Jeff King venit, vidit, dixit 29.03.2011 21:01:

On Tue, Mar 29, 2011 at 10:35:47AM -0400, Jeff King wrote:

quoted

On Tue, Mar 29, 2011 at 12:05:09PM +0200, Michael J Gruber wrote:

quoted

-		if (!prefixcmp(arg+13, "refs/"))
+		if (!prefixcmp(arg+offset, "refs/"))
 			/* happy */;
-		else if (!prefixcmp(arg+13, "notes/"))
+		else if (!prefixcmp(arg+offset, "notes/"))
 			strbuf_addstr(&buf, "refs/");
 		else
 			strbuf_addstr(&buf, "refs/notes/");
-		strbuf_addstr(&buf, arg+13);
+		strbuf_addstr(&buf, arg+offset);
 		string_list_append(revs->notes_opt.extra_notes_refs,
 				   strbuf_detach(&buf, NULL));

This issue is not introduced by your patch, but maybe it is a good
opportunity to refactor this to use expand_notes_ref from notes.c?

Oops, I just realized this is in builtin/notes.c in master. I had
already written a patch for another topic that made it globally
accessible. :)

Yeah, I (figured and) factored it out myself meanwhile, and rebased. I'm
wondering though where we are going. Junio seems to be in a mood for
major changes to the notes ui, so maybe I should hold on until we
decided about a ui restructuring.

I think, though, that any notes ui revamp is correlated with our
(stalled?) discussions about the layout of refs/. It affects not only
the default notes ref ("commits" for all notes?) but also the question
what a standard notes ref is, and where to store (and how to specify)
upstream notes refs.

Michael