[v2,0/1] rm: stage submodule removal from '.gitmodules'

Message

Shourya Shukla Feb. 22, 2021, 5:26 p.m. UTC

Hello all,

This is the v2 of the patch with the same title. After suggestions from
Phillipe and Junio, I have improved the commit messages, squashed the
two commits and did the following:

	1. Change the definition and declaration of
	   'remove_path_from_gitmodules()' to account in for the
	   'index_only' variable denoting the presence of '--cached'
	   option in the 'git rm' command. In case of the variable being
	   1, remove the submodule entry from the index copy of the
	   '.gitmodules' else do the same for the working tree copy.

	2. Remove the 'gitmodules_modified' variable and instead call
	   'stage_updated_gitmodules()' just after the
	   'remove_path_from_gitmodules()' call.

	3. Account for the above changes in 't3600' and make changes in
	   the same.

I am facing some problem with point (2) in the sense that what Junio
suggested in his mail:
https://lore.kernel.org/git/xmqqblchdoej.fsf@gitster.g/

-----8<-----
 - Since the previous point will maintain the correct contents in
   the index in all cases, get rid of gitmodules_modified and calls
   to stage_updated_gitmodules().  The call to write_locked_index()
   at the end will take care of the actual writing out of the index.
----->8-----

I am not able to get rid of the 'stage_updated_gitmodules()' call
without failing tests in t3600 (t3600.4 is the first one to fail). What
am I doing wrong here?

Comments and reviews are appreciated. Thank you Phillipe and Junio for
the constructive feedback on the v1!

Regards,
Shourya Shukla

Shourya Shukla (1):
  rm: stage submodule removal from '.gitmodules' when using '--cached'

 builtin/rm.c  | 42 +++++++++++++++++++++---------------------
 submodule.c   |  5 +++--
 submodule.h   |  2 +-
 t/t3600-rm.sh |  6 ++----
 4 files changed, 27 insertions(+), 28 deletions(-)