On 04/11, Junio C Hamano wrote: > Brandon Williams <bmw...@google.com> writes: > > > v3 builds upon v2 by adding 3 additional patches to add functionality to > > also > > propagate the remote and refspec down to children process's working to push > > submodules. The remote and refspec will only be propagated if the provided > > remote is configured. A remote provided in the form of a URL will cause the > > remote and refspec to not be propagated down to the submodules and instead > > the > > default remote and refspec will be used, preserving existing functionality. > > Currently the only supported LHS of a refspec must exactly match a single > > ref > > in the submodule (e.g. a branch name). > > > > Patches [1/5] and [2/5] remain unchanged from v2. > > 2/5 seems to be a bit different in 5545 ll.156,+6, though.
Oh, oops you're right. Looks like I forgot that I added a single line to a test in 2/5: git -C parent/workbench remote add up ../../upstream && Which just configures a remote in one of the submodules. I must have added that to prep for the propagating of the remote/refspec in later patches. > > > > Brandon Williams (5): > > push: unmark a local variable as static > > push: propagate push-options with --recurse-submodules > > remote: expose parse_push_refspec function > > submodule--helper: add push-check subcommand > > push: propagate remote and refspec with --recurse-submodules > > > > builtin/push.c | 5 +-- > > builtin/submodule--helper.c | 45 +++++++++++++++++++++++++ > > remote.c | 2 +- > > remote.h | 1 + > > submodule.c | 74 > > +++++++++++++++++++++++++++++++++++++++--- > > submodule.h | 5 ++- > > t/t5531-deep-submodule-push.sh | 52 +++++++++++++++++++++++++++++ > > t/t5545-push-options.sh | 40 +++++++++++++++++++++++ > > transport.c | 4 ++- > > 9 files changed, 219 insertions(+), 9 deletions(-) -- Brandon Williams
