Re: [PATCH 4/4] Fix ie_match_stat for non-checked-out submodule

From: Junio C Hamano <gitster@pobox.com>
Date: May 2, 2008, 21:57 UTC
Message-ID: <7vod7owerk.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <1209735336-4690-5-git-send-email-pkufranky@gmail.com>

Ping Yin <pkufranky@gmail.com> writes:

> For submodules non checked out, ie_match_stat should always return 0.
> So in this case avoid calling is_racy_timestamp.

It is conceptually wrong to have this check in that function, I think.

Look at what ce_match_stat_basic() does. For S_IFGITLINK entries, we do not even compare the timestamps, so is_racy_timestamp() check should not even care and should return Ok for them.

Perhaps this patch would be a better, in that it covers the other caller on the write_index() callpath as well.

---

diff --git a/read-cache.c b/read-cache.c
index a92b25b..0a0ea3b 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -258,6 +258,7 @@ static int ce_match_stat_basic(struct cache_entry *ce, struct stat *st)
 static int is_racy_timestamp(const struct index_state *istate, struct cache_entry *ce)
 {
 	return (istate->timestamp &&
+		!S_ISGITLINK(ce->ce_mode) &&
 		((unsigned int)istate->timestamp) <= ce->ce_mtime);
 }