Re: The journey with vendor branch in git

[YONETANI Tomokazu <qhwt+dfly@xxxxxxxxxx>]

On Fri, Dec 19, 2008 at 09:11:57AM +0200, Hasso Tepper wrote:
> YONETANI Tomokazu wrote:
> > On Thu, Dec 18, 2008 at 12:39:06PM +0200, Hasso Tepper wrote:
> > > 4) Now I have to switch to master and merge? This is how my journey
> > > stopped:
> > >
> > > $ git checkout master
> > > $ git merge --no-commit OPENPAM
> >
> > Since you know exactly which commit you want to merge, what about this?
> > $ git merge --no-commit OPENPAM^ OPENPAM
> 
> It doesn't complain, yeah, but I'm not sure whether it's a right thing to 
> do. Is it?

Well, I think both should give you the same result; the same changes
and the same merge parents.  We know that the common ancestor 7bce828
has already been committed to master as 863a10d, but...

| $ GIT_MERGE_VERBOSITY=10 git merge --no-commit origin/vendor/OPENPAM
| Merging HEAD with origin/vendor/OPENPAM
| Merging:
| 249a1dc wrong initializer in VFS_SET; maybe someone'll commit it :)
| f23594c Import OpenPAM Hydrangea.
| found 1 common ancestor(s):
| 7bce828 Import pam_deny, pam_permit and pam_unix from FreeBSD, use them instead
| CONFLICT (rename/delete): Renamed contrib/openpam/modules/pam_deny/pam_deny.c->lib/pam_module/pam_opieaccess/pam_opieaccess.c in HEAD and deleted in origin/vendor/OPENPAM
| CONFLICT (rename/delete): Renamed contrib/openpam/modules/pam_permit/pam_permit.c->usr.bin/truss/extern.h in HEAD and deleted in origin/vendor/OPENPAM
| Automatic merge failed; fix conflicts and then commit the result.

for some reason git thinks a deletion and an addition of two unrelated files
to be a rename (as others mentioned on commits@), I don't really know about
this part.  Anyway if you take the default merge method, you have to ignore
this confusing message and commit.  If you take the `octopus' method, you
don't have to see the conflict in this case, but you need to know which
versions to merge.