[Bioperl-l] Splitting Bioperl and Test related Suggestions

[Chris Fields]

On Jul 5, 2007, at 9:30 AM, Nathan S. Haigh wrote:

> ...
>> I actually like Sendu's idea more, or the idea of each test suite  
>> having it's own directory.
>> Tests which need to guess/validate the format are probably best  
>> left sequestered to a specific suite focused on format guessing/ 
>> validation, at least in my opinion.
>> chris
>
>
> How easily would this lend itself to using the same data for  
> multiple tests, or is it likely to lead to/exacerbate a culture of  
> adding duplicate data files in each "test suite" rather than reusing?
>
> Nath

If there is a group of test data used for more than one test suite we  
can group those together into a common use folder, or we can go by  
format.  I'm pretty open to anything, really, as long as it is more  
organized.

My point is really concerned more with validation/guessing.  I think  
we should limit those tests to their respective specific test suites,  
or even to sections within a particular test suite (for instance,  
genbank.t), but not to force sequence guessing or validation in other  
cases.  To me validation, guessing, and parsing are three distinct  
issues (much like XML parsers handle things), so they require three  
distinct tests.

As for true sequence validation, there is no official format  
validation scheme yet in BioPerl.  It's sort of unofficially  
intergrated into the sequence parsers themselves (something which I  
find to be problematic for several reasons too long to outline here).

chris