Hi
On Sat, Jan 4, 2020 at 4:45 AM Jonathan Nieder <jrnieder@xxxxxxxxx> wrote:
>
> Hi,
>
> Marc-André Lureau wrote:
>
> > One of my most frequent mistake is to commit undesired submodules
> > changes when doing "commit -a", and I have seen a number of people doing
> > the same mistake in various projects. I wish there would be a config to
> > change this default behaviour.
>
> Can you say more about the overall workflow this is part of? What
> causes the submodules to change state in the first place here?
The most common case is, I guess, when you work on different branches
that have different (compatible) versions of the submodules. It is
easy to go unnoticed then, although I am usually quite careful what I
include in my commit, and will usually add changes interactively with
add -i instead.
I often rely on git commit -a during an interactive rebase. I check
the project at various points in history, find a small fix, and git
commit -a. At this point, I may have included a submodule change
inadvertently that may happen later in the series for example.
>
> [...]
> > --- a/builtin/commit.c
> > +++ b/builtin/commit.c
> [...]
> > @@ -1475,6 +1478,11 @@ static int git_commit_config(const char *k, const char *v, void *cb)
> > return 0;
> > }
> >
> > + if (!strcmp(k, "commit.all-ignore-submodules")) {
> > + commit_all_ignore_submodules = git_config_bool(k, v);
> > + return 0;
> > + }
>
> nit, less important than the comment above: no other config items use
> this naming scheme. We'd have to come up with a different name if we
> want to pursue this.
Sure, I am open to suggestions.
>
> If I want to disable this setting for a particular "git commit"
> invocation, how do I do that? Typically when adding new settings, we
> add them first as command-line options and then as a separate followup
> can introduce configuration to change the defaults.
--all=no-ignore ?
>
> To summarize: I'm interested in hearing more about the overall
> workflow so we can make the standard behavior without any special
> configuration work better for it, too.
>
> Thanks and hope that helps,
> Jonathan
>
thanks for the feeback
