[cairo] Cairo and the GitLab pipeline

[Bryce Harrington]

On Wed, Mar 13, 2019 at 05:24:51PM +0100, Uli Schlachter wrote:
> Hi everyone,
> 
> since the floating point formats were merged [1], cairo depends on
> pixman 0.36.0. This made the pipeline on GitLab fail earlier than
> before, because now the provided pixman version is too old. In [2], I
> propose to switch the pipeline to Fedora rawhide to get a newer pixman
> version (*). Are there any thoughts / comments / review for this?

I suppose it depends on what the testing goal for this should be(*).

Tracking a development version of a distro, like Rawhide, will get data
from building against the current versions of dependencies, and thus be
helpful from an integration testing POV.  On the other hand, Cairo's
test suite compares render output against ref images that can depend on
Pixman, so if Pixman changes in Rawhide, keeping the Cairo testsuite
passing may be tougher to manage.

Tracking a more stable version of a distro, like Latest, will keep
dependencies more to static, stable versions, and thus the Cairo
testsuite will be more likely to continue to run consistently, and only
fail when there are regressions in Cairo itself.  The downside is that
as Cairo's required dependency versions change, as in this case, it will
require manual updating.

Either approach is good, just depends on what types of issues you're
most interested in.



* -- I see from the MR this change is suggested to be a temporary
workaround, but it's still a good question, and one that's bound to come
up again in the future, so hopefully the above is good food for thought.

>From a scientific experimentation POV, where you want to minimize the
number of variables, an approach might be to do both approaches, but run
stable Cairo on the development distro, and development Cairo on the
stable distro.

> Cheers,
> Uli
> 
> [1]:
> https://gitlab.freedesktop.org/cairo/cairo/commit/a34cb719cd9cb4f0c5b78be80b80ab0ae22464a6
> [2]: https://gitlab.freedesktop.org/cairo/cairo/merge_requests/14
> (*): And I dare to call this measure "temporary". I look forward to
> figuring out how long "temporary" is.
> -- 
> "Are you preparing for another war, Plutarch?" I ask.
> "Oh, not now. Now we're in that sweet period where
> everyone agrees that our recent horrors should never
> be repeated," he says.
> -- 
> cairo mailing list
> cairo at cairographics.org
> https://lists.cairographics.org/mailman/listinfo/cairo