You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@pulsar.apache.org by Enrico Olivelli <eo...@gmail.com> on 2021/09/23 14:51:07 UTC
Problems with PIP ID Management
Hello,
Today I had to rename my PIP from PIP-93 to PIP-95 because PIP-93 was
picked up.
Then I realised that PIP95 already exists on GitHub, then I moved to PIP-97.
There are multiple PIP-95 on the Wiki page.
https://github.com/apache/pulsar/wiki
We need a way to assign new PIP ids
What about sending an email to dev in order to ask for a new PIP id ?
Only committers have write privileges to the Wiki pages, so any PIP id
assignment needs a sponsor that assigns the PIP id and creates the wiki
page.
I propose to add a link to the PIP issue in the wiki page until the PIP is
accepted,
because we recently decided to run PIP discussions as GH issues
Does anyone volunteer to clean up the PIP list page ?
Regards
Enrico
Re: Problems with PIP ID Management
Posted by Michael Marshall <mi...@gmail.com>.
Thanks for your feedback Enrico.
I'd appreciate any other thoughts on this process since I am proposing
a change to a recently implemented PIP.
Also, if we like this new direction, I think we should update our
GitHub issue template to communicate this new process. I already have
a PR for my proposed changes:
https://github.com/apache/pulsar/pull/12176. Please take a look, if
you're able.
Thanks,
Michael
On Sat, Sep 25, 2021 at 1:29 AM Enrico Olivelli <eo...@gmail.com> wrote:
>
> Michael,
>
> Il Ven 24 Set 2021, 19:47 Michael Marshall <mi...@gmail.com> ha
> scritto:
>
> > Hi Enrico,
> >
> > Thank you for raising this concern. Apparently acquiring a distributed
> > lock on a PIP number is non-trivial :)
> >
>
> We could use the first 3 digits of a hash computed on the title of the PIP.
> Just joking:)
>
>
> > > What about sending an email to dev in order to ask for a new PIP id ?
> >
> > In this design, is it up to a committer (someone with write access to
> > the Wiki page) to reply to the request with a PIP number? I think this
> > could lead to confusion, as no one person would have the job and there
> > could be delays in responding.
> >
> > Instead of receiving a PIP number, I think we could ask users to
> > inspect the mailing list archive to determine the next PIP number.
> >
>
> This is a good idea.
> And it perfectly matches the rule that in the ASF is the main source of
> truth.
> So I totally support this idea
>
>
>
>
> Then, the mailing list is the source of truth for PIP numbers, and
> > there isn't any delay. This protocol can still have collisions if
> > multiple people send PIPs at the same time. However, I think these
> > collisions could be resolved quickly: the earliest email in the
> > archive wins, and others need to pick the next number(s).
> >
> > Also, this raises a secondary question for me. In our new PIP process,
> > are we copying PIPs from GitHub issues into the GitHub wiki?
> >
> > In order to push this conversation forward, I wrote a PR with my
> > proposed solution to the problem:
> > https://github.com/apache/pulsar/pull/12176. I am not attached to my
> > solution per se, but I do think a solution to this problem should
> > include an update to the GitHub template to help new PIP writers
> > easily understand the process.
> >
>
> Agreed
>
> Thanks
> Enrico
>
>
> > > Does anyone volunteer to clean up the PIP list page ?
> > Once we settle on a protocol, we will need to clean up, as now we have
> > two PIP 97s.
> >
> > Thanks,
> > Michael
> >
> > On Thu, Sep 23, 2021 at 9:51 AM Enrico Olivelli <eo...@gmail.com>
> > wrote:
> > >
> > > Hello,
> > > Today I had to rename my PIP from PIP-93 to PIP-95 because PIP-93 was
> > > picked up.
> > > Then I realised that PIP95 already exists on GitHub, then I moved to
> > PIP-97.
> > >
> > > There are multiple PIP-95 on the Wiki page.
> > >
> > > https://github.com/apache/pulsar/wiki
> > >
> > > We need a way to assign new PIP ids
> > >
> > > What about sending an email to dev in order to ask for a new PIP id ?
> > >
> > > Only committers have write privileges to the Wiki pages, so any PIP id
> > > assignment needs a sponsor that assigns the PIP id and creates the wiki
> > > page.
> > >
> > > I propose to add a link to the PIP issue in the wiki page until the PIP
> > is
> > > accepted,
> > > because we recently decided to run PIP discussions as GH issues
> > >
> > > Does anyone volunteer to clean up the PIP list page ?
> > >
> > > Regards
> > > Enrico
> >
Re: Problems with PIP ID Management
Posted by Enrico Olivelli <eo...@gmail.com>.
Michael,
Il Ven 24 Set 2021, 19:47 Michael Marshall <mi...@gmail.com> ha
scritto:
> Hi Enrico,
>
> Thank you for raising this concern. Apparently acquiring a distributed
> lock on a PIP number is non-trivial :)
>
We could use the first 3 digits of a hash computed on the title of the PIP.
Just joking:)
> > What about sending an email to dev in order to ask for a new PIP id ?
>
> In this design, is it up to a committer (someone with write access to
> the Wiki page) to reply to the request with a PIP number? I think this
> could lead to confusion, as no one person would have the job and there
> could be delays in responding.
>
> Instead of receiving a PIP number, I think we could ask users to
> inspect the mailing list archive to determine the next PIP number.
>
This is a good idea.
And it perfectly matches the rule that in the ASF is the main source of
truth.
So I totally support this idea
Then, the mailing list is the source of truth for PIP numbers, and
> there isn't any delay. This protocol can still have collisions if
> multiple people send PIPs at the same time. However, I think these
> collisions could be resolved quickly: the earliest email in the
> archive wins, and others need to pick the next number(s).
>
> Also, this raises a secondary question for me. In our new PIP process,
> are we copying PIPs from GitHub issues into the GitHub wiki?
>
> In order to push this conversation forward, I wrote a PR with my
> proposed solution to the problem:
> https://github.com/apache/pulsar/pull/12176. I am not attached to my
> solution per se, but I do think a solution to this problem should
> include an update to the GitHub template to help new PIP writers
> easily understand the process.
>
Agreed
Thanks
Enrico
> > Does anyone volunteer to clean up the PIP list page ?
> Once we settle on a protocol, we will need to clean up, as now we have
> two PIP 97s.
>
> Thanks,
> Michael
>
> On Thu, Sep 23, 2021 at 9:51 AM Enrico Olivelli <eo...@gmail.com>
> wrote:
> >
> > Hello,
> > Today I had to rename my PIP from PIP-93 to PIP-95 because PIP-93 was
> > picked up.
> > Then I realised that PIP95 already exists on GitHub, then I moved to
> PIP-97.
> >
> > There are multiple PIP-95 on the Wiki page.
> >
> > https://github.com/apache/pulsar/wiki
> >
> > We need a way to assign new PIP ids
> >
> > What about sending an email to dev in order to ask for a new PIP id ?
> >
> > Only committers have write privileges to the Wiki pages, so any PIP id
> > assignment needs a sponsor that assigns the PIP id and creates the wiki
> > page.
> >
> > I propose to add a link to the PIP issue in the wiki page until the PIP
> is
> > accepted,
> > because we recently decided to run PIP discussions as GH issues
> >
> > Does anyone volunteer to clean up the PIP list page ?
> >
> > Regards
> > Enrico
>
Re: Problems with PIP ID Management
Posted by Eron Wright <ew...@streamnative.io.INVALID>.
It is funny that we have four unrelated PIP-95 proposals. By the above
principle, the true PIP 95 is "Smart Listener Selection with Multiple Bind
Addresses". :)
On Fri, Sep 24, 2021 at 10:47 AM Michael Marshall <mi...@gmail.com>
wrote:
> Hi Enrico,
>
> Thank you for raising this concern. Apparently acquiring a distributed
> lock on a PIP number is non-trivial :)
>
> > What about sending an email to dev in order to ask for a new PIP id ?
>
> In this design, is it up to a committer (someone with write access to
> the Wiki page) to reply to the request with a PIP number? I think this
> could lead to confusion, as no one person would have the job and there
> could be delays in responding.
>
> Instead of receiving a PIP number, I think we could ask users to
> inspect the mailing list archive to determine the next PIP number.
> Then, the mailing list is the source of truth for PIP numbers, and
> there isn't any delay. This protocol can still have collisions if
> multiple people send PIPs at the same time. However, I think these
> collisions could be resolved quickly: the earliest email in the
> archive wins, and others need to pick the next number(s).
>
> Also, this raises a secondary question for me. In our new PIP process,
> are we copying PIPs from GitHub issues into the GitHub wiki?
>
> In order to push this conversation forward, I wrote a PR with my
> proposed solution to the problem:
> https://github.com/apache/pulsar/pull/12176. I am not attached to my
> solution per se, but I do think a solution to this problem should
> include an update to the GitHub template to help new PIP writers
> easily understand the process.
>
> > Does anyone volunteer to clean up the PIP list page ?
> Once we settle on a protocol, we will need to clean up, as now we have
> two PIP 97s.
>
> Thanks,
> Michael
>
> On Thu, Sep 23, 2021 at 9:51 AM Enrico Olivelli <eo...@gmail.com>
> wrote:
> >
> > Hello,
> > Today I had to rename my PIP from PIP-93 to PIP-95 because PIP-93 was
> > picked up.
> > Then I realised that PIP95 already exists on GitHub, then I moved to
> PIP-97.
> >
> > There are multiple PIP-95 on the Wiki page.
> >
> > https://github.com/apache/pulsar/wiki
> >
> > We need a way to assign new PIP ids
> >
> > What about sending an email to dev in order to ask for a new PIP id ?
> >
> > Only committers have write privileges to the Wiki pages, so any PIP id
> > assignment needs a sponsor that assigns the PIP id and creates the wiki
> > page.
> >
> > I propose to add a link to the PIP issue in the wiki page until the PIP
> is
> > accepted,
> > because we recently decided to run PIP discussions as GH issues
> >
> > Does anyone volunteer to clean up the PIP list page ?
> >
> > Regards
> > Enrico
>
Re: Problems with PIP ID Management
Posted by Michael Marshall <mi...@gmail.com>.
Hi Enrico,
Thank you for raising this concern. Apparently acquiring a distributed
lock on a PIP number is non-trivial :)
> What about sending an email to dev in order to ask for a new PIP id ?
In this design, is it up to a committer (someone with write access to
the Wiki page) to reply to the request with a PIP number? I think this
could lead to confusion, as no one person would have the job and there
could be delays in responding.
Instead of receiving a PIP number, I think we could ask users to
inspect the mailing list archive to determine the next PIP number.
Then, the mailing list is the source of truth for PIP numbers, and
there isn't any delay. This protocol can still have collisions if
multiple people send PIPs at the same time. However, I think these
collisions could be resolved quickly: the earliest email in the
archive wins, and others need to pick the next number(s).
Also, this raises a secondary question for me. In our new PIP process,
are we copying PIPs from GitHub issues into the GitHub wiki?
In order to push this conversation forward, I wrote a PR with my
proposed solution to the problem:
https://github.com/apache/pulsar/pull/12176. I am not attached to my
solution per se, but I do think a solution to this problem should
include an update to the GitHub template to help new PIP writers
easily understand the process.
> Does anyone volunteer to clean up the PIP list page ?
Once we settle on a protocol, we will need to clean up, as now we have
two PIP 97s.
Thanks,
Michael
On Thu, Sep 23, 2021 at 9:51 AM Enrico Olivelli <eo...@gmail.com> wrote:
>
> Hello,
> Today I had to rename my PIP from PIP-93 to PIP-95 because PIP-93 was
> picked up.
> Then I realised that PIP95 already exists on GitHub, then I moved to PIP-97.
>
> There are multiple PIP-95 on the Wiki page.
>
> https://github.com/apache/pulsar/wiki
>
> We need a way to assign new PIP ids
>
> What about sending an email to dev in order to ask for a new PIP id ?
>
> Only committers have write privileges to the Wiki pages, so any PIP id
> assignment needs a sponsor that assigns the PIP id and creates the wiki
> page.
>
> I propose to add a link to the PIP issue in the wiki page until the PIP is
> accepted,
> because we recently decided to run PIP discussions as GH issues
>
> Does anyone volunteer to clean up the PIP list page ?
>
> Regards
> Enrico