Re: [PATCH v4 2/2] travis-ci: build documentation

From: Lars Schneider <larsxschneider@gmail.com>
Date: May 13, 2016, 06:26 UTC
Message-ID: <87079FFA-996D-4F8D-A2A8-7C8917CEF545@gmail.com>
In-Reply-To: <xmqqpostrg67.fsf@gitster.mtv.corp.google.com>

Show 28 quoted lines

> On 10 May 2016, at 19:12, Junio C Hamano <gitster@pobox.com> wrote:
> 
> larsxschneider@gmail.com writes:
> 
>> From: Lars Schneider <larsxschneider@gmail.com>
>> 
>> Build documentation as separate Travis CI job to check for
>> documentation errors.
>> 
>> Signed-off-by: Lars Schneider <larsxschneider@gmail.com>
>> ---
>> .travis.yml              | 15 +++++++++++++++
>> ci/test-documentation.sh | 14 ++++++++++++++
>> 2 files changed, 29 insertions(+)
>> create mode 100755 ci/test-documentation.sh
>> 
>> diff --git a/.travis.yml b/.travis.yml
>> index 78e433b..55299bd 100644
>> --- a/.travis.yml
>> +++ b/.travis.yml
>> @@ -32,6 +32,21 @@ env:
>>     # t9816 occasionally fails with "TAP out of sequence errors" on Travis CI OS X
>>     - GIT_SKIP_TESTS="t9810 t9816"
> 
> Completely offtopic, but this looks like this is made to apply to
> all archs, not limited to OSX?  It of course would be ideal to see
> why they fail only on OSX and fix them, but shouldn't the blacklist
> at least limited to the platform with the problem?

As far as I remember the test was flaky on Linux and OSX. The problem was not Git. The problem was the Perforce Server "p4d" p4 daemon which would refuse to die sometimes.

I run a bigger number of Travis tests and it looks like the problem is gone. Out of 60 runs only just one failed and t9810 / t9816 passed: https://travis-ci.org/larsxschneider/git/jobs/129383851 (BTW: does anyone know how I can make prove show me the running tests? I would like to know which test died there..)

It looks like as Perforce solved the t9810/t9816 problem with the update from 15.2 to 16.1 that we made in 31f3c86 "travis-ci: update Git-LFS and P4 to the latest version". I will post a patch to enable these tests, again.

Thanks, Lars