After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 773674 - meson: don't use subproject fallback for gst-validate if it won't work
meson: don't use subproject fallback for gst-validate if it won't work
Status: RESOLVED OBSOLETE
Product: GStreamer
Classification: Platform
Component: gst-editing-services
git master
Other Linux
: Normal minor
: NONE
Assigned To: GStreamer Maintainers
GStreamer Maintainers
Depends on:
Blocks:
 
 
Reported: 2016-10-29 16:19 UTC by Tim-Philipp Müller
Modified: 2016-11-11 00:38 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
meson: don't use subproject fallback for gst-validate if it won't work (1.46 KB, patch)
2016-10-29 16:19 UTC, Tim-Philipp Müller
accepted-commit_now Details | Review

Description Tim-Philipp Müller 2016-10-29 16:19:26 UTC
Created attachment 338776 [details] [review]
meson: don't use subproject fallback for gst-validate if it  won't work

Otherwise it will fail.
Comment 1 Thibault Saunier 2016-10-30 13:53:26 UTC
Review of attachment 338776 [details] [review]:

I do not really like that, it is becoming ugly to check for json-glib everywhere. I proposed adding meson support in json-glib(0), I believe it will get merge soon, meanwhile I could use my branch in a wrap, not sur what you think?

(0) https://bugzilla.gnome.org/show_bug.cgi?id=773603
Comment 2 Tim-Philipp Müller 2016-10-30 14:07:21 UTC
I don't know about having to check "everywhere" :)

I think we want something like an 'allow_fallback' kwarg added to subproject() in meson.

This patch is mostly about fixing gst-build breakage today. I know it's not very nice, but it's also not super-ugly. If a better solution comes up we can switch to using that.

Alternatively we can also not build ges by default in gst-build.
Comment 3 Tim-Philipp Müller 2016-10-30 14:09:41 UTC
Oh, sorry, I forgot to answer your question: I don't want us to use inofficial branches. I propose to apply this for the time being and when it's merged we switch over to that :)
Comment 4 Thibault Saunier 2016-10-30 14:10:14 UTC
Review of attachment 338776 [details] [review]:

Also this should not be needed as gst-validate is optionnal so if subproject fails it should not fail the build, bug in mesón?
Comment 5 Thibault Saunier 2016-10-30 14:11:42 UTC
(In reply to Tim-Philipp Müller from comment #3)
> Oh, sorry, I forgot to answer your question: I don't want us to use
> inofficial branches. I propose to apply this for the time being and when
> it's merged we switch over to that :)

Fait enough.
Comment 6 Tim-Philipp Müller 2016-10-30 14:19:38 UTC
> Also this should not be needed as gst-validate is optional so if
> subproject fails it should not fail the build, bug in mesón?

Not sure I follow. gst-validate as external dependency is optional, but the fallback ensures that if it's not found as external dependency a subproject() will be used. We will then hard-require the hard-dependencies of the subproject. This is what this patch addresses. I think Meson is working as intended in this case.

(There is another bug concerning the libxml dependency checking which we'll have to work around, but it's not directly related to this as far as I can tell - https://github.com/mesonbuild/meson/issues/964).
Comment 7 Thibault Saunier 2016-10-30 14:27:08 UTC
I think I disagree that behaviour is correct as the the dpendency is explicitly marked as 'required : false' which should mean, fmpov, that in any case, no fallback avalaible, or fallback build failling, we should just keep going without that dep.
Comment 8 Thibault Saunier 2016-10-30 14:31:30 UTC
Review of attachment 338776 [details] [review]:

Go ahead anyway we already started in that dirección and we should finish it.
Comment 9 Tim-Philipp Müller 2016-10-30 14:32:54 UTC
Interesting. I'm not aware that it is supposed to work like that already, but there might be an argument for doing it like you say. Something for Meson devs.
Comment 10 Tim-Philipp Müller 2016-11-10 23:54:52 UTC
Apparently this is supposed to work already, just not sure in which version:
https://github.com/mesonbuild/meson/issues/975
Comment 11 Thibault Saunier 2016-11-11 00:30:03 UTC
OK, and I fixed it? :) It is in 0.35 so I am not sure why you opened that bug? :)
Comment 12 Tim-Philipp Müller 2016-11-11 00:38:57 UTC
Because it was actually failing for me, but I'm not 100% sure now if it was with 0.34 or 0.35. So let's just close this.