Hi,
Your analysis is totally correct.
The natural and easy way with Nagare is to continue to use callbacks
with Ajax, instead of restful URL. In this aspect, the Nagare IDE is
not a good example because the server can automatically be restarted
when a source is changed and the callbacks discarded. That why restful
URL must be used in this specific case.
The first step is to see if the high-level `ajax.Update()` objects can
do the trick. YUI can do "progressive enhancement" meaning the
datasource can be a part of the HTML DOM. For example, a YUI datagrid
can be populated by a HTML table. And this HTML table can be modified
by a `ajax.Update()` action.
Then, as you have found yourself, if the `ajax.Update()` objects don't
fit the need, you can use the low-level API of Nagare:
`h.register_callback()` to create a callback and
`h.add_sessionid_in_url()` to generate the associated URL.
As a side note, I'm not totally satisfied with the exposition of these
API. There are too low-level. I'm thinking to add an easier API,
hiding for example the direct use of callbacks types, which are only
described in the source code at
http://www.nagare.org/trac/browser/tags/nagare/core/0.3.0/nagare/callbacks.py#L162
(and yes, type "1" is for callbacks with a parameter else use type
"2")