Fredrik Gustafsson [off-list ref] writes:
Regarding the discussion of superprojects with submodules that have no
remote tracking branches: A push will still be denied.
I have marked one part of the data synchronized across machines as
"private" submodule (which contains my gpg keychains, encrypted password
files, personaly memos, etc.) and push only the outer "shell" superproject
(which has tools that I use everywhere to go to $HOME/bin among other
things) to certain machines without the private parts, and the
superproject is designed to work without a checkout (nor clone) of a
submodule.
With this patch series, it sounds like I cannot use this repository
structure anymore, which is sad.
Am 28.07.2011 21:58, schrieb Junio C Hamano:
Fredrik Gustafsson [off-list ref] writes:
quoted
Regarding the discussion of superprojects with submodules that have no
remote tracking branches: A push will still be denied.
I have marked one part of the data synchronized across machines as
"private" submodule (which contains my gpg keychains, encrypted password
files, personaly memos, etc.) and push only the outer "shell" superproject
(which has tools that I use everywhere to go to $HOME/bin among other
things) to certain machines without the private parts, and the
superproject is designed to work without a checkout (nor clone) of a
submodule.
With this patch series, it sounds like I cannot use this repository
structure anymore, which is sad.
Thanks for bringing this use case up. Now I understand why you asked if
submodules without remote tracking branches should be checked too. We
discussed that and couldn't think of a scenario where the user doesn't
want to have remote tracking branches in a submodule, but we missed the
use case you described here. Back to the drawing board ...