~bzg/org-build-failures

10 4

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<875yffelh9.fsf@localhost>
DKIM signature
pass
Download raw message
Christian Köstlin <christian.koestlin@gmail.com> writes:

> Yes .. it's debian unstable ... Sometimes they break stuff ... Happened
> already a few weeks ago and went away after a few days.

I am seeing these failures for a few days in a row now.
I do not think that it is adding much value to our tests.

> We could try to use Ubuntu to test emacs 28 instead.

Agree. I guess Bastien needs to update the server scripts.


-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<87bkp78wt5.fsf@gnu.org>
In-Reply-To
<875yffelh9.fsf@localhost> (view parent)
DKIM signature
pass
Download raw message
Ihor Radchenko <yantar92@posteo.net> writes:

>> We could try to use Ubuntu to test emacs 28 instead.
>
> Agree. I guess Bastien needs to update the server scripts.

Done right now, thanks for the heads up.  I also relaunched the tests.

-- 
 Bastien

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<87h6yxby33.fsf@localhost>
In-Reply-To
<87bkp78wt5.fsf@gnu.org> (view parent)
DKIM signature
pass
Download raw message
Bastien Guerry <bzg@gnu.org> writes:

> Ihor Radchenko <yantar92@posteo.net> writes:
>
>>> We could try to use Ubuntu to test emacs 28 instead.
>>
>> Agree. I guess Bastien needs to update the server scripts.
>
> Done right now, thanks for the heads up.  I also relaunched the tests.

I am still seeing failing debian builds. Am I missing something?
https://builds.sr.ht/~bzg/job/885453

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<87mt8nyzf4.fsf@gnu.org>
In-Reply-To
<87h6yxby33.fsf@localhost> (view parent)
DKIM signature
pass
Download raw message
Ihor Radchenko <yantar92@posteo.net> writes:

> I am still seeing failing debian builds. Am I missing something?
> https://builds.sr.ht/~bzg/job/885453

Or maybe I miss something here -- org-mode-tests is up to date on 
the server, not sure what I can/should really do here.

-- 
 Bastien

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<87v8nbkxfi.fsf@localhost>
In-Reply-To
<87mt8nyzf4.fsf@gnu.org> (view parent)
DKIM signature
pass
Download raw message
Bastien <bzg@gnu.org> writes:

> Ihor Radchenko <yantar92@posteo.net> writes:
>
>> I am still seeing failing debian builds. Am I missing something?
>> https://builds.sr.ht/~bzg/job/885453
>
> Or maybe I miss something here -- org-mode-tests is up to date on 
> the server, not sure what I can/should really do here.

There are no more failures recently. Maybe it simply took some time for
changes to take effect.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<87h6yvxki2.fsf@gnu.org>
In-Reply-To
<87v8nbkxfi.fsf@localhost> (view parent)
DKIM signature
pass
Download raw message
Ihor Radchenko <yantar92@posteo.net> writes:

> There are no more failures recently. Maybe it simply took some time for
> changes to take effect.

Great!

-- 
 Bastien

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<CAG741+bTNUSctgL74Q-FPiVRHKKLGA5QDoF1wfKqwpytnnZnDw@mail.gmail.com>
In-Reply-To
<87h6yvxki2.fsf@gnu.org> (view parent)
DKIM signature
pass
Download raw message
From what I understand the problem is the following:
debian-unstable seems to break the emacs installation from time to time.
(For now it happened two times since we work with debian-unstable and
was "automatically" resolved after a few days).
As the buildserver always builds the image from scratch it picks up
those problems ...

We have two options:
1. keep it like it is -> we see when debian/unstable is broken which
is also worth something as I think debian/unstable is used by many
people. If we could find out how to file a bug report for such a case
it would be good.
2. we switch for the emacs-28 tests to a ubuntu based setup (yaml
files for build.sr.ht as well as local docker files are available).

Anyways we have to reconsider our strategy/options when emacs 29 comes
out, as we would like to run 27, 28 and 29 then.

Kind regards,
Christian

On Sat, Nov 19, 2022 at 12:16 PM Bastien <bzg@gnu.org> wrote:
>
> Ihor Radchenko <yantar92@posteo.net> writes:
>
> > There are no more failures recently. Maybe it simply took some time for
> > changes to take effect.
>
> Great!
>
> --
>  Bastien

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<875yfakwa6.fsf@localhost>
In-Reply-To
<CAG741+bTNUSctgL74Q-FPiVRHKKLGA5QDoF1wfKqwpytnnZnDw@mail.gmail.com> (view parent)
DKIM signature
pass
Download raw message
Christian Köstlin <christian.koestlin@gmail.com> writes:

> From what I understand the problem is the following:
> debian-unstable seems to break the emacs installation from time to time.
> (For now it happened two times since we work with debian-unstable and
> was "automatically" resolved after a few days).
> As the buildserver always builds the image from scratch it picks up
> those problems ...

I expect Org test failures to be less frequent.
Debian breaking multiple times a month sounds like unnecessary
maintenance burden.
Our aim is not testing Debian, but testing Org mode.

> We have two options:
> 1. keep it like it is -> we see when debian/unstable is broken which
> is also worth something as I think debian/unstable is used by many
> people. If we could find out how to file a bug report for such a case
> it would be good.
> 2. we switch for the emacs-28 tests to a ubuntu based setup (yaml
> files for build.sr.ht as well as local docker files are available).

I think we now went with 2.

> Anyways we have to reconsider our strategy/options when emacs 29 comes
> out, as we would like to run 27, 28 and 29 then.

Yes. And it is going to happen soon-ish.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<875yfap3qn.fsf@bzg.fr>
In-Reply-To
<CAG741+bTNUSctgL74Q-FPiVRHKKLGA5QDoF1wfKqwpytnnZnDw@mail.gmail.com> (view parent)
DKIM signature
pass
Download raw message
Christian Köstlin <christian.koestlin@gmail.com> writes:

>>From what I understand the problem is the following:
> debian-unstable seems to break the emacs installation from time to time.
> (For now it happened two times since we work with debian-unstable and
> was "automatically" resolved after a few days).
> As the buildserver always builds the image from scratch it picks up
> those problems ...

Thanks for the explanation.  Does debian-unstable breaks just the
Emacs installation or the Emacs+required-packages-for-babel install?

If the former, then we need to find a way to rely on a stable install
of Emacs latest stable version, because that's quite important for us.

If the latter and tests break because some package breaks, then we
need to revise our ambitions to run all Babel tests here too.

> We have two options:
> 1. keep it like it is -> we see when debian/unstable is broken which
> is also worth something as I think debian/unstable is used by many
> people. If we could find out how to file a bug report for such a case
> it would be good.

I would prefer this option, but the highest priority is that every
failing test run means "there is something wrong in Org's code".

> Anyways we have to reconsider our strategy/options when emacs 29 comes
> out, as we would like to run 27, 28 and 29 then.

We can also rely on various ecosystems (Guix, ArchLinux, Debian,
Ubuntu, etc.) for various needs when using SourceHut CI.  

I.e. if ArchLinux provides a way to test against Emacs 29 more safely
than Debian, and Debian being best for Emacs 28, then I don't think
it's a real problem.

Again, consistency here is about the belief that "failed test =
something wrong in Org's code", not about the method for building the
tests.

WDYT?

-- 
 Bastien

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<CAG741+YoMs6ypqeGoHO_xYPu54Ku=ED8Q-+yTLSEQoJNnPyxyQ@mail.gmail.com>
In-Reply-To
<875yfap3qn.fsf@bzg.fr> (view parent)
DKIM signature
pass
Download raw message
On Sun, Nov 20, 2022 at 6:57 AM Bastien <bzg@gnu.org> wrote:
>
> Christian Köstlin <christian.koestlin@gmail.com> writes:
>
> >>From what I understand the problem is the following:
> > debian-unstable seems to break the emacs installation from time to time.
> > (For now it happened two times since we work with debian-unstable and
> > was "automatically" resolved after a few days).
> > As the buildserver always builds the image from scratch it picks up
> > those problems ...
>
> Thanks for the explanation.  Does debian-unstable breaks just the
> Emacs installation or the Emacs+required-packages-for-babel install?
As far as i understand its the basic emacs installation (emacs-28 for
debian/unstable
at the time of writing).

> If the former, then we need to find a way to rely on a stable install
> of Emacs latest stable version, because that's quite important for us.
>
> If the latter and tests break because some package breaks, then we
> need to revise our ambitions to run all Babel tests here too.
>
> > We have two options:
> > 1. keep it like it is -> we see when debian/unstable is broken which
> > is also worth something as I think debian/unstable is used by many
> > people. If we could find out how to file a bug report for such a case
> > it would be good.
>
> I would prefer this option, but the highest priority is that every
> failing test run means "there is something wrong in Org's code".
+1

>
> > Anyways we have to reconsider our strategy/options when emacs 29 comes
> > out, as we would like to run 27, 28 and 29 then.
>
> We can also rely on various ecosystems (Guix, ArchLinux, Debian,
> Ubuntu, etc.) for various needs when using SourceHut CI.
exactly. At the moment I provide manifest files for debian and ubuntu.

> I.e. if ArchLinux provides a way to test against Emacs 29 more safely
> than Debian, and Debian being best for Emacs 28, then I don't think
> it's a real problem.
>
> Again, consistency here is about the belief that "failed test =
> something wrong in Org's code", not about the method for building the
> tests.
+1

Kind regards,
Christian

Re: [org-mode-test-emacs-28] build failed

Details
Message ID
<871qpvsdxj.fsf@gnu.org>
In-Reply-To
<CAG741+YoMs6ypqeGoHO_xYPu54Ku=ED8Q-+yTLSEQoJNnPyxyQ@mail.gmail.com> (view parent)
DKIM signature
pass
Download raw message
Hi Christian,

Christian Köstlin <christian.koestlin@gmail.com> writes:

>> Does debian-unstable breaks just the
>> Emacs installation or the Emacs+required-packages-for-babel install?

> As far as i understand its the basic emacs installation (emacs-28 for
> debian/unstable
> at the time of writing).

Noted, thanks.

>> We can also rely on various ecosystems (Guix, ArchLinux, Debian,
>> Ubuntu, etc.) for various needs when using SourceHut CI.

> At the moment I provide manifest files for debian and ubuntu.

My suggestion was not precise enough: for now orgmode.org uses only
the three debian-based .yml manifests.  I was suggesting that, if it
is needed, we can use Debian for testing Emacs-26 and Emacs-27, and 
another distribution for testing Emacs-28.

I suggest we see how testing behaves after Org 9.6 when we commit
more stuff - and decide then.

All best,

-- 
 Bastien
Reply to thread Export thread (mbox)