[Ovirt-devel] How to handle 3rd party code?

[Jason Guiditta]

On Thu, 2008-10-23 at 12:26 -0400, Darryl Pierce wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Darryl Pierce wrote:
> >> So what we have here is a collection of code from different sources that
> >> actually forms part of our app. On top of that we have some
> >> ovirt-specific customizations to the javascript libraries which does
> >> need to live in our repo.
> > 
> > So, in that case, isn't the solution still a separate RPM but one that's
> > specific to our product? IOW, if there's a javascript package called
> > "foo", then we would potentially have a foo.rpm and a foo-ovirt.rpm that
> > would apply any patches we have to customize it to our need?
> 
> Sorry, in this case, I meant to include that a dependant rpm that would
> bridge the javascript code to our code; i.e., one that would create or
> update the necessary symlinks to bring the code into
> /usr/share/ovirt-server as needed.
> 
> That way the third party code wouldn't need to packaged more than once
> for any project to use it.

Again, the issue here is we still need to version our changes.  I could
see this possibly working for plugins since we only have a few (though
we did have some customizations there at one point as well) and they are
more likely to be available as gems.  Also, for the javascript, there
are tons of little components/plugins, so splintering them all into
separate rpms would be a really large number of dependencies, not to
mention rpm seems like an odd way to handle javascript files.  I agree
with not keeping the code in our main repo, that is already getting
annoying.  This is why I brought up sub-modules, which I believe Alan
and/or Jim had mentioned as a solution we should consider.

-j