I noticed that there is no safe way to give a revision to git-blame:
it can always be interpreted as an existing file:
 - "git blame rev -- file.txt" can fail if "rev" is a file
 - "git blame rev^0 -- file.txt" can file there is a "rev^0" file
   (happens if you type too fast with a tool not designed for keyboard)
 - "git blame file.txt rev" is ambiguos too, for the same reasons.

I did the simple patch (below) to resolve at least the very first one,
just because that is how git-rev-list does it.
But if all forms of git-blame command line are expected to live, a
more serious surgery of the argv[] handling code needed.
And I afraid the patch has a small chance of crashing: I don't check
if there is enough space in argv (don't actually even know how to),
so Junio, please do not apply it (it passes blame tests, though).
I also suspect git-blame is not the only command using revision
machinery which has the same problem, so this message is more like a
discussion starter.

diff --git a/builtin-blame.c b/builtin-blame.c
index 066dee7..83c8905 100644
--- a/builtin-blame.c
+++ b/builtin-blame.c
@@ -1787,6 +1787,7 @@ int cmd_blame(int argc, const char **arg
        /* Now we got rev and path.  We do not want the path pruning
         * but we may want "bottom" processing.
         */
+       argv[unk++] = "--";
        argv[unk] = NULL;
 
        init_revisions(&revs, NULL);

-
To unsubscribe from this list: send the line "unsubscribe git" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html