Let's restart Bioperl !

Steven Brenner S.E.Brenner@bioc.cam.ac.uk
Wed, 13 Nov 1996 17:42:30 +0000 (GMT)

Hi Georg,

  You're going to hate me.  I have just 2 weeks left in Cambrige and
am rushing to finish up several papers, so I have no time at all.  (I
will be spending the rest of the time between now and mid-January
visiting laboratories and going to conferences. I may be able to proof
code on airplane flights.)

> #Q. 1: Is it OK to use accessor functions for input data checking, during
> #      _initialize & parseFasta,etc ? That is, should the current accessors be
> #      enhanced to provide detailed data/type checking, and be
> #      called whenever data are assigned to the internal hash, 
> #      during initialization & parsing data from files ? (The current 
> #      checks in _initialize would then be replaced by accessor calls.)

Eventually, we'll probably want _initialize to do the checking itself
for performance reasons.  However, using accessor functions to do it
is elegant, effective, and compatible with later changes to duplicate
the type-checking code. 

So, yes it is OK in my book!

> #Q. 2: Is the current $SeqType/$SeqForm system satisfactory ?
> #      Basically, now the tables are just used for mapping names (like "Fasta")
> #      into a unique number code (like "7"). It's no longer ok to have
> #      either the number code _or_ the longword as a format specification)
> #      That is, `` grep {$SeqForm[$SeqForm{$_}] = $_} keys(%SeqForm); ''
> #      is no longer done, and `` $self->{ffmt} = $SeqForm[$ffmt] || 
> #      $SeqForm[$SeqForm{$ffmt}] || 'Unknown'; ''  is no longer supported.

I'm not fully clear on this, but as with the above comment, it seems
that this just removes unnecessary functionality, which could always
be added back later.  So, yes, go ahead with this.

Simplicity is a virtue, especially when it will be
forewards-compatible with complexity.