[Bioperl-l] RE: Bio::FeatureHolderI interface confusion (Hilmar Lapp)

[Nathan (Nat) Goodman]

Hi Hilmar

> >> Interfaces in perl to me have nothing to do with compile time
> >> whatever, they mostly provide for contracts
> >
> > The point that you're missing is that BioPerl classes break contracts
> > regularly whenever the developer feels like it.
>
> I'm going to repeat that people who run red lights don't necessarily
> mean you better remove all traffic lights.

Interesting analogy.

This debate seems to resurface every once in a while.  Perhaps we should
post a concise summary of the various points somewhere on the website so
that next time someone raises the concern, he can be pointed there.  I’d be
happy to write up “my” side with the help of Chris Mungall and Steve Mathias
and anyone else who’d care to contribute.  It wouldn't be much fun, though,
unless the other sides posted their views, too.


> > There should be a CPAN-like repository

> Good suggestion. Volunteers welcome :-)

Both you and Ewan said the same thing. Good idea if someone would do it.  I’
ll take the bait.  I'll do it IF the core developers are willing to populate
it with an initial load of modules.

What I have in mind is moving ‘non-essential’ modules out of the main
BioPerl distribution and into optional libraries.  There’s probably an 80-20
rule at work – 80% of BioPerl usage probably involves 20% of the modules.
The main distribution should have those 20% and the repository should have
the rest.

This could be phased in over a period of time – say a year.

I would imagine using the CPAN software – no need to reinvent the wheel!  We
could call it BPAN or BioPAN or BioCPAN or something.

(I’ve been told that some users have trouble getting CPAN to work.  I, too,
used to be CPAN-challenged, but have finally devised a config file that
works reliably except for a handful of gnarly modules that everyone has
trouble with (GD and BerkeleyDB being two of the most recalcitrant).  Also,
things that need root permission)

Best,
Nat