egl compile error on msys CI runners

egl compile error on msys CI runners

[Peter Maydell]

The msys CI runners have started failing to compile with an EGL
related error:

https://gitlab.com/qemu-project/qemu/-/jobs/4015682307

In file included from
C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl.h:46,
from C:/GitLab-Runner/builds/qemu-project/qemu/include/ui/egl-helpers.h:5,
from ../ui/egl-helpers.c:21:
C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl_generated.h:11:10:
fatal error: EGL/eglplatform.h: No such file or directory
11 | #include "EGL/eglplatform.h"
| ^~~~~~~~~~~~~~~~~~~
ninja: build stopped: subcommand failed.

Dan says this is because of a msys2 packaging issue:
https://github.com/msys2/MINGW-packages/issues/16492

Can we have a meson test that doesn't try to enable EGL on systems
like this one where it's broken ? Changing the current meson.build
  if cc.has_header('epoxy/egl.h', dependencies: epoxy)
to whatever the meson is for "actually try to compile with the
header" ought to do the trick, I think.

thanks
-- PMM

Re: egl compile error on msys CI runners

[Marc-André Lureau]

Hi Peter

On Tue, Mar 28, 2023 at 6:39 PM Peter Maydell <peter.maydell@linaro.org> wrote:
>
> The msys CI runners have started failing to compile with an EGL
> related error:
>
> https://gitlab.com/qemu-project/qemu/-/jobs/4015682307
>
> In file included from
> C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl.h:46,
> from C:/GitLab-Runner/builds/qemu-project/qemu/include/ui/egl-helpers.h:5,
> from ../ui/egl-helpers.c:21:
> C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl_generated.h:11:10:
> fatal error: EGL/eglplatform.h: No such file or directory
> 11 | #include "EGL/eglplatform.h"
> | ^~~~~~~~~~~~~~~~~~~
> ninja: build stopped: subcommand failed.
>
> Dan says this is because of a msys2 packaging issue:
> https://github.com/msys2/MINGW-packages/issues/16492
>
> Can we have a meson test that doesn't try to enable EGL on systems
> like this one where it's broken ? Changing the current meson.build
>   if cc.has_header('epoxy/egl.h', dependencies: epoxy)
> to whatever the meson is for "actually try to compile with the
> header" ought to do the trick, I think.
>

This is an unfortunate consequence of using a fast-moving rolling
distribution, there are random breakages. It would make sense to have
tagged / pre-built images instead. That would also speed up the CI a
bit, hopefully.
Thomas, any idea how to do that?


-- 
Marc-André Lureau

Re: egl compile error on msys CI runners

[Thomas Huth]

On 29/03/2023 09.01, Marc-André Lureau wrote:
> Hi Peter
> 
> On Tue, Mar 28, 2023 at 6:39 PM Peter Maydell <peter.maydell@linaro.org> wrote:
>>
>> The msys CI runners have started failing to compile with an EGL
>> related error:
>>
>> https://gitlab.com/qemu-project/qemu/-/jobs/4015682307
>>
>> In file included from
>> C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl.h:46,
>> from C:/GitLab-Runner/builds/qemu-project/qemu/include/ui/egl-helpers.h:5,
>> from ../ui/egl-helpers.c:21:
>> C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl_generated.h:11:10:
>> fatal error: EGL/eglplatform.h: No such file or directory
>> 11 | #include "EGL/eglplatform.h"
>> | ^~~~~~~~~~~~~~~~~~~
>> ninja: build stopped: subcommand failed.
>>
>> Dan says this is because of a msys2 packaging issue:
>> https://github.com/msys2/MINGW-packages/issues/16492
>>
>> Can we have a meson test that doesn't try to enable EGL on systems
>> like this one where it's broken ? Changing the current meson.build
>>    if cc.has_header('epoxy/egl.h', dependencies: epoxy)
>> to whatever the meson is for "actually try to compile with the
>> header" ought to do the trick, I think.
>>
> 
> This is an unfortunate consequence of using a fast-moving rolling
> distribution, there are random breakages. It would make sense to have
> tagged / pre-built images instead. That would also speed up the CI a
> bit, hopefully.
> Thomas, any idea how to do that?

Sorry, I've got no clue how that container stuff works on Windows :-(

  Thomas

Re: egl compile error on msys CI runners

[Daniel P. Berrangé]

On Wed, Mar 29, 2023 at 11:01:35AM +0400, Marc-André Lureau wrote:
> Hi Peter
> 
> On Tue, Mar 28, 2023 at 6:39 PM Peter Maydell <peter.maydell@linaro.org> wrote:
> >
> > The msys CI runners have started failing to compile with an EGL
> > related error:
> >
> > https://gitlab.com/qemu-project/qemu/-/jobs/4015682307
> >
> > In file included from
> > C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl.h:46,
> > from C:/GitLab-Runner/builds/qemu-project/qemu/include/ui/egl-helpers.h:5,
> > from ../ui/egl-helpers.c:21:
> > C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl_generated.h:11:10:
> > fatal error: EGL/eglplatform.h: No such file or directory
> > 11 | #include "EGL/eglplatform.h"
> > | ^~~~~~~~~~~~~~~~~~~
> > ninja: build stopped: subcommand failed.
> >
> > Dan says this is because of a msys2 packaging issue:
> > https://github.com/msys2/MINGW-packages/issues/16492
> >
> > Can we have a meson test that doesn't try to enable EGL on systems
> > like this one where it's broken ? Changing the current meson.build
> >   if cc.has_header('epoxy/egl.h', dependencies: epoxy)
> > to whatever the meson is for "actually try to compile with the
> > header" ought to do the trick, I think.
> >
> 
> This is an unfortunate consequence of using a fast-moving rolling
> distribution, there are random breakages. It would make sense to have
> tagged / pre-built images instead. That would also speed up the CI a
> bit, hopefully.

IMHO if the distro is considered susceptible to ongoing random
breakage we must not use it as a gating CI job. IOW, these
msys jobs should have 'allow-failure: true' added to their
definition.

With regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|

Re: egl compile error on msys CI runners

[Thomas Huth]

On 29/03/2023 10.07, Daniel P. Berrangé wrote:
> On Wed, Mar 29, 2023 at 11:01:35AM +0400, Marc-André Lureau wrote:
>> Hi Peter
>>
>> On Tue, Mar 28, 2023 at 6:39 PM Peter Maydell <peter.maydell@linaro.org> wrote:
>>>
>>> The msys CI runners have started failing to compile with an EGL
>>> related error:
>>>
>>> https://gitlab.com/qemu-project/qemu/-/jobs/4015682307
>>>
>>> In file included from
>>> C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl.h:46,
>>> from C:/GitLab-Runner/builds/qemu-project/qemu/include/ui/egl-helpers.h:5,
>>> from ../ui/egl-helpers.c:21:
>>> C:/GitLab-Runner/builds/qemu-project/qemu/msys64/mingw64/include/epoxy/egl_generated.h:11:10:
>>> fatal error: EGL/eglplatform.h: No such file or directory
>>> 11 | #include "EGL/eglplatform.h"
>>> | ^~~~~~~~~~~~~~~~~~~
>>> ninja: build stopped: subcommand failed.
>>>
>>> Dan says this is because of a msys2 packaging issue:
>>> https://github.com/msys2/MINGW-packages/issues/16492
>>>
>>> Can we have a meson test that doesn't try to enable EGL on systems
>>> like this one where it's broken ? Changing the current meson.build
>>>    if cc.has_header('epoxy/egl.h', dependencies: epoxy)
>>> to whatever the meson is for "actually try to compile with the
>>> header" ought to do the trick, I think.
>>>
>>
>> This is an unfortunate consequence of using a fast-moving rolling
>> distribution, there are random breakages. It would make sense to have
>> tagged / pre-built images instead. That would also speed up the CI a
>> bit, hopefully.
> 
> IMHO if the distro is considered susceptible to ongoing random
> breakage we must not use it as a gating CI job. IOW, these
> msys jobs should have 'allow-failure: true' added to their
> definition.

I'm not a fan of 'allow-failure: true' - that basically means nobody will 
look at the failing jobs, so we could rather disable the jobs by default 
instead (and just run them manually when you want to have a look at them).

  Thomas

Re: egl compile error on msys CI runners

[Peter Maydell]

On Wed, 29 Mar 2023 at 09:20, Thomas Huth <thuth@redhat.com> wrote:
> I'm not a fan of 'allow-failure: true' - that basically means nobody will
> look at the failing jobs, so we could rather disable the jobs by default
> instead (and just run them manually when you want to have a look at them).

Mmm; in fact I think we could probably look at upgrading some of
our current 'allow_failure: true' jobs to normal ones -- the
cirrus bsd and macos ones seem to pass pretty reliably these days.

thanks
-- PMM