[Biopython-dev] interface/SQL development policies

[Yves Bastide]

James Casbon wrote:
> Hi, 
> 
> Say I have a module X, which depends heavily on flat file databases.  I wish 
> to decompose into two classes XFile and XDatabase, where XFile uses the files 
> and XDatabase use and SQL backend.  This raises a number of questions:
> 
> * the two classes imply a common interface X, what is the best way of handling 
> this in python?  An abstract base class?  

Separate classes, the interface being checked via unit tests

> 
> * How are regression tests managed for SQL based classes (ie how do you 
> provide a dummy database, given there might not be an sql server on a 
> particular client)?

No good answer.

1. Skip the tests if there's no db available

2. Use Gadfly?

> 
> These questions might be best answered with an example, so if there are any 
> examples covering this kind of stuff already in biopython, please point them 
> out.
> 
> Thanks,
> James
> 

yves