automatic tests keep failling

112 views
Skip to first unread message

John Cremona

unread,
Dec 19, 2023, 8:54:59 AM12/19/23
to SAGE devel
I am finding it impossible to get simple PRs to get past the automatic testing workflow system, which are repeatedly failing for reasons which have noting to do with any code I have changed.

Two current examples are 

 - https://github.com/sagemath/sage/pull/36786 ( a simple bugfix which I submitted 3 weeks ago and which could easily have made it into the last release) 
 - https://github.com/sagemath/sage/pull/36879 (an update to standard dependency package eclib, requiring no code changes at all)

It used to be such a simple matter to locate a bug, fix it and get the fix merged.  Now it seems almost impossible (for me at least).  This is a huge disincentive to people who would like to contribute to Sage.

John

Dima Pasechnik

unread,
Dec 19, 2023, 9:20:21 AM12/19/23
to sage-...@googlegroups.com
On Tue, Dec 19, 2023 at 1:54 PM John Cremona <john.c...@gmail.com> wrote:
>
> I am finding it impossible to get simple PRs to get past the automatic testing workflow system, which are repeatedly failing for reasons which have noting to do with any code I have changed.
>
> Two current examples are
>
> - https://github.com/sagemath/sage/pull/36786 ( a simple bugfix which I submitted 3 weeks ago and which could easily have made it into the last release)
> - https://github.com/sagemath/sage/pull/36879 (an update to standard dependency package eclib, requiring no code changes at all)
>
> It used to be such a simple matter to locate a bug, fix it and get the fix merged. Now it seems almost impossible (for me at least).

Hi John,

You have enough rights to ask people for reviews, right on the PR page.
You should also put up a "needs review" tag, there, on the right side
of the page.
Without "needs review" tag chances are that noone will look at it.

Dima


> This is a huge disincentive to people who would like to contribute to Sage.
>
> John
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAD0p0K6umsq4yN27iHt%3Dpj3nh_8ZFTWSSzTGcBDicMsu0HMX0w%40mail.gmail.com.

John Cremona

unread,
Dec 19, 2023, 9:49:51 AM12/19/23
to sage-...@googlegroups.com
On Tue, 19 Dec 2023 at 14:20, Dima Pasechnik <dim...@gmail.com> wrote:
On Tue, Dec 19, 2023 at 1:54 PM John Cremona <john.c...@gmail.com> wrote:
>
> I am finding it impossible to get simple PRs to get past the automatic testing workflow system, which are repeatedly failing for reasons which have noting to do with any code I have changed.
>
> Two current examples are
>
>  - https://github.com/sagemath/sage/pull/36786 ( a simple bugfix which I submitted 3 weeks ago and which could easily have made it into the last release)
>  - https://github.com/sagemath/sage/pull/36879 (an update to standard dependency package eclib, requiring no code changes at all)
>
> It used to be such a simple matter to locate a bug, fix it and get the fix merged.  Now it seems almost impossible (for me at least).

Hi John,

You have enough rights to ask people for reviews, right on the PR page.
You should also put up a "needs review" tag, there, on the right side
of the page.
Without "needs review" tag chances are that noone will look at it.

I would have assumed that no-one would look at a PR whose automatic tests were failing, since that *should* indicate that something is wrong with the PR.
 
My complaint is that even with a PR based on the current develop branch, tests are failing which are not caused by the PR.

But I will certainly now ask you to review both of these!

John


Dima


> This is a huge disincentive to people who would like to contribute to Sage.
>
> John
>
> --
> You received this message because you are subscribed to the Google Groups "sage-devel" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/CAD0p0K6umsq4yN27iHt%3Dpj3nh_8ZFTWSSzTGcBDicMsu0HMX0w%40mail.gmail.com.

--
You received this message because you are subscribed to the Google Groups "sage-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+...@googlegroups.com.

Dima Pasechnik

unread,
Dec 19, 2023, 10:03:01 AM12/19/23
to sage-...@googlegroups.com
One way to fix this would be CI that only tests a relevant subset of things, and it's not quite there yet.

So one has to look at results of CI to gauge the results. One reason for this is that the workflow of the release manager is not quite aligned with the CI,
so our beta releases might well fail our CI !


>
>But I will certainly now ask you to review both of these!
>
>John
>
>
>> Dima
>>
>>
>> > This is a huge disincentive to people who would like to contribute to
>> Sage.
>> >
>> > John
>> >
>> > --
>> > You received this message because you are subscribed to the Google
>> Groups "sage-devel" group.
>> > To unsubscribe from this group and stop receiving emails from it, send
>> an email to sage-devel+...@googlegroups.com.
>> > To view this discussion on the web visit
>> https://groups.google.com/d/msgid/sage-devel/CAD0p0K6umsq4yN27iHt%3Dpj3nh_8ZFTWSSzTGcBDicMsu0HMX0w%40mail.gmail.com
>> .
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "sage-devel" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to sage-devel+...@googlegroups.com.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/sage-devel/CAAWYfq1NXRmodMio1sL4nLtWA01%3Daay2T6PSaYK8wEtsmw-i8g%40mail.gmail.com
>> .
>>
>

Matthias Koeppe

unread,
Dec 19, 2023, 2:44:00 PM12/19/23
to sage-devel
Meanwhile various PRs that make improvements to the CI are waiting for review:

John Cremona

unread,
Dec 19, 2023, 2:59:33 PM12/19/23
to SAGE devel


On Tue, 19 Dec 2023, 19:44 Matthias Koeppe, <matthia...@gmail.com> wrote:
Meanwhile various PRs that make improvements to the CI are waiting for review:


I see that one of those is relevant to one of my PRs as it mentions changes in the m4 configure script.

I'm afraid that I am not competent to review these.

John


Kwankyu Lee

unread,
Dec 19, 2023, 4:21:13 PM12/19/23
to sage-devel
I would have assumed that no-one would look at a PR whose automatic tests were failing, since that *should* indicate that something is wrong with the PR.

+++1 
 
 My complaint is that even with a PR based on the current develop branch, tests are failing which are not caused by the PR.

+++1

Be relieved, as people are learning to ignore checks, painfully. 

All this is related with the danger in the bottom of our ship (disputed PRs on CI). 

Dima Pasechnik

unread,
Dec 19, 2023, 4:37:46 PM12/19/23
to sage-...@googlegroups.com
Call me a Luddite, but let me remind you that "disputed PRs on CI" appear to be the only way to engage in discussions where the project is heading, and whether it is sustainable this way.

My point of view it's got to either be split off into parts, getting rid of one mega-repo, or trimmed down to bring the number of packages down.
For the latter, I propose to remove gcc/gfortran packages, for starters. They are among the biggest, and among the most useless.




>

Kwankyu Lee

unread,
Dec 19, 2023, 4:57:54 PM12/19/23
to sage-devel
Call me a Luddite, but let me remind you that "disputed PRs on CI" appear to be the only way to engage in discussions where the project is heading, and whether it is sustainable this way.

That is a noble discussion. I humbly respect that. But we should not stop the machines, for that.
Reply all
Reply to author
Forward
0 new messages