Re: Another look?

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

Harald Nordgren <haraldnordgren@gmail.com> writes:

>> Please don't.  Unless my assumption, which is that in the old code
>> "!sti" and "!ours && !theirs" is equivalent, is wrong, all you need
>> to do around that part is to first check "if (!ours && !theirs)" and
>> say "your branch is up to date with...", and then have the check
>> "else if (abf == ABQ)" next.  That way, when we check abf we know
>> the branches are different.
>
> It seems to be an incorrect assumption. This code change breaks several
> tests including old ones:



>
> ```
> diff --git a/remote.c b/remote.c
> index 1f87b85b22..8db4fcd7b5 100644
> --- a/remote.c
> +++ b/remote.c
> @@ -2303,7 +2303,7 @@ static void format_branch_comparison(struct strbuf *sb,
>  				     enum branch_mode_flags advice_flags,
>  				     int show_divergence_advice)
>  {
> -	if (!sti) {
> +	if (!ours && !theirs) {
>  		strbuf_addf(sb,
>  			_("Your branch is up to date with '%s'.\n"),
>  			branch_name);
> ```
>
>
> Harald

That is unexpected.

Looking at what stat_branch_pair() does, before returning 0, the
function always clears *num_theirs and *num_ours, so there is
something else going on.

If your caller is *not* initializing ours and theirs, and if it is
not detecting an error return from stat_tracking_info, and the test
code is trying to see when stat_branch_info() signals failure by
returning -1, then I can understand why the above change makes a
difference (i.e., your code above with or without sti -> ours/theirs
change is broken), but then, that should be handled at the caller of
stat_tracking_info() by noticing its return value being negative, I
would have to say.