[Open-bio-l] bioperl-live Bio::DB::Registry.pm bug finding "/etc/bioinformatics/seqdatabase.ini" and suggestion for discovery of new database adaptors

Malcolm Cook malcolm.cook@genxy.com
Fri, 14 Jun 2002 10:51:00 -0700


This module loads the user's registry in his $HOME directory after testing for
one in /etc/bioinformatics...

Oh, the woes of cut and paste...

the 2nd instance of the line

 open(F,"$home/.bioinformatics/seqdatabase.ini");

should read

 open(F,"/etc/bioinformatics/seqdatabase.ini");

Can fix?

On the subject, elsewhere in the module, the approach of hardwiring the
implementations in the hash %implements can be usefully extended to 'discover'
new BioDatabaseAdaptor modules (a borrowed naming convention) allowing their
dynamic loading from sub-directory of Bio::DB named identically after the
protocol as it appears in seqdatabase.ini.

As follows...

put this in the modules header:

use UNIVERSAL::require;

and apply the following patch to  _load_registry


     if (defined $implement{$hash->{'protocol'}}) {
  $class = $implement{$hash->{'protocol'}};
  } elsif (($class =
"Bio::DB::".$hash->{'protocol'}."::BioDatabaseAdaptor")[0]->require) {
  $class->import;
  } elsif ($UNIVERSAL::require::ERROR !~ m/Can\'t locate .* in \@INC/) {
  die "Error while requiring $class: $UNIVERSAL::require::ERROR";
  }
     else {
  $self->warn("Registry does not support protocol ".$hash->{'protocol'});
  next;

Why the trouble?  I'm considering the approach creating/registering/using an
adaptor for SRS/getz.

Comments?  Criticism?

Thanks & Regards,

--
Malcolm Cook – Software Engineer
Genset, Inc – San Diego, USA
Tel: 858-597-2664


(Some day I'll request a CVS account and also start to use patch / diff...)