Grouper Developers Forum

[Chris Hyzer <>]

That is an impressive amount of work for 2 hours, and good information.  
Thanks.

Gary also mentioned that REST typically uses the various HTTP methods (PUT, 
DELETE, etc), as in your discussion notes.

I am hoping we can just use what Axis2 does for REST which is the other more 
loosely defined REST strategy (described in second paragraph):

http://en.wikipedia.org/wiki/Representational_State_Transfer

According to this Axis2 supports GET and POST and does RPC type requests

http://ws.apache.org/axis2/0_94/rest-ws.html

I would think the amount of work to code and manage a client should be the 
roughly the same if the params are URL params and not an input XML (let me 
know if I am missing something).
Let me know if we need to discuss this, otherwise I will proceed.

I do like the idea of putting in strategies to allow change without changing 
the wsdl (e.g. how strict REST can return XHTML, and if the XHTML changes 
slightly, it might not break an existing client).  At Penn we do loose-REST, 
and pass in some unused param names and values reserved for future use (a lot 
of protocols do this).  So I will think about how this could work, and if you 
object to this, also let me know.

Thanks!
Chris

Ps. Option #1 (grouper-ws in different project as ui, with separate library 
versions) in the previous email is in the lead for the vote tally...


> -----Original Message-----
> From: RL 'Bob' Morgan 
> [mailto:]
> Sent: Thursday, January 10, 2008 8:01 PM
> To: Grouper Dev
> Subject: [grouper-dev] early work on RESTful groups service interface
>
>
> As I've mentioned, REST is all the rage around here.  Here is a writeup
> of
> a recent design exercise on a RESTful interface to our existing
> non-Grouper groups service:
>
>    https://wiki.cac.washington.edu/x/TpNh
>
> The meat of the resulting design, such as it is, is in sections 4 and
> 5.
>
> For background, UW's current groups service is minimal, basically a way
> to
> import group info from various sources and export into an LDAP
> directory.
> So the design so far is about reading and writing; no searching, no
> namespace (stem) management, no subject source access, etc.  But I
> thought
> it could be useful for folks to consider the approach.  In particular
> eyebrows might be raised about the choice of XHTML for output rather
> than
> XML; I guess this is a common suggestion for RESTful interfaces.
>
>   - RL "Bob"