splitting out Bio::OntologyIO::InterProParser into separate distribution

[carandraug]

The Bio::OntologyIO::InterProParser brings in the requirement on the libxml-perl distribution (for the XML::Parser::PerlSAX module) which is something that causes people issues (see #314).

I'm looking at it and I'm not sure what's the right group of modules to move with it as I'm not familiar with this. I guess there is Bio::OntologyIO::Handlers::InterProHandler. But there is also a Bio::Ontology::InterProTerm which I don't quite know what it's used for.

@cjfields can you help with this?

[cjfields]

I saw that a number of XML dependencies have left the distribution, awesome! I can check this (might be a week or two, inundated with spring work and I'm playing catchup this weekend and maybe next). I do know the various IO modules normally load the format-specific plugin module dynamically , so just looking for the name of the module likely isn't enough.

[carandraug]

If you just tell me the list of modules you want to split on that distribution, I can do it. It's not only figuring out the modules that use Bio::OntologyIO::InterProParse it is also that sometimes modules, only (or mostly only) make sense when grouped together even if they are not requirements of each other.

[cjfields]

...it is also that sometimes modules, only (or mostly only) make sense when grouped together even if they are not requirements of each other.

Yes, I agree. Split out the full Bio::Ontology* set of modules along with tests, etc. The central issue is from the extra dependency, not that Bio::OntologyIO::InterProParser has a bug itself, so it helps solve the initial problem.

On top of this I highly suspect the Bio::Ontology modules aren't used, it simply may be better to cleave them out anyway. Worst case, someone complains about not being able to use these in the latest BioPerl release and we point them in the right direction.

[carandraug]

Split out the full Bio::Ontology* set of modules along with tests, etc.

What about Bio::Annotation::OntologyTerm and Bio::SeqIO::locuslink and other modules that grep tells me reference Ontology modules?

[cjfields]

What about Bio::Annotation::OntologyTerm and Bio::SeqIO::locuslink and other modules that grep tells me reference Ontology modules?

From a quick search it appears those could be moved along with them.

The only way to know if removing these would have any fallout would be to to a clean test run w/o them (e.g., one where you'd run standard bioperl tests w/o the various Bio::Ontology modules and where they wouldn't be in the PERL5LIB).

[cjfields]

A similar search for locuslink shows that it's pretty limited in use as well. That makes sense, NCBI's LocusLink is no longer available, was taken over by Entrez Gene I think

[carandraug]

What about Bio::Annotation::OntologyTerm and Bio::SeqIO::locuslink and other modules that grep tells me reference Ontology modules?

From a quick search it appears those could be moved along with them.

The only way to know if removing these would have any fallout would be to to a clean test run w/o them (e.g., one where you'd run standard bioperl tests w/o the various Bio::Ontology modules and where they wouldn't be in the PERL5LIB).

This approach only tells me whether bioperl tests would fail when I move them but does not tell me whether it makes sense to make 1, 2, or 3 distributions out of those modules, or if I'm moving Bio::Annotation::OntologyTerm out whether there are other modules that make sense to be moved out with it (even if they don't tests to fail).

A similar search for locuslink shows that it's pretty limited in use as well.

I have now moved locuslink to its own separate distribution. If you notice other modules that only make sense to support services that are no longer available, let me know to have them moved out too.