Re: [PATCH v2 0/7] Teach fetch/pull the on-demand mode and make it the default

[Marc Branchaud <marcnarc@xiplink.com>]

On 11-03-01 05:59 PM, Jens Lehmann wrote:
> So here is version 2 of the on-demand recursive fetch for submodules.
> 
> Changes since the last version:
> 
> *) Add a name parameter to parse_fetch_recurse_submodules_arg()
> *) Enhance description of the "--submodule-default" fetch option
> *) Free the changed_submodule_paths string_list at the end of
>    fetch_populated_submodules() to aid libification and to make
>    valgrind happy
> *) Add a comment for the while loop in check_for_new_submodule_commits()
> *) Use "--all" instead of "--branches" and "--remotes" for rev-list to
>    include tags too
> *) Compute the argc value properly instead of hard coding it in
>    check_for_new_submodule_commits()
> *) Change size_t casts to intptr_t where appropriate
> *) Use "rev-list --not --all" when checking if a submodule commit exists
>    instead of a simple lookup/rev-parse to make sure they are not only
>    present but are reachable from a ref
> *) Describe that fetch and pull only work on populated submodules for
>    now in the BUGS section of their man pages (done in a separate commit
>    because this doesn't belong to a special commit in this series but
>    also to the functionality that has already been merged)
> 
> Thanks to Jonathan an Junio for the feedback that lead to these changes.
> 
> Issues that were raised but not addressed yet:
> 
> *) Should this option be called "on-demand" or "changed" or maybe even
>    "updated"?
>    I have no strong feelings about this, me just thinks the name should
>    tell somehow that only submodules where new commits are recorded in
>    the superproject which aren't already present locally will be fetched.
>    Opinions?

First, thanks again for this work!

I have a slight preference for "updated" or "updated-only", but no really
strong feelings either way.

		M.