Re: [PATCH] builtin-clone: fix for new unpack_trees() semantics

[Johannes Schindelin <Johannes.Schindelin@gmx.de>]

Hi,

On Sun, 9 Mar 2008, Daniel Barkalow wrote:

> On Sun, 9 Mar 2008, Johannes Schindelin wrote:
> 
> > On Sun, 9 Mar 2008, Daniel Barkalow wrote:
> > 
> > > On Sun, 9 Mar 2008, Johannes Schindelin wrote:
> > > 
> > > > In git.git's "next" branch, unpack_trees() must specify source and target 
> > > > index.
> > > > 
> > > > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > > > ---
> > > > 
> > > > 	To be squashed into 10/11
> > > > 
> > > >  builtin-clone.c |    2 ++
> > > >  1 files changed, 2 insertions(+), 0 deletions(-)
> > > > 
> > > > diff --git a/builtin-clone.c b/builtin-clone.c
> > > > index e4047ed..3890e12 100644
> > > > --- a/builtin-clone.c
> > > > +++ b/builtin-clone.c
> > > > @@ -534,6 +534,8 @@ int cmd_clone(int argc, const char **argv, const char *prefix)
> > > >  		opts.verbose_update = !option_quiet;
> > > >  		opts.merge = 1;
> > > >  		opts.fn = twoway_merge;
> > > > +		opts.src_index = &the_index;
> > > > +		opts.dst_index = &the_index;
> > > 
> > > Actually, I think the sensible thing is to just not do a merge here, since 
> > > we know there's no index beforehand and the two trees are the same.
> > > 
> > > I think the odd twoway merge of two copies of HEAD is just an artifact of 
> > > clone originally just doing "git checkout HEAD", and that got translated 
> > > various times failing to notice the special cases.
> > 
> > Okay, but would oneway_merge not want to write the index, too (rightfully 
> > so)?
> 
> I'm thinking:
> 
>   memset(&opts, 0, sizeof opts);
>   opts.update = 1;
>   opts.verbose_update = !option_quiet;
>   opts.dst_index = &the_index;
> 
>   init_tree_desc(&t[0], tree->buffer, tree->size);
>   unpack_trees(1, t, &opts);
> 
> That is, write it, but not read it, and only have one tree.

Yes.  And this sets dst_index (what I tried to hint at with my patch).

Ciao,
Dscho