Re: [PATCH 0/2] improve relocate_git_dir to fail into a sane state.

Stefan Beller [off-list ref] writes:

This goes on top of sb/submodule-embed-gitdir.
When the absorbing of a git dir fails, try to recover into a sane state,
i.e. try to undo the move of the git dir.

Are these unconditionally good improvements?  I ask because the
series is still not in 'next' and it is the last chance to fold
these into existing patches if we wanted to.

Thanks,
Stefan

Stefan Beller (2):
  dir.c: split up connect_work_tree_and_git_dir
  dir.c: add retry logic to relocate_gitdir

 dir.c       | 118 +++++++++++++++++++++++++++++++++++++++++++++++++++++-------
 dir.h       |   6 ++--
 submodule.c |   3 +-
 3 files changed, 110 insertions(+), 17 deletions(-)

On Mon, Dec 19, 2016 at 2:54 PM, Junio C Hamano [off-list ref] wrote:

Stefan Beller [off-list ref] writes:

quoted

This goes on top of sb/submodule-embed-gitdir.
When the absorbing of a git dir fails, try to recover into a sane state,
i.e. try to undo the move of the git dir.

Are these unconditionally good improvements?  I ask because the
series is still not in 'next' and it is the last chance to fold
these into existing patches if we wanted to.

Actually I forgot to mark these as RFC-ish as it is a response to
https://public-inbox.org/git/20161219053507.GA2335@duynguyen.vn.dektech.internal/
(which was the only review comment this round)

So I'd say these patches are rather experimental in my mind
unlike the absorb series, which I assume is ok as is already.

Thanks,
Stefan