Re: tracking repository

From: Junio C Hamano <gitster@pobox.com>
Date: Mar 16, 2008, 23:01 UTC
Message-ID: <7vbq5eff3e.fsf@gitster.siamese.dyndns.org>
In-Reply-To: <7vzlsyfgjg.fsf@gitster.siamese.dyndns.org>

Junio C Hamano <gitster@pobox.com> writes:

Show 7 quoted lines

> Daniel Barkalow <barkalow@iabervon.org> writes:
>
>> Is "refs/*:refs/*" (mirror everything, including weird stuff) supposed to 
>> be prohibited?
>
> No.  In fact "remote add --mirror" actively creates such.  See my other
> message about design level issues.

I think something like this is needed. It still has an independent issue that this is now called by "git remote show" or "git remote prune", and it will die with a nonsense "refusing to create" error message, though.

The error, as far as I can tell, is half about a misconfigured config (e.g. "fetch = refs/heads/*:refs/remotes/[]?/*") and half about screwy remote repository (e.g. a misnamed "[]?" branch on the remote end can try to update a broken "refs/remotes/origin/[]?" even the configuration is a perfectly valid "fetch = refs/heads/*:refs/remotes/origin/*"). It may make sense to reword the error message to "ignoring" from "refusing" and do just that without dying here. I dunno.

 remote.c |    9 ++++++---
 1 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/remote.c b/remote.c
index f3f7375..fbcb03c 100644
--- a/remote.c
+++ b/remote.c
@@ -1007,9 +1007,12 @@ int get_fetch_map(const struct ref *remote_refs,
 	}
 
 	for (rm = ref_map; rm; rm = rm->next) {
-		if (rm->peer_ref && check_ref_format(rm->peer_ref->name + 5))
-			die("* refusing to create funny ref '%s' locally",
-			    rm->peer_ref->name);
+		if (rm->peer_ref) {
+			int st = check_ref_format(rm->peer_ref->name + 5);
+			if (st && st != CHECK_REF_FORMAT_ONELEVEL)
+				die("* refusing to create funny ref '%s'"
+				    " locally", rm->peer_ref->name);
+		}
 	}
 
 	if (ref_map)