Johannes Schindelin <Johannes.Schindelin@xxxxxx> writes:
> diff --git a/builtin/remote.c b/builtin/remote.c
> index eddd40c8f87..fead15adb97 100644
> --- a/builtin/remote.c
> +++ b/builtin/remote.c
> @@ -1344,7 +1344,7 @@ static int show(int argc, const char **argv)
>
> static int set_head(int argc, const char **argv)
> {
> - int i, opt_a = 0, opt_d = 0, result = 0;
> + int i, opt_a = 0, opt_d = 0, result = 0, is_mirror = 0;
> struct strbuf buf = STRBUF_INIT, buf2 = STRBUF_INIT;
> char *head_name = NULL;
>
> @@ -1357,8 +1357,16 @@ static int set_head(int argc, const char **argv)
> };
> argc = parse_options(argc, argv, NULL, options, builtin_remote_sethead_usage,
> 0);
> - if (argc)
> - strbuf_addf(&buf, "refs/remotes/%s/HEAD", argv[0]);
> + if (argc) {
> + struct remote *remote = remote_get(argv[0]);
> +
> + if (!remote || !remote->mirror)
> + strbuf_addf(&buf, "refs/remotes/%s/HEAD", argv[0]);
> + else {
> + is_mirror = 1;
> + strbuf_addstr(&buf, "HEAD");
> + }
> + }
Good.
Regardless of the "mirror" issue, it makes tons of sense to ask the
remote API how the remote-tracking refs for the given remote is set
up, instead of assuming that it must be "refs/remotes/<remote>"
blindly like in the original code.
That way, we could even handle a case like so:
[remote "frotz"]
fetch = +refs/heads/*:refs/remotes/nitfol/*
Their HEAD should be mapped to refs/remotes/nitfol/HEAD on our end,
so set-head should be able to notice that, too, if we go further
with your "do not assume, instead ask remote API" approach.
Thanks.
