Perl test case suggestions?

Ralf Corsepius rc040203 at freenet.de
Thu Apr 7 05:06:37 UTC 2011


On 04/06/2011 05:08 PM, James Laska wrote:
> On Wed, 2011-04-06 at 09:48 -0500, Bruno Wolff III wrote:
>> On Wed, Apr 06, 2011 at 10:45:16 -0400,
>>    James Laska<jlaska at redhat.com>  wrote:
>>> Greetings folks,
>>>
>>> I'm not a perl user, and it's been *some* time since I last knowingly
>>> opened up a perl script.  While providing some karma feedback for a
>>> recent perl update, it occurred that it might be nice to have a few
>>> sample perl test cases linked from the bodhi update request.
>>>
>>> Does anyone have any perl test case suggestions that we can add to
>>> https://fedoraproject.org/wiki/Category:Package_perl_test_cases ?
>>
>> It is fairly normal for perl modules to come with test cases. I am not
>> sure of what you need to do to run them as they are installed, but it is
>> normal to run them when building them from source.
>
> Is that something a tester should validate from the koji build logs?
It's convention for all fedora perl packages to run these test
in %check.

I.e. usually (i.e. there are exceptions), a perl-package which fails its 
testsuite will fail building.

> Meaning, check whether %check ran, and passed?
So far, this has been considered responsibility of the Fedora package 
maintainer as well as has making sure %check runs a perl package's 
testsuite and produces reasonable results, always been been part of the 
Fedora package reviews.

> Anything else to consider?
I am not sure what you are aim at.

The current number one issue with perl packages in Fedora are package 
dependencies (perl module dependencies) and the impact of changes in 
rpm's perl dependency-generator between different versions of rpm.

Bugs in functionality usually are much covered by perl packages' 
upstreams than they are by Fedora users or Fedora package maintainers. 
In other words, the usual situation with perl packages is upstreams 
releasing updates, which fix bugs no Fedora user had reported nor the 
Fedora maintainer has tripped over.

Ralf



More information about the test mailing list