gitk (was Re: Linux v2.6.12-rc6)

gitk (was Re: Linux v2.6.12-rc6)

[Linus Torvalds]

Btw, Paul,
 one thing I end up missing in gitk is that looking at merges doesn't show 
any output, yet in many ways merges are actually the most interesting 
ones where you'd like to see where a file comes from..

Now, handling multi-way merges may be hard, but at least the normal 
two-way merges end up pretty straightforward, I think, and only have four 
cases for each file:
 - not touched by either side (ie it doesn't show up in either of 
   "git-diff-tree TREE PARENT1" or "git-diff-tree TREE PARENT2")
 - changed by PARENT1 only (ie it shows up in the git-diff-tree between 
   TREE and PARENT2)
 - changed by PARENT2 only
 - changed since both

and it would be cool if the "filename list" panel on the right side 
colorized the names by these things (only three cases - leave the "not 
touched" files off entirely, of course).

Also, it should actually possible to do a diff for these things with just
doing a "diff3" on the two parents and on the merge result (diff3 normally
expects the "original", but hey, the "original" might as well be the
result), and thus color a merge file by whether the lines came from one
side or the other or both (or neither, which would be a manual merge
fixup).

Maybe I'm crazy. But it _seems_ like a good idea that shouldn't be 
fundamentally hard.

		Linus

Re: gitk (was Re: Linux v2.6.12-rc6)

[Paul Mackerras]

Linus Torvalds writes:

> and it would be cool if the "filename list" panel on the right side 
> colorized the names by these things (only three cases - leave the "not 
> touched" files off entirely, of course).

Hmmm.  How about this: each parent is given a color, and the file
names are colored according to which parent they differ in.  If a file
differs in more than one parent it gets listed more than once in the
colors for the parents in which it differs, with those entries being
consecutive in the list.

If instead I list the file once in yet another color, I don't see how
to make it intuitively obvious that that's what the color means, and I
get a combinatorial explosion in the number of colors required as the
number of parents grows (I would potentially need 2^n - 1 colors for n
parents).

> Also, it should actually possible to do a diff for these things with just
> doing a "diff3" on the two parents and on the merge result (diff3 normally
> expects the "original", but hey, the "original" might as well be the
> result), and thus color a merge file by whether the lines came from one
> side or the other or both (or neither, which would be a manual merge
> fixup).

I already have code in dirdiff to display n-way diffs in a single
window, so it's certainly doable.  I think I will do it a little
differently in this case, though, because the final result has a
special status.  If I color lines according to which parent(s) they
are present in and then have some other way to show which version was
chosen (maybe a black box around those lines? or a little tick (check
mark) or cross icon at the left hand edge of the line?) I think that
should be pretty clear.  (I'll need another color for the cases where
the result is different to all of the parents, of course.)

Sounds interesting.  I'll have a look at it.

Paul.