Re: [PATCH v2 02/12] rerere: convert flags into an enum
Patrick Steinhardt <ps@pks.im> writes:
Show 35 quoted lines
> On Fri, Nov 06, 2026 at 05:51:13AM -0500, Karthik Nayak wrote:
>> Patrick Steinhardt <ps@pks.im> writes:
>>
>> > diff --git a/rerere.h b/rerere.h
>> > index dd9da03d6db..f13790c5dcd 100644
>> > --- a/rerere.h
>> > +++ b/rerere.h
>> > @@ -39,10 +33,27 @@ struct rerere_id {
>> > int variant;
>> > };
>> >
>> > +enum rerere_flags {
>> > + /*
>> > + * Auto-update the index with previously recorded resolutions. This can
>> > + * be used to explicitly override "rerere.autoUpdate".
>> > + */
>> > + RERERE_AUTOUPDATE = (1 << 0),
>> > +
>> > + /* Reverse of RERERE_AUTOUPDATE. */
>> > + RERERE_NOAUTOUPDATE = (1 << 1),
>> > +
>>
>> Nit: From my understanding, RERERE_NOAUTOUPDATE outputs the conflict
>> resolution information to the stderr instead of updating the index.
>> Perhaps that's something to document?
>
> Well, the fact that it prints to stderr is not specific to NOAUTOUPDATE.
> We print messages in both cases, just at different times:
>
> - When `RERERE_AUTOUPDATE` is set we print the messages in
> `update_paths()`.
>
> - When `RERERE_NOAUTOUPDATE` is set (or `RERERE_AUTOUPDATE` is simply
> unset) then we print the message in `do_rerere_one_path()`.
>I didn't dig enough to find the former.
Show 7 quoted lines
> The messages are different of course, but I don't really think that this
> matters for the flag here. What we might want to say though is that we
> don't update the index, but the working tree does get updated.
>
> Will adapt, thanks.
>
> Patrick