threads / discuss / 30220

non-fast-forward advice breaks tests on OS X

Subject: non-fast-forward advice breaks tests on OS X

## tl;dr

6 messages between Apr 12, 2012 and Apr 12, 2012.

replies: 5people: 3as markdown or json

Brian Gernhardt· Apr 12, 2012, 05:49 UTC · lore

After a break in regularly compiling git on my Mac 10.7.3 laptop, I had a few tests fail:  t5540.15, t5541.{12,15}.  All of these tests use test_i18ngrep to test the output of git-push for the string "To prevent you from losing history, non-fast-forward updates were rejected".

Bisecting traced it back to f25950f: "push: Provide situational hints for non-fast-forward errors".

I'm sending this off to the list in case someone else has some idea of the root cause.  I'll get back to this problem myself sometime tomorrow.

~~ Brian Gernhardt

Christopher Tiwald· Apr 12, 2012, 13:37 UTC · re: Brian Gernhardt · lore

Re: non-fast-forward advice breaks tests on OS X

On Thu, Apr 12, 2012 at 01:49:14AM -0400, Brian Gernhardt wrote:

> After a break in regularly compiling git on my Mac 10.7.3 laptop, I had a few tests fail:  t5540.15, t5541.{12,15}.  All of these tests use test_i18ngrep to test the output of git-push for the string "To prevent you from losing history, non-fast-forward updates were rejected".
> 
> Bisecting traced it back to f25950f: "push: Provide situational hints for non-fast-forward errors".

I also run 10.7.3 on a MacBook Pro. I will investigate this now and see if I can get tests to break. I know the original patch passed the test suite on this machine.

-- Christopher Tiwald

Christopher Tiwald· Apr 12, 2012, 14:00 UTC · re: Christopher Tiwald · lore

Re: non-fast-forward advice breaks tests on OS X

On Thu, Apr 12, 2012 at 09:37:01AM -0400, Christopher Tiwald wrote:

Show 8 quoted lines

> On Thu, Apr 12, 2012 at 01:49:14AM -0400, Brian Gernhardt wrote:
> > After a break in regularly compiling git on my Mac 10.7.3 laptop, I had a few tests fail:  t5540.15, t5541.{12,15}.  All of these tests use test_i18ngrep to test the output of git-push for the string "To prevent you from losing history, non-fast-forward updates were rejected".
> > 
> > Bisecting traced it back to f25950f: "push: Provide situational hints for non-fast-forward errors".
> 
> I also run 10.7.3 on a MacBook Pro. I will investigate this now and see
> if I can get tests to break. I know the original patch passed the test
> suite on this machine.

Ah ha, I wasn't running the proper tests. I'll send a patch to the affected ones later today.

*** t5540-http-push.sh *** # passed all 0 test(s) 1..0 # SKIP Network testing disabled (define GIT_TEST_HTTPD to enable) *** t5541-http-push.sh *** # passed all 0 test(s) 1..0 # SKIP Network testing disabled (define GIT_TEST_HTTPD to enable) *** t5550-http-fetch.sh *** # passed all 0 test(s) 1..0 # SKIP Network testing disabled (define GIT_TEST_HTTPD to enable) *** t5551-http-fetch.sh *** # passed all 0 test(s) 1..0 # SKIP Network testing disabled (define GIT_TEST_HTTPD to enable)

Junio C Hamano· Apr 12, 2012, 15:35 UTC · re: Christopher Tiwald · lore

Re: non-fast-forward advice breaks tests on OS X

Christopher Tiwald <christiwald@gmail.com> writes:

Show 6 quoted lines

>> I also run 10.7.3 on a MacBook Pro. I will investigate this now and see
>> if I can get tests to break. I know the original patch passed the test
>> suite on this machine.
>
> Ah ha, I wasn't running the proper tests. I'll send a patch to the
> affected ones later today.

Please do so. I assume that the test vectors were expecting specific error/advice messages that need to be updated?

Christopher Tiwald· Apr 12, 2012, 17:56 UTC · re: Junio C Hamano · lore

[PATCH] Fix httpd tests that broke when non-ff push advice changed

On Thu, Apr 12, 2012 at 08:35:11AM -0700, Junio C Hamano wrote:

> Please do so.  I assume that the test vectors were expecting specific
> error/advice messages that need to be updated?

Looks to be the case. The below fixes the tests on my machine. It might be worth adding tests to check the specific advice messages, but that's a different topic and regardless wouldn't belong in the http-push series.

---- >8 ----
Signed-off-by: Christopher Tiwald <christiwald@gmail.com>
---
 t/lib-httpd.sh       |    2 +-
 t/t5541-http-push.sh |    2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/t/lib-httpd.sh b/t/lib-httpd.sh
index f7dc078..094d490 100644
--- a/t/lib-httpd.sh
+++ b/t/lib-httpd.sh
@@ -160,6 +160,6 @@ test_http_push_nonff() {
 	'
 
 	test_expect_success 'non-fast-forward push shows help message' '
-		test_i18ngrep "To prevent you from losing history, non-fast-forward updates were rejected" output
+		test_i18ngrep "Updates were rejected because" output
 	'
 }
diff --git a/t/t5541-http-push.sh b/t/t5541-http-push.sh
index cc6f081..57c3e48 100755
--- a/t/t5541-http-push.sh
+++ b/t/t5541-http-push.sh
@@ -167,7 +167,7 @@ test_expect_success 'push fails for non-fast-forward refs unmatched by remote he
 '
 
 test_expect_success 'push fails for non-fast-forward refs unmatched by remote helper: our output' '
-	test_i18ngrep "To prevent you from losing history, non-fast-forward updates were rejected" \
+	test_i18ngrep "Updates were rejected because" \
 		output
 '

-- 
1.7.10.167.gf245c

--
Christopher Tiwald

Junio C Hamano· Apr 12, 2012, 21:51 UTC · re: Christopher Tiwald · lore

Re: [PATCH] Fix httpd tests that broke when non-ff push advice changed

Christopher Tiwald <christiwald@gmail.com> writes:

Show 7 quoted lines

> On Thu, Apr 12, 2012 at 08:35:11AM -0700, Junio C Hamano wrote:
>> Please do so.  I assume that the test vectors were expecting specific
>> error/advice messages that need to be updated?
>
> Looks to be the case. The below fixes the tests on my machine. It might
> be worth adding tests to check the specific advice messages, but that's a
> different topic and regardless wouldn't belong in the http-push series.

Thanks.