Re: What's cooking in git/spearce.git (topics)

[Jeff King <peff@peff.net>]

On Mon, Oct 22, 2007 at 02:32:22AM -0400, Shawn O. Pearce wrote:

> * lt/rename (Sun Oct 21 16:59:03 2007 -0700) 2 commits
>  - Linear-time/space rename logic (exact renames only)
>  - Split out "exact content match" phase of rename detection
> 
> t4001-diff-rename.sh failed while running tests on pu.  I'm pretty
> sure its this topic from Linus.  Need to look at it further.

Hrm, the problem is that it's not favoring basenames anymore. But I
think it is because the loop in find_identical_files is inside out. For
every source file, it picks the best destination file. But I think we
want to do the opposite: for every destination file, pick the best
source file. Otherwise, if a non-basename source file is connected with
a particular destination file, we no longer try that destination file
again.

Patch is below (please just squash with the one from Linus).

diff --git a/diffcore-rename.c b/diffcore-rename.c
index 05d39db..8881818 100644
--- a/diffcore-rename.c
+++ b/diffcore-rename.c
@@ -252,17 +252,18 @@ static int find_identical_files(struct file_similarity *src,
 {
 	int renames = 0;
 	do {
-		struct diff_filespec *one = src->filespec;
+		struct diff_filespec *one = dst->filespec;
 		struct file_similarity *p, *best;
 		int i = 100;
 
 		best = NULL;
-		for (p = dst; p; p = p->next) {
+		for (p = src; p; p = p->next) {
 			struct diff_filespec *two = p->filespec;
 
-			/* Already picked as a destination? */
+			/* Already picked as a source? */
 			if (!p->src_dst)
 				continue;
+
 			/* False hash collission? */
 			if (hashcmp(one->sha1, two->sha1))
 				continue;
@@ -276,10 +277,10 @@ static int find_identical_files(struct file_similarity *src,
 		}
 		if (best) {
 			best->src_dst = 0;
-			record_rename_pair(best->index, src->index, MAX_SCORE);
+			record_rename_pair(dst->index, best->index, MAX_SCORE);
 			renames++;
 		}
-	} while ((src = src->next) != NULL);
+	} while ((dst = dst->next) != NULL);
 	return renames;
 }