From: Junio C Hamano <gitster@pobox.com>
To: Jonathan Tan <jonathantanmy@google.com>
Cc: git@vger.kernel.org
Subject: Re: [PATCH] remote: prefer exact matches when using refspecs
Date: Tue, 31 Jul 2018 14:53:41 -0700 [thread overview]
Message-ID: <xmqq600v9imy.fsf@gitster-ct.c.googlers.com> (raw)
In-Reply-To: <20180731211832.142014-1-jonathantanmy@google.com> (Jonathan Tan's message of "Tue, 31 Jul 2018 14:18:32 -0700")
Jonathan Tan <jonathantanmy@google.com> writes:
> When matching a non-wildcard LHS of a refspec against a list of refs,
> find_ref_by_name_abbrev() returns the first ref that matches using the
> DWIM rules used by refname_match() in refs.c, even if an exact match
> occurs later in the list of refs.
When you have refs/heads/refs/heads/s and refs/heads/s, and if you
ask for refs/heads/s, you want that exact match (i.e. the latter) to
take precedence over DWIMmed refs/heads/refs/heads/s.
What is unfortunate is that ref_rev_parse_rules[] array already
expresses that preference by listing the "fullname" choice "%.*s"
before other DWIM choices like "refs/heads/%.*s".
Now we iterate over refs we say from ls-remote output, and with the
updated one, the logic _manually_ gives the precedence to the first
entry in ref_rev_parse_rules[], so in that "I have a branch s and
also another branch refs/heads/s" case, that may happen to work, but
would the updated code do the right thing when you have entries that
can match, say, both second and third entry in the rules[] and
tiebreak correctly the same way? Say you ask for "tags/T" when
there are "refs/tags/T" and "refs/heads/tags/T" at the same time in
the refs linked list. None of the ref on refs list trigger
!strcmp() as there is no exact mqatch, and refname_match() would say
"Yeah I see a match" when checking "refs/heads/tags/T" and say it is
the best match. Then it finds "refs/tags/T" also on the refs list
and finds it also matches user-supplied "tags/T".
In order to resolve this correctly with the precedence rules, I
think you need to make refname_match() return the precedence number
(e.g. give 1 to "%.*s", 2 to "refs/%.*s", etc., using the index in
ref_rev_parse_rules[] array), and make this loop keep track of the
"best" match paying attention to the returned precedence.
> This causes unexpected behavior when (for example) fetching using the
> refspec "refs/heads/s:<something>" from a remote with both
> "refs/heads/refs/heads/s" and "refs/heads/s". (Even if the former was
> inadvertently created, one would still expect the latter to be fetched.)
>
> This problem has only been observed when the desired ref comes after the
> undesired ref in alphabetical order. However, for completeness, the test
> in this patch also checks what happens when the desired ref comes first
> alphabetically.
>
> Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
> ---
> remote.c | 7 +++++--
> t/t5510-fetch.sh | 28 ++++++++++++++++++++++++++++
> 2 files changed, 33 insertions(+), 2 deletions(-)
>
> diff --git a/remote.c b/remote.c
> index 3fd43453f..eeffe3488 100644
> --- a/remote.c
> +++ b/remote.c
> @@ -1687,12 +1687,15 @@ static struct ref *get_expanded_map(const struct ref *remote_refs,
>
> static const struct ref *find_ref_by_name_abbrev(const struct ref *refs, const char *name)
> {
> + const struct ref *best_match = NULL;
> const struct ref *ref;
> for (ref = refs; ref; ref = ref->next) {
> - if (refname_match(name, ref->name))
> + if (!strcmp(name, ref->name))
> return ref;
> + if (refname_match(name, ref->name))
> + best_match = ref;
> }
> - return NULL;
> + return best_match;
> }
>
> struct ref *get_remote_ref(const struct ref *remote_refs, const char *name)
> diff --git a/t/t5510-fetch.sh b/t/t5510-fetch.sh
> index e402aee6a..da88f35f0 100755
> --- a/t/t5510-fetch.sh
> +++ b/t/t5510-fetch.sh
> @@ -535,6 +535,34 @@ test_expect_success "should be able to fetch with duplicate refspecs" '
> )
> '
>
> +test_expect_success 'LHS of refspec prefers exact matches' '
> + mkdir lhs-exact &&
> + (
> + cd lhs-exact &&
> + git init server &&
> + test_commit -C server unwanted &&
> + test_commit -C server wanted &&
> +
> + git init client &&
> +
> + # Check a name coming after "refs" alphabetically ...
> + git -C server update-ref refs/heads/s wanted &&
> + git -C server update-ref refs/heads/refs/heads/s unwanted &&
> + git -C client fetch ../server refs/heads/s:refs/heads/checkthis &&
> + git -C server rev-parse wanted >expect &&
> + git -C client rev-parse checkthis >actual &&
> + test_cmp expect actual &&
> +
> + # ... and one before.
> + git -C server update-ref refs/heads/q wanted &&
> + git -C server update-ref refs/heads/refs/heads/q unwanted &&
> + git -C client fetch ../server refs/heads/q:refs/heads/checkthis &&
> + git -C server rev-parse wanted >expect &&
> + git -C client rev-parse checkthis >actual &&
> + test_cmp expect actual
> + )
> +'
> +
> # configured prune tests
>
> set_config_tristate () {
next prev parent reply other threads:[~2018-07-31 21:53 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-07-31 21:18 [PATCH] remote: prefer exact matches when using refspecs Jonathan Tan
2018-07-31 21:31 ` Jonathan Nieder
2018-07-31 21:53 ` Junio C Hamano [this message]
2018-07-31 22:28 ` Junio C Hamano
2018-07-31 23:33 ` Jonathan Tan
2018-08-01 0:32 ` Junio C Hamano
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=xmqq600v9imy.fsf@gitster-ct.c.googlers.com \
--to=gitster@pobox.com \
--cc=git@vger.kernel.org \
--cc=jonathantanmy@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).