On Tue, 2010-03-16 at 08:14 -0400, Kamil Paral wrote:
----- "James Laska" <jlaska(a)redhat.com> wrote:
> On Tue, 2010-03-16 at 11:45 +0100, Kamil Páral wrote:
> > autoqa.conf and repoinfo.conf are now installed by 'make install'
> provided
> > that they don't already exist. The same holds for autoqa.cron, it is
> now
> > installed only if it does not already exist.
> > Rationale: We don't want to have config files overwritten every time
> we
> > do 'make install' when developing. We want similar behavior as RPM
> > installation does.
>
> Seems sensible to me, and you've wrapped it around a conditional. I
> can't recall if we also install irb.cfg in the 'make install'
> process.
> If so, that should probably be included in this?
>
> $ rpm -qpc autoqa-0.3.1-1.fc12.noarch.rpm
> /etc/autoqa.conf
> /etc/cron.d/autoqa
> /etc/repoinfo.conf
> /usr/share/autotest/client/site_tests/rats_sanity/irb.cfg
That's covered by this line in Makefile:
for t in tests/*; do cp -a $$t $(PREFIX)$(TEST_DIR); done
It copies all the tests as a pack. I can modify it to skip
irb.conf and then copy it manually. But shouldn't we just
place this file into /etc ? I can see for example this
structure:
/etc/cron.d/autoqa
/etc/autoqa/autoqa.conf
/etc/autoqa/repoinfo.conf
/etc/autoqa/tests/rats_sanity/irb.cfg
Nice, I like that structure better for test specific configs.
Are any changes in the irb tests (rats_*) needed to change to
accommodate moving the config around? I think Will played around with
this a while back. The reason irb.cfg exists inside the test directory
is due to the nature in which autotest rsync's content to the test
clients. I think it would be okay to move this cfg into an appropriate
system-wide config directory ... but just something to consider when
moving it.
Thanks,
James