On Wed, Mar 11, 2009 at 07:28:31AM -0400, John Szakmeister wrote:
> On Wed, Mar 11, 2009 at 5:23 AM, Branko Cibej <brane_at_xbc.nu> wrote:
> > John Szakmeister wrote:
> >> (Can we do something like re-encode
> >> the unidiff as a delta stream and run it through our normal merge
> >> algorithm?) I can see this being useful outside of Subversion as
> >> well.
> >>
> >
> > Interesting thought, but I don't quite understand how that would work. I
> > assume my "delta stream" you mean just adding the ancestry metadata?
>
> I haven't really thought much about it. The thought I had running
> through my head was if there was a way to put the unidiff in a format
> that the client could more readily apply, then it might help to reduce
> the dependence on the patch exectuable for 'svn patch'. Thinking more
> about it, the delta is almost always converted back into a full-text,
> and then run through the line-based merging algorithm... so it's
> probably of no help. But I do think that getting rid of that
> dependence would be a Good Thing. Especially since the non-cygwin
> based version (I use the one from UnxUtils on sourceforge) breaks in
> fun ways.
Sounds like a fun research project! Can you add this to tasks.html?
More importantly though, we need mentors. So far, I think the only
person who explicitly stated they'd be willing to mentor was me.
Anyone else?
If we don't have enough mentors, it's not worth the hassle.
We'll need a few people willing to attentively review patches
posted by gsoc students. Not just one person.
I'd say two or three committers per student would fit the bill
and make success very likely. Applying this rule, it looks like
we currently have space for a grand total of 0 students.
Stefan
Received on 2009-03-11 12:48:37 CET