[Bioperl-l] RC2 test results on WinXP

Nathan Haigh n.haigh at sheffield.ac.uk
Wed Oct 18 14:17:35 UTC 2006


Chris Fields wrote:
>> Nathan Haigh wrote:
>>     
>>> I get all tests passing except for BioDBSeqFeature_mysql which fails all
>>> tests (1-46).
>>>
>>> During perl Makefile.PL I get:
>>> "I see you have Berkeleydb installed. I will create the DBD tests for
>>> Bio::DB::SeqFeature::Store..."
>>>
>>> I notice under the "needs investigation" there is mention about tests
>>> been generated even if DBD::mysql isn't installed. I assume this is the
>>> problem?
>>>       
>> Probably. I'm looking into it. Not sure why it wasn't causing a problem
>> before now.
>>     
>
> Funny, I don't have problems with the BioDBSeqFeature_mysql tests on WinXP
> because 'perl Makefile.PL' doesn't detect my MySQL installation, so the
> MySQL-based tests don't run even though I have DBD::mysql installed.  I
> thought this might just be a WinXP issue, but apparently not.  If I can get
> to it I'll run a few checks.
>
>   
This was on WinXP.
>>  > If this is the problem should DBD::mysql be added to the
>>  > dependencies in Makefile.PL?
>>
>> No. You can use the modules in question without mysql (presumably; ie.
>> you have a different sql setup), so it makes no sense to warn people
>> they don't have a module they absolutely do not need.
>>     
>
> Agreed, though I don't know if other relational DB's are supported like
> PostgreSQL.
>
>   
>>> Is there an easy way to find out what tests are being skipped due to
>>> absent modules?
>>>       
>> Ideally, when the skip occurs the test script will issue a message. I
>> think that happens in most, if not all cases.
>>     
>
> Yes, though we may run into the same issue we had with XEMBL tests not
> reporting the reasons it skipped.  Each test suite should run an eval{} to
> check the required modules, then only skip blocks of tests that rely on
> those modules.  I think we have caught most of those, but who knows w/o
> doing a complete test suite audit?
>
> Our eventual complete switchover to Test::More should hopefully clean these
> up.  I don't consider it a pressing issue for this release, though Sendu may
> feel differently.
>
> Christopher Fields
> Postdoctoral Researcher - Switzer Lab
> Dept. of Biochemistry
> University of Illinois Urbana-Champaign 
>
>
>   




More information about the Bioperl-l mailing list