[Open-bio-l] BioSQL schema: some questions

[Hilmar Lapp]

What about constructing an appropriate test suite in the biosql repository itself that covers e.g. the queries issued by the different bio* adaptor implementations?

I would think for this kind of interoperability project it is paramount to have appropriate unit tests. I actually started directory t/ with a tiny first test that at least checks whether the DDL passes syntax. Likewise, one would need a bunch of test data and all the different queries with their supposed result.

	-hilmar
-- 
-------------------------------------------------------------
Hilmar Lapp                            email: lapp@gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------



> -----Original Message-----
> From: Thomas Down [mailto:td2@sanger.ac.uk]
> Sent: Tuesday, April 30, 2002 12:11 PM
> To: Chris Mungall
> Cc: Hilmar Lapp; OBDA BioSQL (E-mail); 
> gmod-devel@lists.sourceforge.net
> Subject: Re: [Open-bio-l] BioSQL schema: some questions
> 
> 
> On Tue, Apr 30, 2002 at 11:26:22AM -0700, Chris Mungall wrote:
> > 
> > > Shouldn't there be a UK constraint on biodatabase.name?
> > >
> > > I've found couple of other little things that probably need to be
> > > cleaned up, and as I'm going over every entity I may find 
> more. I can
> > > try to list them all, or if you prefer (would probably be 
> easier for me,
> > > too) I can fix them directly in the repository.
> > 
> > fixing in the repository is best, so long as the bioperl-db 
> test suite
> > still passes.
> 
> Ahem.
> 
> bioperl-db isn't the only codebase to hang off these schemas.
> There are certainly equivalents in BioJava and Biopython, and
> (I think) some others as well (Ruby?).  Everyone will be using
> rather different adaptor designs, and different `styles' of
> SQL query to access to database.
> 
> Could I reiterate my request that any changes, however minor,
> get notified to open-bio-l?  Actually, this repository might
> be one case where it's really worth having one of those automatic
> CVS-logging mailing lists.  Checkins should (hopefully) be
> sufficiently rare than it won't be a full-time job reading
> the commit logs, but it would avoid the ``hey, this is such
> a minor fix that nobody else will notice'' syndrome.
> 
> Sorry to be a nag about this -- I'm just a little paranoid about
> our hard-won interoperability :-).
> 
>     Thomas.
>