On Sat, 29 Sep 2018 at 10:22:06 -0400, Jeremy Bicha wrote:
> I believe we don't consider this a bug in glib2.0.
The more I look at this bug, the less sure I am about the rationale
given for preferring absolute paths in these particular variables, so
I've reopened the upstream bug asking for clarification.
The recommendation given is appropriate for directories into which a
dependent package wants to install e.g. GIO plugins, but I'm not so sure
it's appropriate for build-time tools.
I'm not reverting the change and also not patching the dependent projects
right now, because we need to get this resolved upstream one way or
another (either going back to the previous .pc file contents consistently
across Autotools and Meson builds, or keeping the new .pc file contents
consistently across Autotools and Meson builds, and either way providing
clear recommendations for dependent projects).
smcv
