Leszek Gawron wrote:
The whole module idea is really great but (please correct me if I'm wrong):You would appear to be right.
<map:generate type="file" src="module:request:inputStream"/>The o.a.c.components.modules.input.RequestModule provides this:
return ObjectModelHelper.getRequest(objectModel);
getRequest returns o.a.c.environment.Request which is an interface that is not
bound to http. You cannot reach the request's body from here.
See my answer to Leszek. I must say that I find it a little bit amusing that you read the source code instead of trusting what I claimed in the anouncement and trying the implementation ;) I guess it is a behaviour that you develop when you have lived at the "bleading edge" for a long time, I most of the time do the same myself.
My answer is "lets add getInputStream to o.a.c.environment.request". After all, you've got stuff about cookies, sessions, hosts, etc, etc, which, whilst there's a level of abstraction there, it is still pretty HTTP focussed (which is fine). Adding this (and probably other similar ones) would help us in the direction of HTTP protocol independence. For example, the CLI doesn't work when some blocks are compiled in, because of their explicit HTTP references. If we redirected these via the Request interface, other environments could handle them as they wish, and we'd have greater environmental abstraction, which is a good thing.
I completely agree with that getInputStream should be added to the request interface. I gave some arguments for doing it in http://marc.theaimsgroup.com/?l=xml-cocoon-dev&m=104213670731308&w=2, there where also a discussion in http://marc.theaimsgroup.com/?t=104029502400001&r=1&w=2
Oh, and the above module:request:inputStream would work too!It allready does, or a least did some weeks ago ;)
/Daniel
