Bioperl: Possible Seq.pm reimplementation

[Ewan Birney]

I want to bring up the topic of possibly reimplementing Seq.pm for
the 0.06 series of bioperl. Because Seq.pm is the most important
object to bioperl, I'm cross posting this to the discussion list
to get more people to look at it. We'll need alot of eyes on the
design of this (plus some probably clear thinking and persuasion)
if it will work...


I have made a web page outlining some of my ideas. I hope we will
really be able to discuss this alot in heidelberg.

URL is:

http://bio.perl.org/Projects/Sequence/reimpl.html

Could people try to direct the discussion to the guts list where this
does belong, but if you do have an opinion, even as a part time (or
general observer) do make it known ... the more views we get the better
idea what problems we are up against.

[As an aside - I'd like to make it clear that I am thinking hard about
XML interfaces generically to bioperl stuff, and I am pretty sure I have
a good solution that we can fit onto all of bioperl. I want to discuss it
first with some XML gurus before I suggest it, but it is an orthogonal
issue - though extremely important one - to interface definitions.]


Here is the start of the html text:



This page is about the possible reimplementation of Seq.pm. I
(Ewan) strongly feel this should happen because: 

     There is too much code in Seq.pm, some useful, some not 

     The design of the object is not that clean 
               
     There is no clear interface definition for the object to
allow other implementations 

     We need to seriously deprecate the IO handling of Seq.pm to
using the SeqIO system. 

That said, I want to make it clear that we owe a great debt to the
designers and coders of Bio::Seq.pm, which has been the workhorse in
bringing bioperl together, the first object of bioperl and undoubtedly the
most used. This of course makes the challenge of reimplementation that
much harder, as this object is key for all of bioperl.

But like netscape with the switch to the NGL for Mozilla5 and other good
projects, we need to bite the bullet, and move on. I think a full scale
reimplementation is needed, as the code is semi-spagetti down there,
documentation way out of whack (and in my opinion, far too long!), and I
think a certain design shift is needed.


<<flamesuit on>>



Ewan Birney
<birney@sanger.ac.uk>
http://www.sanger.ac.uk/Users/birney/

=========== Bioperl Project Mailing List Message Footer =======
Project URL: http://bio.perl.org/
For info about how to (un)subscribe, where messages are archived, etc:
http://www.techfak.uni-bielefeld.de/bcd/Perl/Bio/vsns-bcd-perl.html
====================================================================