On 17/3/03 2:06, "Vadim Gritsenko" <[EMAIL PROTECTED]> wrote:
Practical proposal:
* Add "flow-continuation" to the objectModel, and getter method to the ObjectModelHelper, as a replacement for "kont" environment attribute.
Read only properties of the "script" (global) ??? Or rather a function (I'd lean towards the first approach)...
In the flow script itself, "kont" is equivalent to the return value of sendPageAndWait() and "bean-dict" is equivalent to the second argument of sendPage*():
var k = sendPageAndWait("someURI", bean);
In Xsp:
<jpath:continuation> refers to the same object as "k" <jpath:value-of select='.'> refers to the same object as "bean"
In Velocity:
$continuation refers to the same object as "k" $this refers to the same object as "bean"
In XMLForm:
<xf:textbox ref="/"> ref="/" refers to the same object as "bean"
<xf:submit id="next" continuation="forward"/> id will be overwritten with the same value as k.id
* Add "flow-bean" to the objectModel, as a replacement for "bean-dict" environment attribute.
+1 :-)
This raises a question... Do we want to keep the IDL sources in line with what's in the implementation (as it is right now) or do we want to work on an abstract model in IDL and then implement it?
Maybe it would be a good idea to move everything as of the current implementation in a "cocoon.flow.javascript" module, keep that in sync with the implementation, hack around the "cocoon.flow" module, leaning towards a good design, and make those two models converge.
How does it sound? Ideas?
Agree. I wrote the original IDL to reflect exactly what was in CVS. We should move that somewhere else and start designing the proper interfaces.
Regards,
Chris
