[Bioperl-l] WinXP test results

[Nathan S. Haigh]

Sendu Bala wrote:
> Nathan S. Haigh wrote:
>> Here are test results for Bioperl 1.5.2-c on WinXP Pro SP2. In all 
>> cases, Perl, Bioperl and only known dependencies (and BerkeleyDB) 
>> from Makefile.Pl were newly installed.
>>
>> Running ActivePerl 5.8.8.819:
>> Failed Test             Stat Wstat Total Fail  Failed  List of Failed
>> ------------------------------------------------------------------------------- 
>>
>> t/BioDBSeqFeature_BDB.t               43   43 100.00%  1-43
>> t/protgraph.t                         66   24  36.36%  10-13 20-21 26 
>> 33 36-37
>>                                                        45 48-56 59-60 
>> 65-66
>> 5 subtests skipped.
>> Failed 2/238 test scripts, 99.16% okay. 67/12359 subtests failed, 
>> 99.46% okay.
>>
>> The BioDBSeqFeature tests are the same fails already reported by 
>> Chris, but I also get fails in t/protograph.t. Further details at the 
>> bottom of this e-mail.
>>
>>
>> Running ActivePerl 5.6.1:
>> Trying to get Bioperl installed on ActivePerl 5.6.1 is a total pain 
>> (which I still failed to do, even after a few days of trying). There 
>> are several reasons, a few dependencies (and more than a few 
>> dependencies of Bioperl dependencies) for which there are no PPD's 
>> for Perl 5.6 to install via PPM. Installing via CPAN, didn't work for 
>> me since nmake dies with many test suites. Therfore, we should 
>> probably recommend ActivePerl 5.8 for Windows.
>
> Thanks Nathan. I think what I'll do is say that Bio::Graph* is 
> deprecated with this release and simply drop t/protgraph.t. 
> Nonetheless, is it possible for you to require Class::AutoClass 1.01 
> for Bio::Graph*?
I've already done this and prepared a PPD for PPM4, but if you do 
deprecate it for this release, I'll remove the dependency from the PPD. 
Is it just Bio::Graph that requires Class::AutoClass?
>
> Is everyone happy with saying the perl requirements are 5.6.1 for all 
> platforms except WindowsXP, which needs ActivePerl 5.8+?
>
> In that case, how close are install-related things (Makefile.PL, PPDs 
> et al.) to being ready?
With the new PPM4, you new specify the name of the modules which are 
dependants rather than the package (in some cases the package name 
doesn't match the module it provides e.g. IO::Scalar is provided by the 
IO::Stringy package), now the PREREQ_PM can include the same modules 
names that are in %packages in Makefile.PL. Even though most 
dependencies will go in Bundle::BioPerl, it makes things a little easier 
when making the PPD to have them all specified in PREREQ_PM and keep 
them all in one place - at least for the time being (i'll commit changes 
later, once i've tested). However, it now means there are 2 lists of the 
dependencies to maintain within the same Makefile.PL, is there some way 
that PREREQ_PM can use %packages for it's list of dependencies? Maybe 
something to think about for the 1.6 release?

Anyway, I'm almost ready with PPD's for PPM4 for Bioperl, some missing 
dependencies and hopefully also a Bundle::BioPerl. Should definately 
have this sorted by the end of this week.
>
> I think the Bio::DB::SeqFeature failures are sufficient to delay RC1 
> however. Sorry, I had forgotten about them. Chris/Nathan, are you 
> looking into it? Shall I take a look as well?
I'll try to have a little look into it, but I have limited time this 
week and then I'm on holiday for 2 weeks. So, it might be worth you 
having a look as well.
>
> I'll push back RC1 a week to October 2nd.