[jsr-314-open] [ajax] Fwd: JSF Ajax Meeting Minutes

[Jim Driscoll]

FYI, these are the minutes from the Ajax meeting we held on Nov 24th.

I'm still working off this list of issues, btw.

Jim

-------- Original Message --------
Subject: [jsr-314-open] JSF Ajax Meeting Minutes
Date: Tue, 24 Nov 2009 11:03:10 -0800
From: Jim Driscoll <Jim.Driscoll at Sun.COM>
To: jsr-314-ajax-ext at sun.com


Nov 24th, JSF Ajax Meeting minutes

Attending

Nick
Alex
Jay
Ted
Andy
Roger
Jim

During the meeting, we primarily went over the following list of issues:

http://www.jboss.org/community/wiki/JSFAjaxpointsfordiscussionwithEG


First up:  PartialResponseWriter should be more intelligent about
handling new elements when inside a component.  spec bug 658.

One idea:  if a new element is started, write it out after the update is
finished writing.  This is a small spec change that could be a spec
clarification, and a fix to the impl (the impl fix may take a little
while - it's liable to be somewhat risky change).  (AI: Jim will write
up a tentative proposal)

Other idea (from Alex):  Allow for some override of PartialViewContext,
to allow for a more robust rendering options.  There were some concerns
about multi-libarary compatability, and this would be an attempt to
allow for better multi-libarary interop.  Alex tentatively agreed to
flesh out this idea, if the above proposal does not meet RichFaces' needs.

There's a request for a new event - after complete, and after each
individual dom update (success happens after *all* dom updates).  We'll
need more details (I'll add a new spec RFE for this).

Errors - in the event that 3 updates happen, if the 2nd fails, the third
should still be attempted.  This falls into the level of spec
clarification, and impl fix.  I'll file appropriate bugs.

Add request options to event payload.  Alex will take AI to elaborate.

Add begin, complete, success constants.  ex: jsf.ajax.BEGIN ?  I'll add
a spec RFE.

Add ability to abort a request before XHR is created:  from the begin
event, a token could be aquired, and an abort() function called.  Jim
will take the AI to write this up.

Response XML needs a version attribute.  I'll file a spec RFE.

Add ability to skip response processing - prehaps an abort like begin?
(Jim AI)

Support for request timeout (will file as spec RFE)   Can be quite
serious, since a network error can result in a hung app.  Would like a
config param (from the server) on a per-app basis, js api for
per-request.  Since this is so serious in effect, I'll also file a impl
RFE as well.

Form serialization should happen at the last possible second (just
before the begin event), not on submission time.  There was wide
agreement that this is a problem, and requires a spec change.  May only
be a spec clarification - I'll have to check the spec to be sure.
Regardless, this is a spec bug, and I'll file it.

Be able to get a list of all registered JSF events.  (and, while we're
at it, the ability to deregister an event)...  This will probably
require two new functions client-side.  Jim will file RFE on spec.


Address IE freeze for inline script eval - either setTimeout or
execScript.  (Jim will file impl bug).

Add ability to append <style> tag as part of the component.  (Jim will
file as impl bug)

The group agreed to meet at the JSF Summit (without Alex,
unfortunately), tentatively on Thursday.

Please reply-all with any corrections or additions to these minutes.


Jim