Re: catching broken deps

[Hans de Goede <j w r degoede hhs nl>]

Jesse Keating wrote:
> On Thu, 25 Oct 2007 16:06:11 +0200
> Hans de Goede <j w r degoede hhs nl> wrote:
>
> > Not at the moment, but before sinking time into this I would first
> > like to see some commitment to / interest in checks like these.
>
> I don't think anybody is going to say no to adding some checks in to at
> least warn a maintainer that the build they submitted may cause
> problems.  I don't think we want to fail any builds yet, especially
> since we don't have an easy way to "unfail" the build.  So
> informational only.

Good I was expecting / hoping to hear something like that.

> > If we are going to do this I have a few questions:
> > -What languages would be acceptable to write such checks in?
>
> Python would be preferred.

Hmm, not my field of expertise I'm afraid.

> > -What kind of access to the repository for which is being build is
> > available on the buildsys. (mirror mounted as filesystem / yum
> > only / ..)
>
> Within the buildroot itself there should be access to the build
> repository, but this doesn't necessarily match the released repos.  My
> best recommendation would be to construct a yum object and point to a
> known local repo (http) to gather repodata information.

Hmm, even further away from my field of expertise I'm afraid, I was thinking 
about some ugly bash scripting using repoquery and who knows what other kind of 
ugliness. Any volunteers for writting somthing clean to run on the buildsys?

Maybe in the mean time its an idea to add an ugly version to the Makefile's 
make <arch> command's, then atleast those can warn the maintainer when he does 
a local test build, maybe we can even patch it into make mockbuild?

Regards,

Hans