interesting discussion on git mailing list

James Westby jw+debian at jameswestby.net
Tue Jul 10 16:44:49 UTC 2007


Hi Martin,

On (10/07/07 11:55), martin f krafft wrote:
> also sprach Damyan Ivanov <dam at modsoftsys.com> [2007.07.10.1142 +0200]:
> > b) patches are easily disabled/enabled
> 
> the way to do this in git would be to have an integration branch and
> only merge those branches you want enabled. You cannot unmerge
> a branch, I think, though it should be trivial too. Then you could
> disable patches/branches.

'git reset --hard ORIG_HEAD' just after a pull undoes that merge. For
later on 'git revert' can do it, but I don't know about reverting a
whole branch at once with it.

> > e) patches are reviewable on source-package level
> 
> This is an important point, I agree. It would make sense actually to
> export source packages by converting branches to dpatches, like
> Canonical's HCT tried. Unfortunately, I am not sure where it stands
> or whether it's usable, but it should actually not be that hard to
> implement in a script.
> 

That would be good. A script should be possible. However HCT appears to
be dead and abandoned, and the source is not available as far as I can
see. There has been talk of moving to a successor like the loom plugin
on NoMoreSourcePackages, however I have not been able to get any
contact from anyone involved in that originally to discuss problems with
it and maybe move towards an implementation. I have had some
conversations with Reinhard Tartler about it, and he may work on an
implementation soon, but he has not commented on my concerns as yet.

The scheme described there is actually a lot more applicable to git that
to bzr, so it might be interesting to implement it on git first anyway.

Thanks,

James

-- 
  James Westby   --    GPG Key ID: B577FE13    --     http://jameswestby.net/
  seccure key - (3+)k7|M*edCX/.A:n*N!>|&7U.L#9E)Tu)T0>AM - secp256r1/nistp256



More information about the vcs-pkg-discuss mailing list