[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Maxim Cournoyer]

Hi,

This was fully built for x86_64 as some insurance that it didn't
introduce any problem:
https://ci.guix.gnu.org/jobset/meson-parallel-tests

It's ready to me merged.

-- 
Thanks,
Maxim

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Christopher Baines]

[-- Attachment #1: Type: text/plain, Size: 388 bytes --]

Maxim Cournoyer <maxim.cournoyer@gmail.com> writes:

> Hi,
>
> This was fully built for x86_64 as some insurance that it didn't
> introduce any problem:
> https://ci.guix.gnu.org/jobset/meson-parallel-tests
>
> It's ready to me merged.

The branch name in the issue title doesn't quite match the actual name
of the branch. Could you retitle the issue to make it match up?

Thanks,

Chris

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 987 bytes --]

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Maxim Cournoyer]

Hi Chris,

Christopher Baines <mail@cbaines.net> writes:

> Maxim Cournoyer <maxim.cournoyer@gmail.com> writes:
>
>> Hi,
>>
>> This was fully built for x86_64 as some insurance that it didn't
>> introduce any problem:
>> https://ci.guix.gnu.org/jobset/meson-parallel-tests
>>
>> It's ready to me merged.
>
> The branch name in the issue title doesn't quite match the actual name
> of the branch. Could you retitle the issue to make it match up?

Thanks for the heads-up.  I believe that's now fixed.

PS: Your email went to my spam folder, for some reasons.

-- 
Thanks,
Maxim

[bug#73765] Rebased

[Andreas Enge]

Hello,

this is the next branch in the QA queue. I have rebased it on master;
hopefully it will be picked up by QA. Or you could have it built again
on CI, as QA has trouble keeping up recently.

Andreas

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Ludovic Courtès]

Hi,

Andreas Enge <andreas@enge.fr> skribis:

> this is the next branch in the QA queue. I have rebased it on master;
> hopefully it will be picked up by QA. Or you could have it built again
> on CI, as QA has trouble keeping up recently.

It currently fails to evaluate (commit 0363f28):

--8<---------------cut here---------------start------------->8---
   376:14  3 (arguments #<package mesa@24.2.5 gnu/packages/gl.scm:30?>)
In ice-9/boot-9.scm:
  1685:16  2 (raise-exception _ #:continuable? _)
  1780:13  1 (_ #<&compound-exception components: (#<&undefined-vari?>)
In unknown file:
           0 (backtrace #<undefined>)

(exception unbound-variable (value #f) (value "Unbound variable: ~S") (value (meson-1.5)) (value #f))
--8<---------------cut here---------------end--------------->8---

(From <https://ci.guix.gnu.org/eval/1837912/log/raw>.)

Ludo’.

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Andreas Enge]

Am Fri, Nov 22, 2024 at 03:45:26PM +0100 schrieb Ludovic Courtès:
> It currently fails to evaluate (commit 0363f28):
> --8<---------------cut here---------------start------------->8---
>    376:14  3 (arguments #<package mesa@24.2.5 gnu/packages/gl.scm:30?>)
> In ice-9/boot-9.scm:
>   1685:16  2 (raise-exception _ #:continuable? _)
>   1780:13  1 (_ #<&compound-exception components: (#<&undefined-vari?>)
> In unknown file:
>            0 (backtrace #<undefined>)
> 
> (exception unbound-variable (value #f) (value "Unbound variable: ~S") (value (meson-1.5)) (value #f))
> --8<---------------cut here---------------end--------------->8---

That is strange - I did a "make" in my branch before pushing.
How can this be explained?

Andreas

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Andreas Enge]

Am Fri, Nov 22, 2024 at 04:03:11PM +0100 schrieb Andreas Enge:
> That is strange - I did a "make" in my branch before pushing.
> How can this be explained?

"guix build mesa" fails due to a missing (and now superfluous) variable.

I have pushed a fix, which I could not test due to lack of space.
If it turns out not to work on CI, and since we rebase and thus force
push branches anyway, please feel free to remove the commit again.

Andreas

[bug#73765] Results

[Andreas Enge]

The branch has been built on CI, with 36 new failures:
   https://ci.guix.gnu.org/eval/1838237?status=newly-failed

Maybe something could be done about them before merging?

Andreas

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Ludovic Courtès]

Hi!

Andreas Enge <andreas@enge.fr> skribis:

> The branch has been built on CI, with 36 new failures:
>    https://ci.guix.gnu.org/eval/1838237?status=newly-failed
>
> Maybe something could be done about them before merging?

BTW, I noticed that the jobset was configured to build for x86_64-linux
only (with 95% success currently¹).  I’ve added i686-linux now, though
it’ll only take effect on the next evaluation.

Feel free to add aarch64-linux as well once you know this is likely the
last evaluation before merge.

Ludo’.

¹ https://ci.guix.gnu.org/jobset/meson-parallel-tests

bug#73765: Request for merging "meson-enable-parallel-tests" branch

[Andreas Enge]

Hello!

I have gone back to the original merge request and noticed, following
Maxim's remark, that indeed no build failures occurred. Updating the meson
version resulted in build failures, but which looked more like random test
failures in somewhat brittle packages that fail on and off.

So I have taken the liberty and rebased on master and pushed.

I had forgotten that the branch was not yet built on i686, sorry.

Apart from that, I think it is good to clear the space for the following
branches waiting in line.

Andreas

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Maxim Cournoyer]

Hi,

Ludovic Courtès <ludo@gnu.org> writes:

> Hi!
>
> Andreas Enge <andreas@enge.fr> skribis:
>
>> The branch has been built on CI, with 36 new failures:
>>    https://ci.guix.gnu.org/eval/1838237?status=newly-failed
>>
>> Maybe something could be done about them before merging?

These appear to be flaky tests... should be reported upstream and
disabled in the packages... I tried updating orange, and ended up in an
infernal rust-laced rabbit hole, so I'm burnt out for the rest for now
;-).

-- 
Thanks,
Maxim

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[John Kehayias via Guix-patches via]

Hello,

On Wed, Nov 27, 2024 at 11:52 AM, GNU bug Tracking System wrote:

> Your bug report
>
> #73765: Request for merging "meson-enable-parallel-tests" branch
>
> which was filed against the guix-patches package, has been closed.
>
> The explanation is attached below, along with your original report.
> If you require more details, please reply to 73765@debbugs.gnu.org.
>
> --
> 73765: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=73765
> GNU Bug Tracking System
> Contact help-debbugs@gnu.org with problems
>
> From: Andreas Enge <andreas@enge.fr>
> Subject: Re: bug#73765: Request for merging "meson-enable-parallel-tests"
>  branch
> To: Ludovic Courtès <ludo@gnu.org>
> Cc: 73765-done@debbugs.gnu.org, Maxim Cournoyer <maxim.cournoyer@gmail.com>
> Date: Wed, 27 Nov 2024 12:51:30 +0100 (17 hours, 46 minutes, 36 seconds ago)
>
> Hello!
>
> I have gone back to the original merge request and noticed, following
> Maxim's remark, that indeed no build failures occurred. Updating the meson
> version resulted in build failures, but which looked more like random test
> failures in somewhat brittle packages that fail on and off.
>
> So I have taken the liberty and rebased on master and pushed.
>
> I had forgotten that the branch was not yet built on i686, sorry.
>
> Apart from that, I think it is good to clear the space for the following
> branches waiting in line.
>
> Andreas
>

I just noticed the big build queue on Cuirass, is that all for the
non-x86_64 architectures? (The change to mesa, if it was on the branch
before, is ~4k packages alone). Should we have waited for that to
get up to reasonable coverage before pushing to master?

PS: Andreas I saw some messages earlier in this thread that again didn't
go to others. Annoyingly, the debbugs bug number address doesn't go to
anyone (that I know of) and I've frequently missed or miss-sent messages
because of that. Just thought I'd let you know in case you missed that.

Thanks for everyone's work on this!
John

>
> ----------
>
> From: Maxim Cournoyer <maxim.cournoyer@gmail.com>
> Subject: Request for merging "meson-enable-parallel-tests" branch
> To: guix-patches <guix-patches@gnu.org>
> Date: Sat, 12 Oct 2024 14:36:12 +0900
> Date: Sat, 12 Oct 2024 14:36:12 +0900 (6 weeks, 5 days, 1 minute ago)
>
> Hi,
>
> This was fully built for x86_64 as some insurance that it didn't
> introduce any problem:
> https://ci.guix.gnu.org/jobset/meson-parallel-tests
>
> It's ready to me merged.

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Maxim Cournoyer]

Hi John,

John Kehayias <john.kehayias@protonmail.com> writes:

[...]

> I just noticed the big build queue on Cuirass, is that all for the
> non-x86_64 architectures? (The change to mesa, if it was on the branch
> before, is ~4k packages alone). Should we have waited for that to
> get up to reasonable coverage before pushing to master?

I had set up the build job on CI for x86-64 only in my original
configuration and had forgotten about it -- it's only been enabled for
other archs recently.

> PS: Andreas I saw some messages earlier in this thread that again didn't
> go to others. Annoyingly, the debbugs bug number address doesn't go to
> anyone (that I know of) and I've frequently missed or miss-sent messages
> because of that. Just thought I'd let you know in case you missed
> that.

That's an annoyance of Debbugs yes.  I wish it'd handle CCing all
participants itself when any message reaches a bug they are
participating in.

I think 'mumi compose' does this, although I haven't yet used it.

-- 
Thanks,
Maxim

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[Andreas Enge]

Am Thu, Nov 28, 2024 at 04:46:07PM +0900 schrieb Maxim Cournoyer:
> I had set up the build job on CI for x86-64 only in my original
> configuration and had forgotten about it -- it's only been enabled for
> other archs recently.

Me too - it had been built out for x86_64, and I had forgotten about i686.
For aarch64, I think we do not have the build power at ci to wait for it,
and QA is unfortunately almost not working right now.

> > PS: Andreas I saw some messages earlier in this thread that again didn't
> > go to others. Annoyingly, the debbugs bug number address doesn't go to
> > anyone (that I know of) and I've frequently missed or miss-sent messages
> > because of that. Just thought I'd let you know in case you missed
> > that.
> That's an annoyance of Debbugs yes.  I wish it'd handle CCing all
> participants itself when any message reaches a bug they are
> participating in.

Is this true? I think I usually get messages directed at bugs I have opened
or participated in. Or is it just when people cc me explicitly? So are you
saying that when sending mail to a bug, one should manually add all
participants in cc? This would be a lot of work - since issues.guix.gnu.org
does not show email addresses, one would need to download the raw message,
or look at debbugs.gnu.org and type the addresses by hand.

Andreas

[bug#73765] Request for merging "meson-enable-parallel-tests" branch

[John Kehayias via Guix-patches via]

On Thu, Nov 28, 2024 at 09:51 AM, Andreas Enge wrote:

> Am Thu, Nov 28, 2024 at 04:46:07PM +0900 schrieb Maxim Cournoyer:
>> I had set up the build job on CI for x86-64 only in my original
>> configuration and had forgotten about it -- it's only been enabled for
>> other archs recently.
>
> Me too - it had been built out for x86_64, and I had forgotten about i686.
> For aarch64, I think we do not have the build power at ci to wait for it,
> and QA is unfortunately almost not working right now.
>

I see. Yeah, it takes a long time even when things are not backed up,
to build for non-x86. Usually Bordeaux gets the best coverage there.
Probably we need to be able to manage jobs on Bordeaux more directly
to clear out outdated evaluations when we are working on big branches.

>> > PS: Andreas I saw some messages earlier in this thread that again didn't
>> > go to others. Annoyingly, the debbugs bug number address doesn't go to
>> > anyone (that I know of) and I've frequently missed or miss-sent messages
>> > because of that. Just thought I'd let you know in case you missed
>> > that.
>> That's an annoyance of Debbugs yes.  I wish it'd handle CCing all
>> participants itself when any message reaches a bug they are
>> participating in.
>
> Is this true? I think I usually get messages directed at bugs I have opened
> or participated in. Or is it just when people cc me explicitly? So are you
> saying that when sending mail to a bug, one should manually add all
> participants in cc? This would be a lot of work - since issues.guix.gnu.org
> does not show email addresses, one would need to download the raw message,
> or look at debbugs.gnu.org and type the addresses by hand.
>
> Andreas

Indeed, it is a real issue. I responded with some more detail in our
other thread but I'll reiterate here we should make sure this is
in/very clear in the manual. This seems like a good assumption
(bugnumber is an email list) and it really should be as who knows how
many messages were "lost". Browsing debbugs on Emacs helps me see
these, and after I got bit by this on both ends I try to remember to
check there or the mumi frontend for all messages.

(Perhaps this is a very real practical reason to have alternatives to
debbugs, though mumi should be in a good position without drastic
changes on our end already.)

John