[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Qt-5 in core-updates
From: |
Ludovic Courtès |
Subject: |
Re: Qt-5 in core-updates |
Date: |
Mon, 25 Jan 2016 10:25:20 +0100 |
User-agent: |
Gnus/5.13 (Gnus v5.13) Emacs/24.5 (gnu/linux) |
address@hidden (宋文武) skribis:
> Andreas Enge <address@hidden> writes:
>
>> Hello,
>>
>> qt-5 fails to build on both x86_64 and armhf in core-updates with the
>> following error:
>> Traceback (most recent call last):
>> File
>> "/tmp/nix-build-qt-5.5.1.drv-0/qt-everywhere-opensource-src-5.5.1/qtwebkit/Source/WebKit2/Scripts/generate-messages-header.py",
>> line 28, in <module>
>> import webkit2.messages
>> EOFError: EOF read where object expected
>> Makefile.WebKit2.DerivedSources:881: recipe for target
>> 'generated/PluginProxyMessages.h' failed
>> make[3]: *** [generated/PluginProxyMessages.h] Error 1
>>
>> Surprisingly, it succeeds on i686 (and its build is disabled on mips).
>>
> This seems like a parallel build issue after I find that,
> Writing a pyc file is not atomic in python:
>
> <https://bugs.python.org/issue13146>
Ouch.
> It was fixed in Python 3.3+, but not in Python 2.7.x (the one for
> building WebKit).
>
> I think we can try:
> - Patch our python-2.7.x like Ubuntu does.
>
> <https://launchpad.net/ubuntu/+archive/primary/+files/python2.7_2.7.10-4ubuntu1.diff.gz>
Best if we can avoid it for now since that would further delay the merge
of ‘core-updates’.
> - Patch the build phases of qt and webkitgtk to pre-generate *.pyc
> files before the actually building.
If that’s simple enough, we could add a phase that run “make
whatever-target-it-is”, indeed.
> Or disable parallel build completely.
Sounds like the easiest thing to do. Not sure how long it would take to
build Qt, though.
> - Restart the builds on hydra, hope it will pass.
I think we should avoid that.
Andreas, are you on it?
Ludo’.