From: Christian Couder Date: Thu, 02 Apr 2026 06:55:09 GMT Subject: Re: [PATCH 11/16] promisor-remote: refactor should_accept_remote() control flow Message-ID: In-Reply-To: On Thu, Mar 26, 2026 at 1:21 PM Patrick Steinhardt wrote: > > On Mon, Mar 23, 2026 at 09:05:14AM +0100, Christian Couder wrote: > > diff --git a/promisor-remote.c b/promisor-remote.c > > index 8f2c1280c3..c2f0eb7223 100644 > > --- a/promisor-remote.c > > +++ b/promisor-remote.c > > @@ -665,6 +665,12 @@ static int should_accept_remote(enum accept_promisor accept, > > const char *remote_name = advertised->name; > > const char *remote_url = advertised->url; > > > > + if (!remote_url || !*remote_url) { > > + warning(_("no or empty URL advertised for remote '%s', " > > + "ignoring this remote"), remote_name); > > + return 0; > > + } > > + > > if (accept == ACCEPT_ALL) > > return all_fields_match(advertised, config_info, NULL); > > You mention that it shouldn't change behaviour in well-defined cases > where the remote sends non-empty URLs. But does it change behaviour in > ill-defined cases where the remote sends empty ones? Yes, it could change the behavior in the ill-defined case where the remote sends empty URLs. So I have added a new patch to catch advertised empty URLs and empty remote names at parsing time. It is patch 4/10 ("promisor-remote: reject empty name or URL in advertised remote") in the preparatory series I will send very soon now. > In other words, does this fix a bug that can be hit in the real world? No, because I think Git has the only implementation of the "promisor-remote" capability, and a Git server doesn't advertise empty URLs. (promisor_config_info_list() ignores empty URLs.) So the ill-defined case could only happen if people use their own fork of either Git (with a hacked the "promisor-remote" capability) or an implementation of Git (that implements the "promisor-remote" capability differently). So not likely in practice. Thanks.