Re: [PATCH v2 3/3] sequencer.c: don't die on invalid cleanup_arg

[Junio C Hamano <gitster@pobox.com>]

Denton Liu <liu.denton@gmail.com> writes:

> When get_cleanup_mode was provided with an invalid cleanup_arg, it used
> to die. Warn user and fallback to default behaviour if an invalid
> cleanup_arg is given.
>
> Signed-off-by: Denton Liu <liu.denton@gmail.com>
> ---
>  sequencer.c | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/sequencer.c b/sequencer.c
> index 5c04bae7ac..f9bdfa90ad 100644
> --- a/sequencer.c
> +++ b/sequencer.c
> @@ -502,8 +502,11 @@ enum commit_msg_cleanup_mode get_cleanup_mode(const char *cleanup_arg,
>  	else if (!strcmp(cleanup_arg, "scissors"))
>  		return use_editor ? COMMIT_MSG_CLEANUP_SCISSORS :
>  				    COMMIT_MSG_CLEANUP_SPACE;
> -	else
> -		die(_("Invalid cleanup mode %s"), cleanup_arg);
> +	else {
> +		warning(_("Invalid cleanup mode %s, falling back to default"), cleanup_arg);
> +		return use_editor ? COMMIT_MSG_CLEANUP_ALL :
> +				    COMMIT_MSG_CLEANUP_SPACE;
> +	}
>  }

In what different contexts does this get called, I wonder?

I think

	$ git cherry-pick --cleanup=bogus ...

should error out, instead of falling back to anything else, while
having

	[commit] cleanup = bogus

in .git/config should *not* say anything if you are not running a
command that does not get affected by the commit.cleanup variable,
and with such a bogus configuration, a command that does use the
variable should either also die, or fallback with a warning.

I have a suspicion that the change is breaking the error detection
done for the command line argument parsing, and if that is the case,
then it is a bad idea.  But I may have misread the code.

Thanks.