On Tue, Feb 09, 2016 at 12:47:39AM +0200, Mikko Rapeli wrote: > On Mon, Feb 08, 2016 at 05:36:50PM -0500, Matt McCutchen wrote: > > On Mon, 2016-02-08 at 14:22 -0800, Junio C Hamano wrote: > > > Matt McCutchen <matt@xxxxxxxxxxxxxxxxx> writes: > > > > > > > I found no evidence of such behavior in the source code. > > > > > > > > Signed-off-by: Matt McCutchen <matt@xxxxxxxxxxxxxxxxx> > > > > --- > > > > > > That was added last year at bcd57cb9 (Documentation/git-clean.txt: > > > document that -f may need to be given twice, 2015-02-26). It would > > > be better to know what got changed since then--that is, was the > > > additional text unnecessary even back then, or we made changes to > > > the system since then and forgot to remove the added text. > > > > > > Mikko, is this need to give -f twice still the case? > > > > I know you probably want confirmation from Mikko, but I'll offer my > > understanding. There were two statements added in bcd57cb9: > > > > 1. -f may need to be given twice to delete nested worktrees and > > embedded repositories. This is still true. > > > > 2. Deletion of submodule repositories under .git/modules is conditional > > on -f being given twice. AFAICT, this was wrong even back then: "git > > clean" has never deleted such repositories under any conditions. > > This is the use case which I've used double -f at work with several build > jobs but with older 1:1.7.9.5-1ubuntu0.2 (Ubuntu 12.04) and 1:1.9.1-1ubuntu0.2 > (Ubuntu 14.04) versions of git. Sorry, can't reproduce the problem where submodules stayed in the tree until git clean was called with two -f's. You are right in removing the second part. > But I can confirm that git version 1:2.7.0~rc3-1 (Debian unstable) is no > longer removing the git submodule trees from .git/modules with double -f. > > At work, we really want to remove the .git/modules subtrees since we want to > test changes to .git/modules structure via normal commits to the git trees. > Thus we need a way removing all non-tracked files from the git tree > which includes obsolete (or for test only) git submodule trees. This usecase does not exists in git then. Using gerrit topics git branches to test changes to submodule structure does not work since there is no way of undoing these changes from the working tree. I should not have started using submodules in the first place, sigh. -Mikko -- 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