Re: [PATCH v2 17/21] Convert refresh_index to take struct pathspec

On Fri, Jan 11, 2013 at 06:21:11PM +0700, Nguyễn Thái Ngọc Duy wrote:

-	for (i = 0; i < specs; i++) {
+	for (i = 0; i < pathspec->nr; i++) {
 		if (!seen[i])
-			die(_("pathspec '%s' did not match any files"), pathspec[i]);
+			die(_("pathspec '%s' did not match any files"), pathspec->raw[i]);
 	}

This needs the following fixup on top. I don't want to send another
reroll just a couple hours after I flooded git@vger. I did not plan to
work on the series this soon but somehow another problem got me back
here.

-- 8< --

diff --git a/builtin/add.c b/builtin/add.c
index 1235eb9..e1bcdb9 100644
--- a/builtin/add.c
+++ b/builtin/add.c

@@ -159,7 +159,8 @@ static void refresh(int verbose, const struct pathspec *pathspec)
 		      pathspec, seen, _("Unstaged changes after refreshing the index:"));
 	for (i = 0; i < pathspec->nr; i++) {
 		if (!seen[i])
-			die(_("pathspec '%s' did not match any files"), pathspec->raw[i]);
+			die(_("pathspec '%s' did not match any files"),
+			    pathspec->items[i].match);
 	}
         free(seen);
 }
-- 8< --

and the baaad reason: pathspec->items[] are sorted because of 86e4ca6
(tree_entry_interesting(): fix depth limit with overlapping pathspecs
- 2010-12-15). But raw[] are _not_. So raw[i] does not correspond to
item[i].

Now seen[] array returned from match_pathspec() has the order
corresponding to raw[]. On the other hand match_pathspec_depth()
returns seen[] corresponds to items[]. This patch converts
match_pathspec() to match_pathspec_depth() so we need to use the
correct pathspec array.

I'll put these explanation in the next reroll. And don't worry about
this subtle difference. My next email kills match_pathspec() for good.