Other than Brandon's issue on 3/5 (which I noticed myself but wanted to double-check that my mailer wasn't playing tricks on me and so lost the race to report), this version looks good. > -----Original Message----- > From: Stefan Beller [mailto:sbeller@xxxxxxxxxx] > Sent: Monday, December 19, 2016 6:28 PM > To: gitster@xxxxxxxxx > Cc: git@xxxxxxxxxxxxxxx; bmwill@xxxxxxxxxx; sandals@xxxxxxxxxxxxxxxxxxxx; > David Turner; Stefan Beller > Subject: [PATCHv4 0/5] git-rm absorbs submodule git directory before > deletion > > v4: > * reworded commit messages of the last 2 patches > * introduced a new patch introducing {run,start,finish}_command_or_die > * found an existing function in dir.h to use to remove a directory > which deals gracefully with the cornercases, that Junio pointed out. > > v3: > * removed the patch to enhance ok_to_remove_submodule to absorb the > submodule > if needed > * Removed all the error reporting from git-rm that was related to > submodule > git directories not absorbed. > * instead just absorb the git repositories or let the absorb function die > with an appropriate error message. > > v2: > * new base where to apply the patch: > sb/submodule-embed-gitdir merged with sb/t3600-cleanup. > I got merge conflicts and resolved them this way: > #@@@ -709,9 -687,10 +687,9 @@@ test_expect_success 'checking out a com > # git commit -m "submodule removal" submod && > # git checkout HEAD^ && > # git submodule update && > #- git checkout -q HEAD^ 2>actual && > #+ git checkout -q HEAD^ && > # git checkout -q master 2>actual && > # - echo "warning: unable to rmdir submod: Directory not empty" > >expected && > # - test_i18ncmp expected actual && > # + test_i18ngrep "^warning: unable to rmdir submod:" actual && > # git status -s submod >actual && > # echo "?? submod/" >expected && > # test_cmp expected actual && > # > > * improved commit message in "ok_to_remove_submodule: absorb the submodule > git dir" > (David Turner offered me some advice on how to write better English off > list) > * simplified code in last patch: > -> dropped wrong comment for fallthrough > -> moved redundant code out of both bodies of an if-clause. > * Fixed last patchs commit message to have "or_die" instead of or_dir. > > v1: > The "checkout --recurse-submodules" series got too large to comfortably > send it out for review, so I had to break it up into smaller series'; this > is the first subseries, but it makes sense on its own. > > This series teaches git-rm to absorb the git directory of a submodule > instead of failing and complaining about the git directory preventing > deletion. > > It applies on origin/sb/submodule-embed-gitdir. > > Any feedback welcome! > > Thanks, > Stefan > > Stefan Beller (5): > submodule.h: add extern keyword to functions > submodule: modernize ok_to_remove_submodule to use argv_array > run-command: add {run,start,finish}_command_or_die > submodule: add flags to ok_to_remove_submodule > rm: absorb a submodules git dir before deletion > > builtin/rm.c | 82 +++++++++++++++--------------------------------------- > ----- > run-command.c | 28 ++++++++++++++++++++ run-command.h | 4 +++ > submodule.c | 27 ++++++++++---------- > submodule.h | 58 ++++++++++++++++++++++++------------------ > t/t3600-rm.sh | 39 +++++++++++----------------- > 6 files changed, 114 insertions(+), 124 deletions(-) > > -- > 2.11.0.rc2.53.gb7b3fba.dirty