On Thu, Feb 12, 2009 at 02:55:58PM +0530, Aneesh Kumar K.V wrote: > On Thu, Feb 12, 2009 at 09:48:11AM +0100, martin f krafft wrote: > > also sprach Aneesh Kumar <aneesh.kumar@xxxxxxxxx> [2009.02.12.0909 +0100]: > > > doing a tg update with latest git gives the below error > > > > > > [extent_validate@xxxxxxxxx]$ tg update > > > fatal: Refusing to point HEAD outside of refs/heads/ > > > [extent_validate@xxxxxxxxx]$ > > > > Which version? And could you please provide (a lot) more information > > about your repository or make it available? > > > > Latest git and topgit. Moving to git version v1.6.1.3 fixed the issue. > I can reproduce the problem on any test repo. Just do a tg update after > committing something in the dependent branch. This error message and safety valve are not in any released version of git yet. So by moving back to 1.6.1.3, you are just predating the addition of that message. :) I think I know what is going on. A safety valve was added in afe5d3d to disallow setting HEAD to anything that would violate git's "is this a git directory" detector: symbolic ref: refuse non-ref targets in HEAD When calling "git symbolic-ref" it is easy to forget that the target must be a fully qualified ref. E.g., you might accidentally do: $ git symbolic-ref HEAD master Unfortunately, this is very difficult to recover from, because the bogus contents of HEAD make git believe we are no longer in a git repository (as is_git_dir explicitly checks for "^refs/heads/" in the HEAD target). So immediately trying to fix the situation doesn't work: $ git symbolic-ref HEAD refs/heads/master fatal: Not a git repository and one is left editing the .git/HEAD file manually. Released versions of git just check "refs/" in HEAD. _But_ as part of this patch series, b229d18 also tightened the "refs/" check to "refs/heads/". So what I suspect is happening is that topgit is trying to set HEAD to "refs/top-bases/whatever". Aneesh, can you confirm by running your test with GIT_TRACE=1? I suspect you will see a call like "git symbolic-ref HEAD refs/top-bases/foo". Junio, I think we should probably revert b229d18 (and loosen symbolic-ref's check to just "refs/"). Even if you want to argue that topgit should be changed to handle this differently, we are still breaking existing topgit installations, and who knows what other scripts which might have relied on doing something like this. -Peff -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html