Question about registry support of general RESTful services.

[Douglas Tody]

I'd suggest that the first question to ask, for each such service, is
whether a fully RESTful (file/resource oriented) or object oriented
(object with methods) model is more appropriate for the specific
service.  In the latter case the DAL interfaces define an existing
standard pattern which being OO is not tied so closely to HTTP, but
which is nonetheless integrated with the existing VO RESTful standards
such as for UWS and VOSI.

Both are good models but they are quite different - OO is more
appropriate for accessing things which are well defined objects (e.g. an
image), the file/resource model is more appropriate for structured
information.

 	- Doug


On Wed, 4 Jan 2012, Theresa Dower wrote:

>
>
> I ask because a colleague (Steve Handy) at STScI and I are looking for a way to catalogue our non-VO-standard services for internal use and wondering if the registry paradigm fits the services we already have and ones we're planning to build soon. He's suggested looking to WSDL 2.0 for answers on RESTful service descriptions if there's not already a VO way to do things. I'd like to start working on this once I've read through it and our use cases and the existent special-case VO interface descriptions, and welcome help, input, a hostile takeover, from anyone. Anybody? Bueller?
>
> -----Original Message-----
> From: Ray Plante [mailto:rplante at ncsa.uiuc.edu]
> Sent: Wednesday, January 04, 2012 4:29 PM
> To: Theresa Dower
> Cc: registry at ivoa.net
> Subject: Re: Question about registry support of general RESTful services.
>
> On Wed, 4 Jan 2012, Theresa Dower wrote:
>> Is there anywhere in the IVOA standards docs that I've somehow missed for
>> describing generic RESTful services? The ParamHTTP standard doesn’t quite
>> fit for a lot of real-world cases. Beyond expanding queryTypes for ParamHTTP
>> to include UPDATE/DELETE/ETC,  I also mean some kind of  support for
>> parameters that are XML or JSON blobs.
??
> No, in general.  ParamHTTP represents a special though very common
> subset.  This was done to keep the description of the interface very
> simple and cover the 80%, including DAL service interfaces.
>
>> If not, I’m beginning to see
>> the need for a general extension for it.
>
> I agree.  I'd love to see a strawman floated for it.
>
> cheers,
> Ray
>