On Jan 5, 2013, at 10:38 PM, Jerome Velociter <jerome(a)velociter.fr> wrote:
Le 05/01/13 21:17, Vincent Massol a écrit :
On Jan 5, 2013, at 6:20 PM, Jerome Velociter
<jerome(a)velociter.fr> wrote:
Hi Vincent,
That's a good idea and that's something I had in mind for some time too :)
What part(s) exactly of our REST module do you imagine moving to commons ?
Just
the part that bridges Restlet and our component model.
In my opinion that's already too much.
* That bridge uses a trick that forces to have the FQN of the class as a hint, it's
not a proper integration with the CM.
* That would imply we want to continue with Restlet in the future. Personally I'm not
sure I'd vouch for that, for several reasons. In particular the fact that JAX-RS is
sort of a second class citizen in Restlet land ; so since we are building on JAX-RS it
means both we can't benefit from many Restlet features that would be interesting, and
we don't benefit either from the features of an (other) framework that focuses solely
on JAX-RS. See for instances injection providers in Jersey
http://codahale.com/what-makes-jersey-interesting-injection-providers/
2 points here:
1) I've seen mentions of dynamic addition of REST resources in Restlet. I need to talk
to Jerome about that
2) the goal here was not to redo the REST system and it doesn't have to be. It's
not because something is in commons or rendering that we can't change it after.
It's fine to want to improve our REST support but it doesn't have to be a
prerequisite to introduce a Rendering REST APIs (which we already partly have AFAIK),
especially since from a user POV all that counts are the URLs. The rest (ahaha ;)) can
change without affecting the user.
I still haven't made my mind but I was thinking of maybe proposing this for Devoxx FR,
which would mean a short timeframe.
I think the "Future of XWiki REST architecture" is interesting and should be a
new email thread.
Thanks
-Vincent
Jerome
>
>> I ask because to be honest I'm not a big fan of how our REST API is built and
what API it exposes ; and moving it to commons would be to "promote" it in a
way.
> What don't you like specifically? Maybe worth a separate thread?
>
>> So, in my opinion, and since the subject in itself drags interest (at least from
me), I think it would be a good occasion to try and find out what direction we want to to
go regarding REST.
>> If there is interest, I can take some time and gather my thoughts/feedback as why
I think the current REST API/implementation is limited.
> Sure. My idea was just to move the generic part that makes us able to add REST
resources as components into commons and then make some proposal for REST URLs for the
Rendering module (after checking what we already have,).
>
> Thanks
> -Vincent
>
>> Jerome
>>
>> Le 05/01/13 16:27, Vincent Massol a écrit :
>>> Hi guys,
>>>
>>> Just got an idea about a talk I could give at a conference: Building a wiki,
with XWiki Rendering and AngularJS.
>>>
>>> The ideas are:
>>> * Demonstrate XWiki Rendering through a use case which is to build a simple
wiki
>>> * Use a recent/modern framework like AngularJS to draw people to the talk ;)
>>> * Talk about XWiki
>>> * Get people to use xwiki rendering and grow the developer base for xwiki
rendering
>>>
>>> I think we're close to be able to implement this but it's missing one
thing: ability to call XWiki Rendering from REST.
>>>
>>> Thus I think it would be a good idea to move our REST framework into XWiki
Commons and then in XWiki Rendering add a REST module to expose rendering operations
through REST.
>>>
>>> WDYT?
>>>
>>> Thanks
>>> -Vincent