registration and declarations to the hub in a single atomic operation

[Mark Taylor]

On Fri, 13 Mar 2009, Yohann Granet wrote:

> > In designing the interface we deliberately separated the registration,
> > metadata declaration and subscription declaration for a few reasons.  The
> > two main ones that I recall were:
> > 
> >  (a) The registration call can take different forms, depending on profile,
> >      and by separating that from the other things, the amount of the
> >      interface which was profile-dependent could be minimised.
> >   
> I didn't know that registration calls were depending on profile...
> In these conditions I understand that splitting registration, declaration and
> subscription simplifies the job in the SAMP internal processes.
> however it's a bit sad that, in order to avoid complex operations inside the
> framework, the solution chosen obliges the SAMP framework users to deal with a
> compulsory split operation of registration and its (little) inconveniences...
> (yet I don't know the benefit(s) (certainly important) of the actual
> registration way in the SAMP framework internal job!!)

just one comment on that.  The decision to split these calls up was not
to make it easier for implementors, rather to make it conceptually
clearer what was going on.  If only the registration call changes
between profiles, then more of the rest of the interface, and more
of the documentation, can be unchanged between profiles.  The idea was
that this makes it easier for users to understand what is actually
profile dependent (registration has to be) and what is not 
(metadata and subscriptions declaration).

Mark

-- 
Mark Taylor   Astronomical Programmer   Physics, Bristol University, UK
m.b.taylor at bris.ac.uk +44-117-928-8776 http://www.star.bris.ac.uk/~mbt/