┌──────┬──────────────────────────┬────────────────┐
>   │ Year │ Q3 release (week 50 Wed) │ Days to Dec 25 │
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2026 │ Dec 9                    │ 16             │                                                                              
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2027 │ Dec 15                   │ 10             │
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2028 │ Dec 13                   │ 12             │
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2029 │ Dec 12                   │ 13             │
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2030 │ Dec 11                   │ 14             │                                                                              
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2031 │ Dec 10                   │ 15             │
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2032 │ Dec 8                    │ 17             │
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2033 │ Dec 14                   │ 11             │                                                                              
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2034 │ Dec 13                   │ 12             │
>   ├──────┼──────────────────────────┼────────────────┤
>   │ 2035 │ Dec 12                   │ 13             │
>   └──────┴──────────────────────────┴────────────────┘



For some years, we would still cut it a little bit tight, but we could 
also apply an exception to the rule, to compensate for it. If we have 
enough buffer room from LYN, we can start the Q1 release train a week 
earlier, reducing the Q3->Q1 timeline by one week.


Curious to hear what people think about it. In the end, if we can agree 
on high level rules like "start Q1 release with X buffer after LYN", and 
"complete Q3 release with Y buffer before Christmas", and still aim for 
an as balance week-allocation per release, we can just put together a 
release plan for the next 10 years and be done with it.


-Matthias



On 5/28/26 3:51 AM, Christo Lolov wrote:
> Hello,
> 
> Thanks for pushing for this! I can confirm that the 4.2.0's release
> coincided with a reasonable amount of holidays, so I would favour a
> 3+2 schedule with releases in March, July and November.
> 
> Best,
> Christo
> 
> On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
>>
>> Hi,
>>
>> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
>> wrote:
>>
>>> I would go for 3 releases of equal lengths in March, July and November. I
>>> favour the 3+2 schedule.
>>>
>>
>> +1
>>
>> Thanks,
>> Ismael


Re: [DISCUSS] Tightening time-based releases

Posted by Christo Lolov <cl...@apache.org>.
Hello,

Thanks for pushing for this! I can confirm that the 4.2.0's release
coincided with a reasonable amount of holidays, so I would favour a
3+2 schedule with releases in March, July and November.

Best,
Christo

On Thu, 28 May 2026 at 07:05, Ismael Juma <me...@ismaeljuma.com> wrote:
>
> Hi,
>
> On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
> wrote:
>
> > I would go for 3 releases of equal lengths in March, July and November. I
> > favour the 3+2 schedule.
> >
>
> +1
>
> Thanks,
> Ismael

Re: [DISCUSS] Tightening time-based releases

Posted by Ismael Juma <me...@ismaeljuma.com>.
Hi,

On Wed, May 27, 2026 at 12:44 AM Andrew Schofield <sc...@apache.org>
wrote:

> I would go for 3 releases of equal lengths in March, July and November. I
> favour the 3+2 schedule.
>

+1

Thanks,
Ismael

Re: [DISCUSS] Tightening time-based releases

Posted by Andrew Schofield <sc...@apache.org>.
Hi Matthias,
Thanks for driving this. I'm a bit less concerned about it than you, but stronger rules would eliminate the typical release schedule alignment discussions we've seen in all of the recent releases.

I would go for 3 releases of equal lengths in March, July and November. I favour the 3+2 schedule.

Thanks,
Andrew

On 2026/05/27 05:43:04 "Matthias J. Sax" wrote:
> Hello,
> 
> I would like to start a discussion about our time-based release 
> schedule. We have a discussion (that I personally find a little bit 
> annoying and tedious) about release plan dates on every single release 
> (the 4.4.0 discussion, is just another example). And we still often end 
> up with none-optimal time-frames (4.4 might also be an example here -- 
> not by itself, but the impact on 4.5 release timeline -- cf my latest 
> reply on the other thread).
> 
> Given that we actually plan to have time-based releases, I think we 
> should agree on a more fixed schedule, which allows us to project all 
> release dates (including KIP/feature/code freeze) into the future and 
> just be done with it. -- And even if we need an exception from the rule, 
> we can easily get back on track on future release afterwards.
> 
> I was also looking into historic data, and while we overall don't do too 
> bad to stay on track for 3 release per year, we had to pay the price of 
> a few short release cycles to "shift back" (the previous 4.3 release is 
> a good example of it), while we still did shift somewhat forward 
> overall. These short releases cause even more pain IMHO.
> 
> To avoid problems with the holiday season end of the year, I would like 
> to propose the following fixed release cadence (this is not my personal 
> favorite, but let's start here):
> 
> 1. KIP freeze is the second Wednesday of January, May, and September.
> 2. Feature freeze = KIP freeze + 3 weeks.
> 3. Code freeze = Feature freeze + 3 weeks.
> 4. Release target = Code freeze + 4 weeks.
> 
> While we never officially formalized the 3+3 weeks between KIP/feature 
> and feature/code freeze dates (and we used shorter time windows in the 
> past, ie, 2+2 weeks), 3+3 established itself since 4.0 release -- I am 
> totally open for other time frames, but believe we should not go shorter 
> than 2+2 weeks. I would also not make them longer than 3+3, as it would 
> just give us the subjective impression that releases are closer to each 
> other.
> 
> A 3+2 schedule would actually be my personal preference. For KIPs, 
> giving only 2 weeks to complete them is tight, even for simple KIPs. And 
> for bug-fixes, I would believe that 2 weeks are enough. Making it 
> overall 5 week instead of 6, also helps to keep releases 1 week "further 
> apart" from each other (not really, but we know how reality works, given 
> our subjective nature).
> 
> The above schedule implies that we would have target release dates late 
> March, July, and November. Even if we need more than 4 weeks after code 
> freeze, we still have some head room in the November release before the 
> holidays (we might hit Thanksgiving, but I think that's ok). For the KIP 
> freeze in January, it should also give enough time after the holidays to 
> complete KIPs and get them accepted (w/o being too close to the 
> holidays, eg, "first Wednesday in January" would effectively mean, that 
> KIPs must get voted before the holiday break).
> 
> If we want to take the holiday season into account even more, we could 
> also make the timeline between "release 3" and "release 1 of the 
> following year" two weeks longer, and the other two timelines 1 week 
> shorter. This would give us the following KIP freeze deadlines:
> 
>   - third Wednesday of January
>   - second Wednesday of May
>   - first Wednesday of September
> 
> It also implies, that Thanksgiving would be less of an issue for 
> "release 3". This schedule is what I would personally prefer, instead of 
> an 100% equal distribution of our 52 weeks across all three releases.
> 
> 
> 
> For the upcoming 4.4 release, we should make it a 5-month cycle 
> targeting October instead of September, and we would also have a 
> slightly longer cycle for 4.6, which we should only start in January. 
> However, starting with 4.6 we would enter the above proposed fixed 
> timeline, putting us into a much better position going forward IMHO.
> 
> 
> Looking forward to your thoughts on this idea.
> 
> 
> -Matthias
> 
> 
>