On Tue, 2007-10-23 at 15:14 -0400, rob myers wrote:
On Tue, 2007-10-23 at 14:30 -0400, Mike Bonnet wrote:
> On Tue, 2007-10-23 at 14:13 -0400, rob myers wrote:
> > attached is a patch that attempts to generalize checking out the
> > components used to build an SRPM. this patch supports CVS, GIT, and
> > SVN, but only CVS and SVN have been tested. the idea is to provide the
> > infrastructure for different SCM systems to be configured at run-time so
> > that users can choose their favorite system.
> It would also be nice if rather than assuming the existence of the
> "common/" module and hard-coding the "make srpm" command, there
> way to specify what modules were necessary to checkout, and what command
> to run to create a srpm. This should probably be configurable
> per-repository. It might make sense to add a new database table to
> store this information in a more flexible way than possible with the
> config files.
the existing design is (currently) sufficient for my needs. what you
describe is more flexible and is probably a better design, but is it
> However, this is a great start. I'm going to review the patch more
> carefully, and I'll see about merging it soon (after possibly removing
> the scmtype config option I mentioned earlier).
thanks for your feedback, and remove cruft as necessary. :)
The patch has been committed to Koji:
I made some pretty significant changes to make the SCM class more
object-y, remove some code duplication, make it more configurable at
runtime, etc., but the framework setup in the original patch is there.
I've tested it fairly completely and everything seems to be working as
expected. Note that when using the "+ssh://" access methods, you're
responsible for getting the required login credentials (host keys,
private keys, Kerberos tickets) into the right place on the builders
Thanks for the patch Rob, and if you have any other suggestions or
improvements, please let me know.
Keep those patches coming! :)