[Biopython-dev] Biopython wrappers' behavior

[Wibowo Arindrarto]

>> As per your second point, I can imagine three general ways to do this
>> (on top of my head):
>>
>> 1. Implement a method to override one parameter setting with its
>> opposing parameter in AbstractCommandline. This is perhaps similar to
>> Bio.Blast.Application's _validate_incompatibilities method, only
>> instead of raising an exception it deletes one of the parameters.
>>
>> 2. Implement a new _AbstractParameter subclass that can handle two
>> different incompatible parameters (this is perhaps too complicated)
>>
>> 3. Implement an incompatibility checking mechanism in
>> AbstractCommandline.__str__, to define parameters that can override
>> its pair (e.g. foo and nofoo). This will keep the opposing parameters
>> stored as the object attribute (so a __repr__ will reveal them both),
>> but it won't get passed on to the console as the __call__ method
>> relies on __str__.
>>
>
> Here's a fourth to consider, similar to your #1 (not to disagree with any of
> your suggestions): add an "_AntiSwitch" class to Bio.Applications, which
> includes a reference or string name of the attribute/parameter it nullifies.
> Would that be easier to specify when writing the application wrapper?
>

That seems doable :). I can't imagine it being too hard to implement
technically, as this will only be used for options that can be grouped
under one name (i.e. opposing boolean parameters).