Suggestion for cvs tag handling

[Jarod Wilson]

On Thursday 06 March 2008 05:57:55 pm Jason L Tibbitts III wrote:
> >>>>> "JS" == Jochen Schmitt,
> >>>>> "JK" == Jesse Keating <jkeating at redhat.com> writes:
>
> JS> The user/developers should be able to create tag in the form
> JS> foo-1_1-1_fc8 and can still use the -F flag.
>
> JS> If a build was started on koji based on the requested cvs tag a new
> JS> tag in the form _build_foo-1_1-1_fc8 may be created.
>
> JK> We don't want koji writing to the SCM, and thus the tag should
> JK> come from the user via make tag.
>
> Well, I guess you could implement Jochen's suggestion by having "make
> build" create the immutable tag, but I kind of wonder as to the point
> of tagging without building.  The only thing I can think of is that
> you want to tag and then have someone else take a look at the package
> before you build, but in that case I don't understand why they
> wouldn't just checkout the head anyway.

I dislike the idea of immutable tags before a successful build completes. More 
than once, I've simply forgot to cvs add a patch before submitting a build, 
and bumping the spec and using a new tag simply to add a patch that is 
already referenced in the spec is obnoxious. I greatly prefer to just cvs add 
the patch, force-tag and resubmit the build as it was supposed to be.

Now, perhaps if there was a pre-commit hook that checked to see that all 
referenced Sources and Patches were actually available, and prevent a checkin 
w/a new patch referenced in a spec that isn't cvs added, then I might be less 
opposed to immutable tagging before building...

-- 
Jarod Wilson
jwilson at redhat.com