[Biocorba-l] BSANE and bioCORBA

[Brad Chapman]

Hey all;

Juha:
> We must really start to work more to get the BSA/BSANE and bioCORBA
> threads to merge.

You've all got my agreement -- it would definately be nice to have one 
spec; I'm glad you brought this forward, Juha.

Mostly I trust you all to hash over the details -- most of what you've 
said so far is sensible (or I was not smart enough to know what you
were talking about :-). Some specific comments on things are below.

Ewan:
> Jason/Brad - you should dig in before we sign ourselves up for something
> we don't particularly like!

Okay :-). BTW, for those who don't know, I'll be coming at all of this
from a Python perspective.

Ewan:
>    I need to revisit CosPropertyService and CosLifeCycle - if I remember
> right they make life a pain in the arse for implementors - I prefer the
> GNOME reference counting system and just a short a simple key - list-value
> system as in BioCorba. 

I am definately on Ewan's side here -- I *really* don't want to have
to implement these services in python just to do my biology-related
work. Also, since we are only pulling out small bits of these
specifications, I think we lose in clarity what we gain in "standards
conformity." Personally, I'd much rather have something simple that we 
just encompass in a single IDL -- so far CORBA itself scares away
enough people, no sense in making it even more complicated.

Juha:
> Perhaps we should use the iterator/list pattern used in the BSA. It is
> actually quite good!

I agree that the single Iterator object with the next_n() function is
an improvement of the separate Vector/Iterator objects that BioCorba
already has. I'd like to see this change

The implementation does feel a bit awkward to me. Right now it uses
an out parameter to return the object from the iterator, and uses a
boolean return value to specify whether the object was returned or
not. I greatly prefer the current BioCORBA model where the return
value is the object, and the call raises an EndOfStream exception if
it is out of objects. Maybe this is a dumb implementation detail, but
the boolean return parameter really caught me off guard when I was
reading it.


In general, I'm happy to see us getting something coming together!
Thanks again for Juha for bringing this forward, and having IDLs for
us to look at.

Still-too-dumb-to-understand-UML-ly y'rs,
Brad