[gentoo-dev] Massive Github PR Queue

[gentoo-dev] Massive Github PR Queue

[orbea]

Hi,

Currently the Gentoo Github PR queue is at 577 open PRs that even
includes one that has been left open in 2018 and neglected since 2021.

While not trying to be rude before contributing to Gentoo I was involved
with Slackbuilds.org for Slackware where everything gets reviewed once
a week with only a handful of maintainers doing the reviewing process.

Why does Gentoo lets PRs indefinitely sit around and collect dust? Its
extremely discouraging for contributors if their work just gets ignored.
With some extra work I imagine its possible to get the PR queue to a
much more manageable size where work doesn't get lost.

Re: [gentoo-dev] Massive Github PR Queue

[Michael Orlitzky]

On Fri, 2023-08-11 at 07:07 -0700, orbea wrote:
> 
> Why does Gentoo lets PRs indefinitely sit around and collect dust? Its
> extremely discouraging for contributors if their work just gets ignored.
> With some extra work I imagine its possible to get the PR queue to a
> much more manageable size where work doesn't get lost.
> 

Reviewing a PR takes longer than doing it yourself would, and if you
had time to do it yourself there wouldn't be a PR in the first place.

But also, Github sucks. Open bugs.

Re: [gentoo-dev] Massive Github PR Queue

[orbea]

On Fri, 11 Aug 2023 10:19:29 -0400
Michael Orlitzky <mjo@gentoo.org> wrote:

> On Fri, 2023-08-11 at 07:07 -0700, orbea wrote:
> > 
> > Why does Gentoo lets PRs indefinitely sit around and collect dust?
> > Its extremely discouraging for contributors if their work just gets
> > ignored. With some extra work I imagine its possible to get the PR
> > queue to a much more manageable size where work doesn't get lost.
> >   
> 
> Reviewing a PR takes longer than doing it yourself would, and if you
> had time to do it yourself there wouldn't be a PR in the first place.

Understandable, but I still don't get how 1 or 2 people can take care
of SBo every week while Gentoo the process stagnates.

> 
> But also, Github sucks. Open bugs.
> 
> 

No disagreements and personally I wouldn't use Github if not for it
being beyond my control, but there are still 235 open PRs with bugs
assigned.

https://github.com/gentoo/gentoo/pulls?q=is%3Apr+is%3Aopen+label%3A%22bug+linked%22

Re: [gentoo-dev] Massive Github PR Queue

[Michael Orlitzky]

On Fri, 2023-08-11 at 07:46 -0700, orbea wrote:
> 
> Understandable, but I still don't get how 1 or 2 people can take care
> of SBo every week while Gentoo the process stagnates.

Probably due to compartmentalization. In Gentoo you're often waiting
for one maintainer. Or worse, waiting for one "project" consisting of
several people none of whom actually care about the package the PR
touches.


> > 
> > But also, Github sucks. Open bugs.
> > 
> 
> No disagreements and personally I wouldn't use Github if not for it
> being beyond my control, but there are still 235 open PRs with bugs
> assigned.
> 
> https://github.com/gentoo/gentoo/pulls?q=is%3Apr+is%3Aopen+label%3A%22bug+linked%22
> 

Linking a bug was a half-hearted attempt to appease the folks who think
that Gentoo development shouldn't depend on proprietary software (per
our social contract). But if the bug just points me to Github, it
doesn't really change anything.

I of course have a comprehensive list of Github complaints in mind 24
hours a day, but they're not especially relevant. What is relevant is
that some other developers have their own list, and sometimes you're
waiting on one developer who doesn't like to use Github for whatever
reason. My personal PR queue is of length zero, so I'm not making
excuses, but when you want something done for free it's in your best
interest to make it easy for the person who has to do it.

Re: [gentoo-dev] Massive Github PR Queue

[Joonas Niilola]

[-- Attachment #1.1: Type: text/plain, Size: 3204 bytes --]

On 11.8.2023 17.07, orbea wrote:
> Hi,
> 
> Currently the Gentoo Github PR queue is at 577 open PRs that even
> includes one that has been left open in 2018 and neglected since 2021.

That's a bit misleading. The PR in question is opened by a Gentoo
developer, and labeled as "do-not-merge". So maybe they lost interest,
or forgot? Ping in the PR if you want to see it finished. But there are
indeed tons of PRs open from 2020.


> 
> While not trying to be rude before contributing to Gentoo I was involved
> with Slackbuilds.org for Slackware where everything gets reviewed once
> a week with only a handful of maintainers doing the reviewing process.

We also have 32,000 PRs closed, while slackbuild is at 3000. And here
also only handful of members are putting effort in general PR review. So
I'd say we're doing pretty good in that regard.


> 
> Why does Gentoo lets PRs indefinitely sit around and collect dust? Its
> extremely discouraging for contributors if their work just gets ignored.
> With some extra work I imagine its possible to get the PR queue to a
> much more manageable size where work doesn't get lost.
> 

It is discouraging I agree. Even for us. If you want to help, go through
the old PRs seeing which are relevant, ping the maintainers if the PR is
waiting for action on their side, and let us know (via IRC for example)
which PRs are mergeable.

Now as to the problems. We've done some cleanups in the past, but in my
opinion it's no point in putting too much energy on that if there's are
many fresh PRs to look at. We shouldn't lose the momentum. There's also
this handy tool,
https://github.com/wimmuskee/gengee
which I used a lot in the past but haven't been able to recently due to
always looking at few week old PRs.

Another big issue is as mjo pointed out, not everyone uses GH. But don't
be fooled, I don't think you're gaining better record attaching your
work to bugzilla either. It really depends on the maintainer whether
they prefer contributions via GH or BZ, and we've tossed some ideas in
the past trying to show maintainer's preference, but nothing came of
those ideas. Then unfortunately some maintainers who'll ignore both exists.

There's pretty much only 1 project dealing with Github PRs (proxy-maint)
and that project only gets notified with packages under
maintainer-needed, or maintained via proxy-maint itself. This project
doesn't get notified for _every_ PR opened, which I believe is a common
misconception. And it shouldn't either. There's simply too much incoming
mail even with few projects. So it's up to projects handling their own
PRs, but again, not everyone uses GH / care about PRs.

What I suggest for everyone to do:
1: open a bug that gets assigned to maintainer,
2: link your PR to the bug with "Bug:" or "Closes:" tag,
3: if no one reviews it for 2-4 weeks, pop into #gentoo-proxy-maint IRC
channel and ask someone to take a look.

Open bug is a requirement for other non-project members to merge your
work, since it allows maintainer timeout. Note that packages maintained
by base-system and toolchain can't be merged by devs outside these projects.

-- juippis

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 618 bytes --]

Re: [gentoo-dev] Massive Github PR Queue

[Sam James]

Joonas Niilola <juippis@gentoo.org> writes:

> [[PGP Signed Part:Undecided]]
> On 11.8.2023 17.07, orbea wrote:
>> Hi,
>> 
>> Currently the Gentoo Github PR queue is at 577 open PRs that even
>> includes one that has been left open in 2018 and neglected since 2021.
>
> That's a bit misleading. The PR in question is opened by a Gentoo
> developer, and labeled as "do-not-merge". So maybe they lost interest,
> or forgot? Ping in the PR if you want to see it finished. But there are
> indeed tons of PRs open from 2020.
>
>
>> 
>> While not trying to be rude before contributing to Gentoo I was involved
>> with Slackbuilds.org for Slackware where everything gets reviewed once
>> a week with only a handful of maintainers doing the reviewing process.
>
> We also have 32,000 PRs closed, while slackbuild is at 3000. And here
> also only handful of members are putting effort in general PR review. So
> I'd say we're doing pretty good in that regard.
>

juippis' whole email nails the issue, but I'd just like to add that
there's kind of a baseline at ~400/~450 or so where everything below
that is PRs for new packages or long-obsolete stuff nobody closed yet,
or where we're waiting on the submitter. It's just that going ahead
and going over those is time-consuming and means we're spending less
time on active PRs which need looking at.

as for bz: it's far harder to review something on bz and if everyone
did that for their contributions, any backlog (which you can't
easily measure on bz either) would be far larger.

> [...]

Re: [gentoo-dev] Massive Github PR Queue

[orbea]

On Fri, 11 Aug 2023 19:10:36 +0300
Joonas Niilola <juippis@gentoo.org> wrote:

> On 11.8.2023 17.07, orbea wrote:
> > Hi,
> > 
> > Currently the Gentoo Github PR queue is at 577 open PRs that even
> > includes one that has been left open in 2018 and neglected since
> > 2021.  
> 
> That's a bit misleading. The PR in question is opened by a Gentoo
> developer, and labeled as "do-not-merge". So maybe they lost interest,
> or forgot? Ping in the PR if you want to see it finished. But there
> are indeed tons of PRs open from 2020.

Good point, I should of looked closer and found a more appropriate
example.

> 
> 
> > 
> > While not trying to be rude before contributing to Gentoo I was
> > involved with Slackbuilds.org for Slackware where everything gets
> > reviewed once a week with only a handful of maintainers doing the
> > reviewing process.  
> 
> We also have 32,000 PRs closed, while slackbuild is at 3000. And here
> also only handful of members are putting effort in general PR review.
> So I'd say we're doing pretty good in that regard.

Using github and gitlab are recent additions for SBo, their main repo
is maintained with cgit.

https://git.slackbuilds.org/slackbuilds/

More regular and trusted contributors push to their own branches which
get merged roughly once a week while other people use the submission
form.

https://slackbuilds.org/guidelines/

Where the pending and approved submissions are listed here.

https://slackbuilds.org/pending/
https://slackbuilds.org/ready/

Again these get merged about once a week where the SBo maintainers fix
any minor issues themselves and contact the submitter usually via irc
or e-mail for anything more involved.

> 
> 
> > 
> > Why does Gentoo lets PRs indefinitely sit around and collect dust?
> > Its extremely discouraging for contributors if their work just gets
> > ignored. With some extra work I imagine its possible to get the PR
> > queue to a much more manageable size where work doesn't get lost.
> >   
> 
> It is discouraging I agree. Even for us. If you want to help, go
> through the old PRs seeing which are relevant, ping the maintainers
> if the PR is waiting for action on their side, and let us know (via
> IRC for example) which PRs are mergeable.
> 
> Now as to the problems. We've done some cleanups in the past, but in
> my opinion it's no point in putting too much energy on that if
> there's are many fresh PRs to look at. We shouldn't lose the
> momentum. There's also this handy tool,
> https://github.com/wimmuskee/gengee
> which I used a lot in the past but haven't been able to recently due
> to always looking at few week old PRs.
> 
> Another big issue is as mjo pointed out, not everyone uses GH. But
> don't be fooled, I don't think you're gaining better record attaching
> your work to bugzilla either. It really depends on the maintainer
> whether they prefer contributions via GH or BZ, and we've tossed some
> ideas in the past trying to show maintainer's preference, but nothing
> came of those ideas. Then unfortunately some maintainers who'll
> ignore both exists.
> 
> There's pretty much only 1 project dealing with Github PRs
> (proxy-maint) and that project only gets notified with packages under
> maintainer-needed, or maintained via proxy-maint itself. This project
> doesn't get notified for _every_ PR opened, which I believe is a
> common misconception. And it shouldn't either. There's simply too
> much incoming mail even with few projects. So it's up to projects
> handling their own PRs, but again, not everyone uses GH / care about
> PRs.
> 
> What I suggest for everyone to do:
> 1: open a bug that gets assigned to maintainer,
> 2: link your PR to the bug with "Bug:" or "Closes:" tag,
> 3: if no one reviews it for 2-4 weeks, pop into #gentoo-proxy-maint
> IRC channel and ask someone to take a look.
> 
> Open bug is a requirement for other non-project members to merge your
> work, since it allows maintainer timeout. Note that packages
> maintained by base-system and toolchain can't be merged by devs
> outside these projects.
> 
> -- juippis

Perhaps the correct answer would be neither Bugzilla or Github?
Bugzilla may be a good way to track bugs, but attaching work to it can
be cumbersome. Meanwhile Github has the benefit of using git which can
be more convenient, but also is proprietary and owned by Microsoft.

However thank you for the in depth explanation.

Re: [gentoo-dev] Massive Github PR Queue

[Rich Freeman]

On Fri, Aug 11, 2023 at 12:54 PM Sam James <sam@gentoo.org> wrote:
> juippis' whole email nails the issue, but I'd just like to add that
> there's kind of a baseline at ~400/~450 or so where everything below
> that is PRs for new packages or long-obsolete stuff nobody closed yet,
> or where we're waiting on the submitter. It's just that going ahead
> and going over those is time-consuming and means we're spending less
> time on active PRs which need looking at.

++

In general the Gentoo workflow tends to be a bit different from a lot
of other projects as well, with quite a bit more individual
ownership/interest.  Many distros just stick to core packages in a
main repo, and consign everything else to user-maintained
repos/overlays/etc, or tiers with lower expectations for quality.

Then you have all the build-related issues that can be hard to
reproduce/test, that other distros simply don't have to worry about
because they only support one or two build configurations.

Another thing I've seen in fairly large projects is auto-closing of
issues/bugs/PRs/etc.  If your submission sits around for more than a
few weeks without getting merged, a bot comes along and helpfully
marks it closed.  That of course results in a nice low open-issue
count, but of course that doesn't mean those issues are actually
fixed.

-- 
Rich

Re: [gentoo-dev] Massive Github PR Queue

[Gordon Pettey]

[-- Attachment #1: Type: text/plain, Size: 814 bytes --]

On Fri, Aug 11, 2023 at 11:11 AM Joonas Niilola <juippis@gentoo.org> wrote:

>
> Another big issue is as mjo pointed out, not everyone uses GH.
>

"Not using GitHub" is not an excuse for a PR to sit ignored when it has a
bug attachment. Anyone can view a PR anonymously and clone the remote from
the read-only https URL. There's no "using github" required for that beyond
clicking the link in the attached bug to see the PR and the remote
repository name. For the 202 "bug linked" (excluding "new package") PRs,
~50% are more than 6 months old. While that doesn't necessarily mean
they've been ignored (for the "not using GitHub" devs, the code could've
been merged that way without any interaction on the GH PR, then it would be
up to the submitter to close the PR), it still doesn't look great.

[-- Attachment #2: Type: text/html, Size: 1180 bytes --]

Re: [gentoo-dev] Massive Github PR Queue

[Michael Orlitzky]

On Fri, 2023-08-11 at 10:35 -0700, orbea wrote:
> 
> Perhaps the correct answer would be neither Bugzilla or Github?

A mailing list (whose archives work :o) with git-send-email threads
would be an improvement over both.

Re: [gentoo-dev] Massive Github PR Queue

[Michael Orlitzky]

On Fri, 2023-08-11 at 18:37 -0500, Gordon Pettey wrote:
> 
> "Not using GitHub" is not an excuse for a PR to sit ignored when it has a
> bug attachment. Anyone can view a PR anonymously and clone the remote from
> the read-only https URL. There's no "using github" required for that beyond
> clicking the link in the attached bug to see the PR and the remote
> repository name.

We went years with one developer refusing to fix tinderbox bugs because
the build logs were hosted on AWS instead of bugzilla.

These are Gentoo Linux developers. People who chose a corner case
operating system and then chose a corner case distribution of it and
then became a corner case of its userbase. As a result, you're going to
see some corner-case moral and technical principles. Telling someone
that those principles aren't an excuse just isn't going to get them to
help you any faster.

Re: [gentoo-dev] Massive Github PR Queue

[Mike Gilbert]

[-- Attachment #1: Type: text/plain, Size: 558 bytes --]

On Fri, Aug 11, 2023 at 21:18 Michael Orlitzky <mjo@gentoo.org> wrote:

> On Fri, 2023-08-11 at 10:35 -0700, orbea wrote:
> >
> > Perhaps the correct answer would be neither Bugzilla or Github?
>
> A mailing list (whose archives work :o) with git-send-email threads
> would be an improvement over both.
>
>
> I don’t think replacing pull requests with mailing list posts would yield
any significant improvement. Personally, I am much less likely to go
through the trouble of fetching patches from my email than fetching a
branch from GitHub.

[-- Attachment #2: Type: text/html, Size: 874 bytes --]

Re: [gentoo-dev] Massive Github PR Queue

[Michał Górny]

On Fri, 2023-08-11 at 21:18 -0400, Michael Orlitzky wrote:
> On Fri, 2023-08-11 at 10:35 -0700, orbea wrote:
> > 
> > Perhaps the correct answer would be neither Bugzilla or Github?
> 
> A mailing list (whose archives work :o) with git-send-email threads
> would be an improvement over both.

I'm sure that having to go through 15 threads where people reposted
the same patches with updates to find all the comments will be a great
productivity booster.

-- 
Best regards,
Michał Górny

Re: [gentoo-dev] Massive Github PR Queue

[Haelwenn (lanodan) Monnier]

[2023-08-11 22:12:27-0400] Mike Gilbert:
>On Fri, Aug 11, 2023 at 21:18 Michael Orlitzky <mjo@gentoo.org> wrote:
>
>> On Fri, 2023-08-11 at 10:35 -0700, orbea wrote:
>> >
>> > Perhaps the correct answer would be neither Bugzilla or Github?
>>
>> A mailing list (whose archives work :o) with git-send-email threads
>> would be an improvement over both.
>>
>>
>> I don’t think replacing pull requests with mailing list posts would yield
>any significant improvement. Personally, I am much less likely to go
>through the trouble of fetching patches from my email than fetching a
>branch from GitHub.

Maybe there could be something that automatically pushes the mailing-list patches to GitHub? (and vice-versa?)

There was such a setup for a bit on Alpine Linux[1] using https://lists.sr.ht/ as software, Drew DeVault being the one that put it together.

1: https://lists.alpinelinux.org/~alpine/aports

Re: [gentoo-dev] Massive Github PR Queue

[orbea]

On Fri, 11 Aug 2023 18:37:10 -0500
Gordon Pettey <petteyg359@gmail.com> wrote:

> On Fri, Aug 11, 2023 at 11:11 AM Joonas Niilola <juippis@gentoo.org>
> wrote:
> 
> >
> > Another big issue is as mjo pointed out, not everyone uses GH.
> >  
> 
> "Not using GitHub" is not an excuse for a PR to sit ignored when it
> has a bug attachment. Anyone can view a PR anonymously and clone the
> remote from the read-only https URL. There's no "using github"
> required for that beyond clicking the link in the attached bug to see
> the PR and the remote repository name. For the 202 "bug linked"
> (excluding "new package") PRs, ~50% are more than 6 months old. While
> that doesn't necessarily mean they've been ignored (for the "not
> using GitHub" devs, the code could've been merged that way without
> any interaction on the GH PR, then it would be up to the submitter to
> close the PR), it still doesn't look great.

You can just add a "Closes:" tag with the github PR url to the commit
message for it to be closed automatically. Using app-portage/pram can
also streamline the process of merging the commit. And if there are any
issues that need to be communicated there is e-mail or even irc in some
cases.