[Bioperl-l] evalues/floating point tests

[Chris Fields]

Oh, my guess is you'll see the same errors.  I didn't have a Windows  
system to test on until yesterday so no changes were made until a  
couple hours after RC3 was packaged up.  Unfortunately every test I've  
seen has been from anything else but Windows.

A checkout and test of main trunk would help more, particularly after  
the evalue and float_eq() commit (which may come down the pipe  
tonight).  I can weed out the tests for modules not in the branch.

chris

On Jan 17, 2009, at 5:08 PM, Mark A. Jensen wrote:

> OK- I'll do a fresh build of RC3 and send the results- (I recall  
> seeing the issue in the HSP tests some while back as well...)
>
> ----- Original Message ----- From: "Chris Fields" <cjfields at illinois.edu 
> >
> To: "Mark A. Jensen" <maj at fortinbras.us>
> Cc: "BioPerl List" <bioperl-l at lists.open-bio.org>
> Sent: Saturday, January 17, 2009 5:22 PM
> Subject: Re: [Bioperl-l] evalues/floating point tests
>
>
>> Mark,
>> If you have any additional errors post them here if you have time   
>> (particularly for Strawberry Perl):
>> http://www.bioperl.org/wiki/Release_1.6_Testing
>> or you can send them in to me.  Related to Win installation I may  
>> try  downloading ActivePerl for Mac OS X and test out PPM  
>> connectivity if I  have time this weekend.
>> I'm pretty sure the evalue issue has popped up before.  In fact,  
>> Scott  Markel reported additional zeros appearing in evals, along  
>> with file  name problems with Windows.
>> chris
>> On Jan 17, 2009, at 4:06 PM, Mark A. Jensen wrote:
>>> I am getting these funky fails as well (ActiveState/Cygwin/Vista  
>>> and  strawberry/Cygwin/Vista); I'm all for the kludge, should  
>>> prevent a  LOT of chatter at support, but I can try to ferret out  
>>> the  underlying issue.
>>>
>>> ----- Original Message ----- From: "Chris Fields" <cjfields at illinois.edu 
>>>  >
>>> To: "BioPerl List" <bioperl-l at lists.open-bio.org>
>>> Sent: Saturday, January 17, 2009 4:53 PM
>>> Subject: [Bioperl-l] evalues/floating point tests
>>>
>>>
>>>> I've noticed an odd issue when testing BioPerl on WinXP  
>>>> yesterday   which doesn't appear to be popping up anywhere else.   
>>>> When  comparing  floating point numbers, several tests fail like  
>>>> the  following:
>>>>
>>>> #   Failed test at t\SearchIO\blast_pull.t line 40.
>>>> #          got: '1e-23'
>>>> #     expected: '1e-023'
>>>>
>>>> #   Failed test at t\SearchIO\blast_pull.t line 67.
>>>> #          got: '6e-59'
>>>> #     expected: '6e-059'
>>>>
>>>> These are generally in lines like (for the example above)
>>>>
>>>> is $result->get_parameter('expect'), 1e-23;
>>>>
>>>> I'm adding a method to Bio::Root::Test for comparing two  
>>>> floating   point numbers based on equality:
>>>>
>>>> float_eq $float1, $float2, $message;
>>>>
>>>> This basically wraps is() but runs both values through   
>>>> 'sprintf("%g",  $foo)' prior to the comparison; it has a  
>>>> prototype  requiring two  values to be passed and an optional  
>>>> message.  Any  objections to the  method name or adding this in?   
>>>> I've already  converted many of the  tests in t/SearchIO over but  
>>>> wanted to pass  it by the group prior to  an actual commit.
>>>>
>>>> chris
>>>> _______________________________________________
>>>> Bioperl-l mailing list
>>>> Bioperl-l at lists.open-bio.org
>>>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>>>>
>>>
>> _______________________________________________
>> Bioperl-l mailing list
>> Bioperl-l at lists.open-bio.org
>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/bioperl-l