threads / discuss / 10255

push fails with unexpected 'matches more than one'

Subject: push fails with unexpected 'matches more than one'

## tl;dr

4 messages between Oct 12, 2007 and Oct 13, 2007.

replies: 3people: 2as markdown or json

Steffen Prohaska· Oct 12, 2007, 06:59 UTC · lore

This adds a test case for unambigous local match but multiple remote matches. To me, it is unexpected that a ref that is perfectly defined on the local side fails with 'matches more than one'.

The following rule could solve this: A ref shall first be unambigously resolved on the local side, and its full name should be used for matching on the remote side. For example 'frotz' resolves locally to 'heads/refs/frotz'. Therefore pretend the user had typed 'heads/refs/frotz'.

But maybe there is some hidden secret about the current rules that I do not see.

Signed-off-by: Steffen Prohaska <prohaska@zib.de>
---
 t/t5516-fetch-push.sh |    8 ++++++++
 1 files changed, 8 insertions(+), 0 deletions(-)

diff --git a/t/t5516-fetch-push.sh b/t/t5516-fetch-push.sh
index ca46aaf..f249216 100755
--- a/t/t5516-fetch-push.sh
+++ b/t/t5516-fetch-push.sh
@@ -244,4 +244,12 @@ test_expect_success 'push with colon-less refspec (4)' '
 
 '
 
+test_expect_success 'push with colon-less refspec (locally unambigous)' '
+
+	mk_test heads/frotz heads/t/frotz &&
+	git branch -f frotz master &&
+	git push testrepo frotz
+
+'
+
 test_done

-- 
1.5.3.4.219.gd0b2

Steffen Prohaska· Oct 12, 2007, 12:06 UTC · re: Steffen Prohaska · lore

Re: push fails with unexpected 'matches more than one'

On Oct 12, 2007, at 8:59 AM, Steffen Prohaska wrote:

Show 12 quoted lines

> This adds a test case for unambigous local match but multiple remote
> matches. To me, it is unexpected that a ref that is perfectly defined
> on the local side fails with 'matches more than one'.
>
> The following rule could solve this:
> A ref shall first be unambigously resolved on the local side, and its
> full name should be used for matching on the remote side.
> For example 'frotz' resolves locally to 'heads/refs/frotz'.
> Therefore pretend the user had typed 'heads/refs/frotz'.
>
> But maybe there is some hidden secret about the current rules that
> I do not see.

Here is a related question:

I read carefully through the documentation of git-send-pack and git-rev-parse. The current implementation of git-send-pack is in line with the documented behaviour, as is the implementation of git-rev- parse.

So formally everything is correct.

But it is completely against my expectation that git-push <remote> <head> can successfully resolve a <head> that git-rev-parse fails to parse. I understand that refs are not revs ;). But nonetheless, I'd expect that a local ref that cannot be parsed by git-rev-parse should also fail to be pushed by git-send-pack. I didn't expect that git-send-pack would locate <head> as someprefix/<head>.

Why is my expectation wrong? Or is the current specification of git-send-pack's ref parsing wrong?

	Steffen

Shawn O. Pearce· Oct 13, 2007, 03:21 UTC · re: Steffen Prohaska · lore

Re: push fails with unexpected 'matches more than one'

Steffen Prohaska <prohaska@zib.de> wrote:

Show 17 quoted lines

> I read carefully through the documentation of git-send-pack and
> git-rev-parse. The current implementation of git-send-pack is in line
> with the documented behaviour, as is the implementation of git-rev- 
> parse.
> 
> So formally everything is correct.
> 
> But it is completely against my expectation that git-push <remote>  
> <head>
> can successfully resolve a <head> that git-rev-parse fails to parse. I
> understand that refs are not revs ;). But nonetheless, I'd expect that a
> local ref that cannot be parsed by git-rev-parse should also fail to be
> pushed by git-send-pack. I didn't expect that git-send-pack would locate
> <head> as someprefix/<head>.
> 
> Why is my expectation wrong?
> Or is the current specification of git-send-pack's ref parsing wrong?

I think its a bug.  But I didn't write the original code.

Meaning I think what happened here was someone wanted to enable git-send-pack to match "master" here with "refs/heads/master" on the remote side. One easy way to do that was to see if any ref ended with "/master", as that was what the ref here was called.

Way back when that code was written most Git repositories probably only ever had that one branch anyway, or maybe two (refs/heads/master and refs/heads/origin) so matching the trailing suffix never came up as a bug. Nobody ever had two refs that could possibly match.

Then the documentation got expanded to actually document the behavior that git-send-pack implemented. Unfortunately that codified the bug as documented behavior.

So I agree with you Steffen, this is a bug in send-pack, and I run up against it every once in a while. I've specifically told my coworkers "NEVER, EVER, EVER, create a branch called 'master' that isn't exactly refs/heads/master OR ELSE I WILL COME BEAT YOU WITH A CLUE STICK". They still create "refs/heads/experiments/master". *sigh*.

I think we should fix it. Anyone that is relying on "git push $url master" to resolve to "refs/heads/experimental/master" on the remote side is already playing with fire. But Junio is (rightfully so) very conservative and doesn't like to break a user's scripts. We may not be able to fix this until Git 1.6.

-- 
Shawn.

Steffen Prohaska· Oct 13, 2007, 16:51 UTC · re: Shawn O. Pearce · lore

Re: push fails with unexpected 'matches more than one'

On Oct 13, 2007, at 5:21 AM, Shawn O. Pearce wrote:

Show 52 quoted lines

> Steffen Prohaska <prohaska@zib.de> wrote:
>> I read carefully through the documentation of git-send-pack and
>> git-rev-parse. The current implementation of git-send-pack is in line
>> with the documented behaviour, as is the implementation of git-rev-
>> parse.
>>
>> So formally everything is correct.
>>
>> But it is completely against my expectation that git-push <remote>
>> <head>
>> can successfully resolve a <head> that git-rev-parse fails to  
>> parse. I
>> understand that refs are not revs ;). But nonetheless, I'd expect  
>> that a
>> local ref that cannot be parsed by git-rev-parse should also fail  
>> to be
>> pushed by git-send-pack. I didn't expect that git-send-pack would  
>> locate
>> <head> as someprefix/<head>.
>>
>> Why is my expectation wrong?
>> Or is the current specification of git-send-pack's ref parsing wrong?
>
> I think its a bug.  But I didn't write the original code.
>
> Meaning I think what happened here was someone wanted to enable
> git-send-pack to match "master" here with "refs/heads/master" on
> the remote side.  One easy way to do that was to see if any ref
> ended with "/master", as that was what the ref here was called.
>
> Way back when that code was written most Git repositories probably
> only ever had that one branch anyway, or maybe two (refs/heads/master
> and refs/heads/origin) so matching the trailing suffix never came
> up as a bug.  Nobody ever had two refs that could possibly match.
>
> Then the documentation got expanded to actually document the behavior
> that git-send-pack implemented.  Unfortunately that codified the
> bug as documented behavior.
>
>
> So I agree with you Steffen, this is a bug in send-pack, and I run
> up against it every once in a while.  I've specifically told my
> coworkers "NEVER, EVER, EVER, create a branch called 'master' that
> isn't exactly refs/heads/master OR ELSE I WILL COME BEAT YOU WITH A
> CLUE STICK".  They still create "refs/heads/experiments/master".
> *sigh*.
>
> I think we should fix it.  Anyone that is relying on "git push
> $url master" to resolve to "refs/heads/experimental/master" on the
> remote side is already playing with fire.  But Junio is (rightfully
> so) very conservative and doesn't like to break a user's scripts.
> We may not be able to fix this until Git 1.6.

I'm working on this and will send patches tomorrow.

	Steffen