On 3 March 2010 00:24, James Laska <jlaska(a)redhat.com> wrote:
The scenario noted in previous emails of installing F-12 and then
attempting to update using gpk-update-viewer has a few too many moving
parts, I suspect it would be a challenge to maintain. Is there some
smaller test scenario to consider that could be used to identify the
conditions that led to this failure? Are there existing PK unittests
that might have discovered this issue that would be desirable to run on
every build?
There are a few hundred tests that PackageKit runs before any release
(make check) -- these primarily test the code and the "dummy" (fake)
backend for consistency.
It would be very easy (but quite a lot of work) to set up a fake repo
for PackageKit and then test the various yum install / update / remove
combination for consistency.
This would be a great things to add as part of a GSoC or something
similar, it just needs somebody to "own" it.
Richard.