From: Patrick Steinhardt Date: Tue, 13 Oct 2026 07:31:30 GMT Subject: Re: [PATCH v7 2/2] rerere: add "gc --skip-locked" for auto maintenance Message-ID: In-Reply-To: <000288119671bad19e5e9e37419766b364daaacc.1791627204.git.gitgitgadget@gmail.com> On Sat, Oct 10, 2026 at 10:13:24AM +0000, Thomas Bachem via GitGitGadget wrote: > diff --git a/rerere.h b/rerere.h > index feeb0e2c9f..61e986475b 100644 > --- a/rerere.h > +++ b/rerere.h > @@ -10,6 +10,8 @@ struct repository; > #define RERERE_AUTOUPDATE 01 > #define RERERE_NOAUTOUPDATE 02 > #define RERERE_READONLY 04 > +/* If MERGE_RR.lock is taken, return -1 as if rerere were disabled */ > +#define RERERE_NOWAIT 010 > > /* > * Marks paths that have been hand-resolved and added to the > @@ -37,7 +39,14 @@ const char *rerere_path(struct strbuf *buf, const struct rerere_id *, > int rerere_forget(struct repository *, struct pathspec *); > int rerere_remaining(struct repository *, struct string_list *); > void rerere_clear(struct repository *, struct string_list *); > -void rerere_gc(struct repository *, struct string_list *); > + > +enum rerere_gc_flags { > + /* Skip the operation in case the MERGE_RR.lock is already taken. */ > + RERERE_GC_NOWAIT = (1 << 0), > +}; Hm. I was hoping that we could avoid introducing both RERERE_NOWAIT and RERERE_GC_NOWAIT. But seeing the result I'm way less sure about my earlier suggestion to introduce the gc-specific flags, as translating the flags is a bit awkward. I think that your version that only has the RERERE_NOWAIT flag was better, sorry. Patrick