From: Daniel Barkalow Date: Mon, 17 Mar 2008 16:23:26 GMT Subject: Re: tracking repository Message-ID: In-Reply-To: <7v3aqpdc4n.fsf@gitster.siamese.dyndns.org> On Mon, 17 Mar 2008, Junio C Hamano wrote: > Daniel Barkalow writes: > > > On Sun, 16 Mar 2008, Junio C Hamano wrote: > > > >> Daniel Barkalow writes: > >> > >> > We don't currently have any concept of an invalid refspec; > >> > >> We don't? or just that parse_ref_spec() does not detect one? > >> > >> > ... we just have > >> > things that fall back to not being patterns and not being possible to > >> > match (due to one or the other side being invalid as a ref name). > >> > >> I am afraid that is an invitation for more bugs and confusions. > > > > Yeah, we're definitely too lenient. t3200-branch has been using the > > refspec "=" since July without anybody noticing that it's wrong. > > You mean these that came in 6f084a5 (branch --track: code cleanup and > saner handling of local branches, 2007-07-10), right? Yup. That actually makes me wonder if git-config should complain if it doesn't do anything due to the value_regex not matching anything. > Will you fix them while you come up with a patch to tighten the parsing? > Fixing these does seem to trigger problems in later parts of the test > sequence. Yeah, we just need to remove those options once that case finishes (or I suppose that test could go at the end). -Daniel *This .sig left intentionally blank*