[Biocorba-l] Bioperl biocorba server code checked in

[Ewan Birney]

On Sat, 10 Feb 2001, Brad Chapman wrote:

> ==========
> Fixed
> -----
> * PrimarySeq::subseq -- was trying to check max_request_length, but 
> there wasn't any such thing. Just defined an internal hash value for
> it, set to some arbitrary big number like 1Mb. 

Done.

> 
> * PrimarySeqDB::version -- getting seg-faults when trying to call
> this. Problem was a reference to seqdb, instead of _seqdb. Also, the
> database tries to return a version like 0.2, but it can't do this since
> only longs are defined, so zero is always returned.

Done.

> 
> * PrimarySeqDB::get_PrimarySeq -- has the same problem with referencing
> seqdb instead of _seqdb.
> 

done

> * PrimarySeqDB::get_PrimarySeq again -- There also doesn't seem to be
> a function in the Bio::Index stuff to get a PrimarySeq directly from an
> id. Instead, I got the Seq, and then got a primary seq from this.
> 

This is equivalent-ish in bioperl. But I've fixed this properely now.


> * PrimarySeq::get_SeqFeatures_in_region_by_type -- Complaints about
> Bio::Range not having a "new". Fixed by including use Bio::Range;
> 

Done.

> * SeqFeatureIterator::next -- wasn't checking for EndOfStream;
> 

Done.


> * SeqFeatureVector::elementAt -- wasn't turning the bioperl SeqFeature
> object into a biocorba SeqFeature object.
> 

Done.


> * Seq::get_SeqFeatures_in_region[_by_type] -- was raising
> UnableToProcess errors instead of OutOfRange errors.
> 

Not done... I've stopped now... machines are back up over here and 
I am off to do genome stuff...



> * Seq::get_SeqFeatures_in_region[_by_type] -- trying to use 
> range->contains on a SeqFeature object and not a range object. Changed
> to create a range for the SeqFeature, and then check.
> 
> Not fixed
> ---------
> * Seq::get_SeqFeatures_in_region/by_type -- All of these functions first
> make a call to all_Features, which retrieves BioCorba objects, and not
> bioperl objects. Then when you try to get a SeqFeature out of a Vector 
> or Iterator created with these objects, the SeqFeature constructor will
> die with: "Must have poa and seq into Seq Feature." I didn't want to get
> too into this, since this would involve rewriting parts of these 
> implementations -- I'm not sure about the best way to handle it.
> 
> * PrimarySeq/AnonymousSeq implementation -- one issue with the way you
> implemented these as one class is that if a client tries to narrow a
> PrimarySeq object to the AnonymousSeq interface (which it should be able
> to do), the server will die since it has no AnonymousSeq implementation
> to narrow to. The solution to this would be to implement AnonymousSeq
> separately and then inherit from it with PrimarySeq.
> 
> * Unknown::query_interface -- Anything I do here causes the server to 
> crash :-<. If you try to return anything but a CORBA::Object, ORBit
> will complain about having problems marshalling the object, and if you
> try to return a CORBA::Object, the server seg-faults. I don't know what
> to do to just keep this from blowing up. Any thoughts?
> 
> 

-----------------------------------------------------------------
Ewan Birney. Mobile: +44 (0)7970 151230, Work: +44 1223 494420
<birney@ebi.ac.uk>. 
-----------------------------------------------------------------