Re: [PATCH v10 3/5] fetch: infer branches to fetch from a refmap-only remote
- From
Harald Nordgren <haraldnordgren@gmail.com>
- Date
- Oct 25, 2026, 17:27 UTC
- Message-ID
- <CAHwyqnUQSninG=3hcxZOXUUwEXGbA6AT7FfmL36gwa9wV3ub4g@mail.gmail.com>
- In-Reply-To
- <CALnO6CCJzzX1qKeX4AR-GViM4LJjKrfdj6iafVXUQZM-c2Su4A@mail.gmail.com>
Show 11 quoted lines
> > Configuring remote.<name>.refmap without remote.<name>.fetch used to > > make a refspec-less "git fetch <name>" fail with "--refmap option is > > only meaningful with command-line refspec(s)", since a refmap only > > says where to put fetched refs, not what to fetch. > > This makes this commit sound like a fixup of [1/5], but I think it's > actually one of the main goals of the series! Can we flip the > justification around to say something like "we want to allow refmaps > to do XYZ […] Incidentally, we also make sense of the case when refmap > is configured without fetch" ? That would make the purpose clearer to > me, at least.
Good idea.
Show 49 quoted lines
> > Make that case infer what to fetch: the local branches whose
> > @{upstream} is already on that remote. This lets a remote be
> > configured to fetch only the branches actually in use, without
> > listing them by hand in remote.<name>.fetch, and without needing to
> > touch the command line every time.
> >
> > Signed-off-by: Harald Nordgren <haraldnordgren@gmail.com>
> > ---
> > Documentation/config/remote.adoc | 5 +-
> > Documentation/fetch-options.adoc | 6 +-
> > builtin/fetch.c | 75 +++++++++++++++----
> > remote.c | 24 +++++++
> > remote.h | 4 ++
> > t/meson.build | 1 +
> > t/t5586-fetch-refmap.sh | 119 +++++++++++++++++++++++++++++++
> > 7 files changed, 219 insertions(+), 15 deletions(-)
> > create mode 100755 t/t5586-fetch-refmap.sh
> >
> > diff --git a/Documentation/config/remote.adoc b/Documentation/config/remote.adoc
> > index 103eada406..7c52e443a5 100644
> > --- a/Documentation/config/remote.adoc
> > +++ b/Documentation/config/remote.adoc
> > @@ -36,7 +36,10 @@ remote.<name>.fetch::
> > remote.<name>.refmap::
> > The default value of the `--refmap` option for linkgit:git-fetch[1].
> > Used to map remote refs being fetched to remote-tracking refs to
> > - store. See the `--refmap` entry in linkgit:git-fetch[1].
> > + store. If `remote.<name>.fetch` is not set either, a refspec-less
> > + fetch infers what to fetch from local branches built on this
> > + remote, instead of fetching every branch it has. See the
> > + `--refmap` entry in linkgit:git-fetch[1].
> >
> > remote.<name>.push::
> > The default set of "refspec" for linkgit:git-push[1]. See
> > diff --git a/Documentation/fetch-options.adoc b/Documentation/fetch-options.adoc
> > index c2101a7b39..90b9e13ac2 100644
> > --- a/Documentation/fetch-options.adoc
> > +++ b/Documentation/fetch-options.adoc
> > @@ -245,8 +245,10 @@ endif::git-pull[]
> > command-line arguments. See section on "Configured Remote-tracking
> > Branches" for details.
> > +
> > -`remote.<name>.refmap` provides the default value for this option, the
> > -same way `remote.<name>.fetch` provides the default refspecs to fetch.
> > +When a refmap is active (from `--refmap` or `remote.<name>.refmap`) but
> > +nothing to fetch is specified on the command line, nor is there a
> > +`remote.<name>.fetch`, branches from the remote that are used as the
>
> "remote-tracking branches" ??Yes, the next version will say "remote-tracking branches from this remote that are also configured as the @{upstream} of a local branch are fetched".
Show 21 quoted lines
> > +`@{upstream}` of our local branches are fetched.
>
> The clause here is complicated ':D I found "infer what to fetch from
> local branches built on this remote" to be a bit clearer, but we
> definitely still need to get to the point, which is that
> remote-tracking branches from this remote that are also configured as
> upstreams of local branches are fetched.
>
> Now, I wonder a bit: the doc says "when a refmap is active" we do that
> "fetch remote-tracking branches that are upstream of local branches"
> [all scoped to the current remote]---but it doesn't really say how the
> refmap affects that? I mean, I imagine it's used for mapping (duh),
> but I don't know if it also narrows the scope, or if the scope is
> deliberately all remote-tracking branches that meet the requirements,
> or…?
>
> IOW, it seems to beg the question "why does the presence of a refmap
> control this if it has no effect on the set of selected branches," and
> I think the answer is partly "because we designed it that way" and
> possibly found in the mails leading up to this point, but we might
> want to try to connect the dots for folks? Idk.I'll take a look at improving this part.
Show 40 quoted lines
> >
> > `-t`::
> > `--tags`::
> > diff --git a/builtin/fetch.c b/builtin/fetch.c
> > index 136a59154a..0fc1fb66d5 100644
> > --- a/builtin/fetch.c
> > +++ b/builtin/fetch.c
> > @@ -524,6 +524,8 @@ static struct ref *get_ref_map(struct remote *remote,
> > */
> > struct refspec *effective_refmap =
> > refmap.nr ? &refmap : remote ? &remote->refmap : NULL;
> > + struct refspec inferred_rs;
> > + int infer_from_refmap = 0;
> >
> > /* opportunistically-updated references: */
> > struct ref *orefs = NULL, **oref_tail = &orefs;
> > @@ -531,15 +533,32 @@ static struct ref *get_ref_map(struct remote *remote,
> > struct hashmap existing_refs;
> > int existing_refs_populated = 0;
> >
> > + refspec_init_fetch(&inferred_rs, the_hash_algo);
> > +
> > filter_prefetch_refspec(rs);
> > if (remote)
> > filter_prefetch_refspec(&remote->fetch);
> >
> > + if (!rs->nr && remote && !remote->fetch.nr &&
> > + effective_refmap && effective_refmap->nr) {
> > + struct string_list tracked = STRING_LIST_INIT_DUP;
> > + struct string_list_item *item;
> > +
> > + branches_tracking_remote(the_repository, remote, &tracked);
> > + for_each_string_list_item(item, &tracked)
> > + refspec_append(&inferred_rs, item->string);
> > + string_list_clear(&tracked, 0);
>
> Interesting choice of string_list again ;) do we need the sorting
> behavior? (In fact, since it could come from an "ordered" config,
> perhaps we should use that order?) A strvec seems more appropriate
> then…Agreed, the next version uses a strvec here too. Nothing is sorted anymore: the branches come in refname order, and each branch's merge entries in config order.
Show 54 quoted lines
> > +
> > + rs = &inferred_rs;
> > + infer_from_refmap = 1;
> > + }
> > +
> > if (rs->nr) {
> > struct refspec *fetch_refspec;
> >
> > for (i = 0; i < rs->nr; i++) {
> > - get_fetch_map(remote_refs, &rs->items[i], &tail, 0);
> > + get_fetch_map(remote_refs, &rs->items[i], &tail,
> > + infer_from_refmap);
> > if (rs->items[i].dst && rs->items[i].dst[0])
> > *autotags = 1;
> > }
> > @@ -574,6 +593,8 @@ static struct ref *get_ref_map(struct remote *remote,
> >
> > for (i = 0; i < fetch_refspec->nr; i++)
> > get_fetch_map(ref_map, &fetch_refspec->items[i], &oref_tail, 1);
> > + } else if (infer_from_refmap) {
> > + /* Already fully handled above. */
> > } else if (refmap.nr) {
> > die("--refmap option is only meaningful with command-line refspec(s)");
> > } else {
> > @@ -669,6 +690,7 @@ static struct ref *get_ref_map(struct remote *remote,
> > if (existing_refs_populated)
> > hashmap_clear_and_free(&existing_refs, struct refname_hash_entry, ent);
> >
> > + refspec_clear(&inferred_rs);
> > return ref_map;
> > }
> >
> > @@ -1978,13 +2000,47 @@ static int do_fetch(struct transport *transport,
> >
> > if (rs->nr) {
> > refspec_ref_prefixes(rs, &transport_ls_refs_options.ref_prefixes);
> > + } else if (transport->remote->fetch.nr) {
> > + struct string_list tracked = STRING_LIST_INIT_DUP;
> > + struct string_list_item *item;
> > +
> > + refspec_ref_prefixes(&transport->remote->fetch,
> > + &transport_ls_refs_options.ref_prefixes);
> > + /*
> > + * See remote.c's handling of remote.<name>.followRemoteHEAD
> > + * for the analogous, still-unresolved case.
>
> Is this a "we haven't resolved what to do" case, or a "some remote (or
> something) hasn't resolved" case? I'm not sure what I should be
> looking at here.
>
> Oh, I see. It's a duplicate of comment and code from the
> else{effective_refmap->nr} block below. Hrm… probably it would be nice
> not to duplicate that? I can't spot any differences between the code
> blocks.Yes, good idea!
Harald