[Biocorba-l] Re: [Biojava-l] Re: CORBADEV: LSR news - new RFPabout
Tue, 19 Sep 2000 10:39:07 -0700
"Dickson, Mike" wrote:
> BTW, I also understand that there has been some
> discussion regarding wether the next RFP should allow changes to the current
> BSA model. I won't attempt to summarize the discussion at the meeting since
> I didn't attend (Scott perhaps you could?). My personal opinion is that
> some modifications are probably necessary.
The general sense at the BSA FTF (Finalization Task Force) meeting last
week was that changes to the BSA IDL should be kept to a minimum by
responses to the new RFP. I do think that some changes to the existing
IDL would be a good thing. For example, if a SequenceAlphabet is added,
BioSequence should provide access to one. If instead BioSequence were
to be extended to pick up the access to a SequenceAlphabet, then both of
BioSequence's extensions (NucleotideSequence and AminoAcidSequence) and
all three typed sequence iterators would probably warrant extensions or
changes. So, in this case, change is probably minimized by making a
simple change directly to BioSequence. The interface would be tagged
with a version number.
> BTW, since I'm giving opinions :) I do have one other. I believe the next
> RFP should focus on submissions that submit models (using UML and the UML
> Corba profile) as opposed to directly requiring IDL. Doing so would support
> a normative mapping to IDL but also allow use of the model in other contexts
> (i.e. anything that supports mapping UML to an implementation language).
I absolutely agree with the modeling approach. Some of the metadata
issues that Martin has been struggling with in BSA implementations would
have been eliminated by a modeling approach in the original spec. One
drawback to taking a modeling approach with the second RFP is that we
don't have a model for BSA.
Scott Markel, Ph.D. NetGenics, Inc.
firstname.lastname@example.org 4350 Executive Drive
Tel: 858 455 5223 Suite 260
FAX: 858 455 1388 San Diego, CA 92121