[Important] Merge Freeze for release

[Ian Maxon <im...@uci.edu>]

Hey everyone,
Just as a reminder, please hold off on submitting things to master (or at
least ask here if you'd prefer) so we can get the release out. Right now
all the features are in but I'm still discussing with Chris and Taewoo
about how to best address the really ugly merges for open branches that
will come out from changing the package from edu.uci.ics to org.apache. If
we submit other patches while still trying to figure this out it makes
things harder than necessary.

Thanks,
- Ian

Re: [Important] Merge Freeze for release

[Ian Maxon <im...@uci.edu>]

Update:
The YARN issue is solved, it actually was just related to some
left-over serialized java objects in the build server that break due
to the class path change. Didn't need a patch (but I should make one
to prevent the files from being leftover). The only thing we're
blocked on is the Feeds docs.

-Ian

On Wed, Aug 26, 2015 at 6:01 AM, Chen Li <ch...@gmail.com> wrote:
> I suggest we do not rush adding untested features before the release.
>
> Chen
>
> On Wed, Aug 26, 2015 at 8:13 AM, Ian Maxon <im...@uci.edu> wrote:
>
>> Most likely it will miss the first pass :( Unless it can be rebased and
>> revised by someone in the next day or so, who isn't off at VLDB soon...
>>
>> -Ian
>> On Aug 26, 2015 2:21 AM, "Heri Ramampiaro" <he...@gmail.com> wrote:
>>
>> > I just wonder if the internal functions would “miss” this release.
>> >
>> > At least, an important change especially (bug fixes) in the
>> "record-merge"
>> > function
>> > would be nice to have in this release. I believe the current version
>> > of this function in the master only supports closed records and will fail
>> > when
>> > used with open records…
>> >
>> > Best,
>> > -heri
>> >
>> >
>> > > On Aug 25, 2015, at 7:44 AM, Ian Maxon <im...@uci.edu> wrote:
>> > >
>> > > That's kind of what I assumed as well until this afternoon. Taewoo
>> tried
>> > > merging the candidate changes, and it was unexpectedly difficult. As it
>> > > turns out, there's some special ugliness that can be encountered if one
>> > > tries to merge in some usually sane ways (mainly just 'git merge
>> master'
>> > or
>> > > similar, Chris found that 'git gerrit update' is more likely to be
>> > > cleaner). I'll summarize what Chris and I found in another mail.
>> > >
>> > > - Ian
>> > >
>> > > On Mon, Aug 24, 2015 at 8:16 PM, Till Westmann <ti...@apache.org>
>> wrote:
>> > >
>> > >> I'm wondering if those merges really have to be bad. It seems that
>> most
>> > >> changes in the source files would be in the imports and thus there
>> > wouldn't
>> > >> be too many conflicts with "regular" code changes.
>> > >> Is that right or am I missing an important point?
>> > >> Did you try those merges?
>> > >>
>> > >> Thanks,
>> > >> Till
>> > >>
>> > >>> On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
>> > >>>
>> > >>> Hey everyone,
>> > >>> Just as a reminder, please hold off on submitting things to master
>> (or
>> > at
>> > >>> least ask here if you'd prefer) so we can get the release out. Right
>> > now
>> > >>> all the features are in but I'm still discussing with Chris and
>> Taewoo
>> > >>> about how to best address the really ugly merges for open branches
>> that
>> > >>> will come out from changing the package from edu.uci.ics to
>> org.apache.
>> > >> If
>> > >>> we submit other patches while still trying to figure this out it
>> makes
>> > >>> things harder than necessary.
>> > >>>
>> > >>> Thanks,
>> > >>> - Ian
>> > >>
>> >
>> >
>>

Re: [Important] Merge Freeze for release

[Chen Li <ch...@gmail.com>]

I suggest we do not rush adding untested features before the release.

Chen

On Wed, Aug 26, 2015 at 8:13 AM, Ian Maxon <im...@uci.edu> wrote:

> Most likely it will miss the first pass :( Unless it can be rebased and
> revised by someone in the next day or so, who isn't off at VLDB soon...
>
> -Ian
> On Aug 26, 2015 2:21 AM, "Heri Ramampiaro" <he...@gmail.com> wrote:
>
> > I just wonder if the internal functions would “miss” this release.
> >
> > At least, an important change especially (bug fixes) in the
> "record-merge"
> > function
> > would be nice to have in this release. I believe the current version
> > of this function in the master only supports closed records and will fail
> > when
> > used with open records…
> >
> > Best,
> > -heri
> >
> >
> > > On Aug 25, 2015, at 7:44 AM, Ian Maxon <im...@uci.edu> wrote:
> > >
> > > That's kind of what I assumed as well until this afternoon. Taewoo
> tried
> > > merging the candidate changes, and it was unexpectedly difficult. As it
> > > turns out, there's some special ugliness that can be encountered if one
> > > tries to merge in some usually sane ways (mainly just 'git merge
> master'
> > or
> > > similar, Chris found that 'git gerrit update' is more likely to be
> > > cleaner). I'll summarize what Chris and I found in another mail.
> > >
> > > - Ian
> > >
> > > On Mon, Aug 24, 2015 at 8:16 PM, Till Westmann <ti...@apache.org>
> wrote:
> > >
> > >> I'm wondering if those merges really have to be bad. It seems that
> most
> > >> changes in the source files would be in the imports and thus there
> > wouldn't
> > >> be too many conflicts with "regular" code changes.
> > >> Is that right or am I missing an important point?
> > >> Did you try those merges?
> > >>
> > >> Thanks,
> > >> Till
> > >>
> > >>> On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
> > >>>
> > >>> Hey everyone,
> > >>> Just as a reminder, please hold off on submitting things to master
> (or
> > at
> > >>> least ask here if you'd prefer) so we can get the release out. Right
> > now
> > >>> all the features are in but I'm still discussing with Chris and
> Taewoo
> > >>> about how to best address the really ugly merges for open branches
> that
> > >>> will come out from changing the package from edu.uci.ics to
> org.apache.
> > >> If
> > >>> we submit other patches while still trying to figure this out it
> makes
> > >>> things harder than necessary.
> > >>>
> > >>> Thanks,
> > >>> - Ian
> > >>
> >
> >
>

Re: [Important] Merge Freeze for release

[Ian Maxon <im...@uci.edu>]

Most likely it will miss the first pass :( Unless it can be rebased and
revised by someone in the next day or so, who isn't off at VLDB soon...

-Ian
On Aug 26, 2015 2:21 AM, "Heri Ramampiaro" <he...@gmail.com> wrote:

> I just wonder if the internal functions would “miss” this release.
>
> At least, an important change especially (bug fixes) in the "record-merge"
> function
> would be nice to have in this release. I believe the current version
> of this function in the master only supports closed records and will fail
> when
> used with open records…
>
> Best,
> -heri
>
>
> > On Aug 25, 2015, at 7:44 AM, Ian Maxon <im...@uci.edu> wrote:
> >
> > That's kind of what I assumed as well until this afternoon. Taewoo tried
> > merging the candidate changes, and it was unexpectedly difficult. As it
> > turns out, there's some special ugliness that can be encountered if one
> > tries to merge in some usually sane ways (mainly just 'git merge master'
> or
> > similar, Chris found that 'git gerrit update' is more likely to be
> > cleaner). I'll summarize what Chris and I found in another mail.
> >
> > - Ian
> >
> > On Mon, Aug 24, 2015 at 8:16 PM, Till Westmann <ti...@apache.org> wrote:
> >
> >> I'm wondering if those merges really have to be bad. It seems that most
> >> changes in the source files would be in the imports and thus there
> wouldn't
> >> be too many conflicts with "regular" code changes.
> >> Is that right or am I missing an important point?
> >> Did you try those merges?
> >>
> >> Thanks,
> >> Till
> >>
> >>> On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
> >>>
> >>> Hey everyone,
> >>> Just as a reminder, please hold off on submitting things to master (or
> at
> >>> least ask here if you'd prefer) so we can get the release out. Right
> now
> >>> all the features are in but I'm still discussing with Chris and Taewoo
> >>> about how to best address the really ugly merges for open branches that
> >>> will come out from changing the package from edu.uci.ics to org.apache.
> >> If
> >>> we submit other patches while still trying to figure this out it makes
> >>> things harder than necessary.
> >>>
> >>> Thanks,
> >>> - Ian
> >>
>
>

Re: [Important] Merge Freeze for release

[Heri Ramampiaro <he...@gmail.com>]

I just wonder if the internal functions would “miss” this release. 

At least, an important change especially (bug fixes) in the "record-merge" function
would be nice to have in this release. I believe the current version
of this function in the master only supports closed records and will fail when
used with open records… 

Best,
-heri


> On Aug 25, 2015, at 7:44 AM, Ian Maxon <im...@uci.edu> wrote:
> 
> That's kind of what I assumed as well until this afternoon. Taewoo tried
> merging the candidate changes, and it was unexpectedly difficult. As it
> turns out, there's some special ugliness that can be encountered if one
> tries to merge in some usually sane ways (mainly just 'git merge master' or
> similar, Chris found that 'git gerrit update' is more likely to be
> cleaner). I'll summarize what Chris and I found in another mail.
> 
> - Ian
> 
> On Mon, Aug 24, 2015 at 8:16 PM, Till Westmann <ti...@apache.org> wrote:
> 
>> I'm wondering if those merges really have to be bad. It seems that most
>> changes in the source files would be in the imports and thus there wouldn't
>> be too many conflicts with "regular" code changes.
>> Is that right or am I missing an important point?
>> Did you try those merges?
>> 
>> Thanks,
>> Till
>> 
>>> On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
>>> 
>>> Hey everyone,
>>> Just as a reminder, please hold off on submitting things to master (or at
>>> least ask here if you'd prefer) so we can get the release out. Right now
>>> all the features are in but I'm still discussing with Chris and Taewoo
>>> about how to best address the really ugly merges for open branches that
>>> will come out from changing the package from edu.uci.ics to org.apache.
>> If
>>> we submit other patches while still trying to figure this out it makes
>>> things harder than necessary.
>>> 
>>> Thanks,
>>> - Ian
>> 

Re: [Important] Merge Freeze for release

[Raman Grover <ra...@gmail.com>]

I have been working on the feed related stuff...

Stay tuned! I should have an update soon!

On Tue, Aug 25, 2015 at 8:12 PM, Ian Maxon <im...@uci.edu> wrote:

> Last two things that are in the way release wise (everything else is in
> master):
>
> - YARN tests seem to be dying on fulliautomatix but not within the typical
> Docker container build, I'll have to investigate that ASAP
> - What to do about the Feeds tutorial (leave it half-open or wait)
>
> -Ian
>
> On Mon, Aug 24, 2015 at 10:44 PM, Ian Maxon <im...@uci.edu> wrote:
>
> > That's kind of what I assumed as well until this afternoon. Taewoo tried
> > merging the candidate changes, and it was unexpectedly difficult. As it
> > turns out, there's some special ugliness that can be encountered if one
> > tries to merge in some usually sane ways (mainly just 'git merge master'
> or
> > similar, Chris found that 'git gerrit update' is more likely to be
> > cleaner). I'll summarize what Chris and I found in another mail.
> >
> > - Ian
> >
> > On Mon, Aug 24, 2015 at 8:16 PM, Till Westmann <ti...@apache.org> wrote:
> >
> >> I'm wondering if those merges really have to be bad. It seems that most
> >> changes in the source files would be in the imports and thus there
> wouldn't
> >> be too many conflicts with "regular" code changes.
> >> Is that right or am I missing an important point?
> >> Did you try those merges?
> >>
> >> Thanks,
> >> Till
> >>
> >> > On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
> >> >
> >> > Hey everyone,
> >> > Just as a reminder, please hold off on submitting things to master (or
> >> at
> >> > least ask here if you'd prefer) so we can get the release out. Right
> now
> >> > all the features are in but I'm still discussing with Chris and Taewoo
> >> > about how to best address the really ugly merges for open branches
> that
> >> > will come out from changing the package from edu.uci.ics to
> org.apache.
> >> If
> >> > we submit other patches while still trying to figure this out it makes
> >> > things harder than necessary.
> >> >
> >> > Thanks,
> >> > - Ian
> >>
> >
> >
>



-- 
Raman

Re: [Important] Merge Freeze for release

[Ian Maxon <im...@uci.edu>]

Last two things that are in the way release wise (everything else is in
master):

- YARN tests seem to be dying on fulliautomatix but not within the typical
Docker container build, I'll have to investigate that ASAP
- What to do about the Feeds tutorial (leave it half-open or wait)

-Ian

On Mon, Aug 24, 2015 at 10:44 PM, Ian Maxon <im...@uci.edu> wrote:

> That's kind of what I assumed as well until this afternoon. Taewoo tried
> merging the candidate changes, and it was unexpectedly difficult. As it
> turns out, there's some special ugliness that can be encountered if one
> tries to merge in some usually sane ways (mainly just 'git merge master' or
> similar, Chris found that 'git gerrit update' is more likely to be
> cleaner). I'll summarize what Chris and I found in another mail.
>
> - Ian
>
> On Mon, Aug 24, 2015 at 8:16 PM, Till Westmann <ti...@apache.org> wrote:
>
>> I'm wondering if those merges really have to be bad. It seems that most
>> changes in the source files would be in the imports and thus there wouldn't
>> be too many conflicts with "regular" code changes.
>> Is that right or am I missing an important point?
>> Did you try those merges?
>>
>> Thanks,
>> Till
>>
>> > On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
>> >
>> > Hey everyone,
>> > Just as a reminder, please hold off on submitting things to master (or
>> at
>> > least ask here if you'd prefer) so we can get the release out. Right now
>> > all the features are in but I'm still discussing with Chris and Taewoo
>> > about how to best address the really ugly merges for open branches that
>> > will come out from changing the package from edu.uci.ics to org.apache.
>> If
>> > we submit other patches while still trying to figure this out it makes
>> > things harder than necessary.
>> >
>> > Thanks,
>> > - Ian
>>
>
>

Re: [Important] Merge Freeze for release

[Ian Maxon <im...@uci.edu>]

That's kind of what I assumed as well until this afternoon. Taewoo tried
merging the candidate changes, and it was unexpectedly difficult. As it
turns out, there's some special ugliness that can be encountered if one
tries to merge in some usually sane ways (mainly just 'git merge master' or
similar, Chris found that 'git gerrit update' is more likely to be
cleaner). I'll summarize what Chris and I found in another mail.

- Ian

On Mon, Aug 24, 2015 at 8:16 PM, Till Westmann <ti...@apache.org> wrote:

> I'm wondering if those merges really have to be bad. It seems that most
> changes in the source files would be in the imports and thus there wouldn't
> be too many conflicts with "regular" code changes.
> Is that right or am I missing an important point?
> Did you try those merges?
>
> Thanks,
> Till
>
> > On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
> >
> > Hey everyone,
> > Just as a reminder, please hold off on submitting things to master (or at
> > least ask here if you'd prefer) so we can get the release out. Right now
> > all the features are in but I'm still discussing with Chris and Taewoo
> > about how to best address the really ugly merges for open branches that
> > will come out from changing the package from edu.uci.ics to org.apache.
> If
> > we submit other patches while still trying to figure this out it makes
> > things harder than necessary.
> >
> > Thanks,
> > - Ian
>

Re: [Important] Merge Freeze for release

[Till Westmann <ti...@apache.org>]

I'm wondering if those merges really have to be bad. It seems that most changes in the source files would be in the imports and thus there wouldn't be too many conflicts with "regular" code changes. 
Is that right or am I missing an important point?
Did you try those merges?

Thanks,
Till

> On Aug 24, 2015, at 18:47, Ian Maxon <im...@uci.edu> wrote:
> 
> Hey everyone,
> Just as a reminder, please hold off on submitting things to master (or at
> least ask here if you'd prefer) so we can get the release out. Right now
> all the features are in but I'm still discussing with Chris and Taewoo
> about how to best address the really ugly merges for open branches that
> will come out from changing the package from edu.uci.ics to org.apache. If
> we submit other patches while still trying to figure this out it makes
> things harder than necessary.
> 
> Thanks,
> - Ian