Re: IPR tests, part 1

Le 9 apr 2013 à 21:35, Jim Barnett a écrit :

>  On tests 150, 151, I will change the language to allow for JSON in-line inside of <data> (for processors that support JSON).  If I do that, will the test then work?

Yes.

> On 351, the point of including the sendid in the SCXML Event I/O processor is that if the receiving side knows the sendid, it can include information in its response letting the original sending side know which <send> it is receiving a response to.  That information could be put in <param> in the original <send>, but only if it's not using <content>.  So including it automatically in the event is a little more flexible.

Allowing <param>s to coexist with <content> would be more flexible, and more generic. Just saying…

Besides, you want to do this only when the sendid is author-defined, so it won't be a one-time sendid, it will be the same for every event sent by that <send> element. Which means you won't in fact be able to track individual events with this sendid.

Assuming you fix that issue and we have truly unique sendids… what would you do with them? the only way to keep track of an arbitrary number of concurrent "event conversations“ on vanilla SCXML would be writing lots of targetless transitions and scripts, so why use SCXML at all instead of Node.js?

It's nice if core SCXML gets some (possibly optional) stateful way to handle many conversations, maybe by invoking a new session for each conversation. But sendid is not it nor even part of it.


> For ECMA <foreach>, the current spec language seems to me to allow the generality you want.  I guess that we can add that shallow copy is not required for immutable objects such as strings.  Is there anything else you would change?

It's better than I remembered, but still too Array-centric. It should says that "the iteration order is the order of the underlying ECMAScript structure, that goes from 0 to length-1 in the case of arrays and strings, and is unspecified for most objects".

The fact that some values may be undefined is not worth mentionning, but I suppose you wanted to make it clear that the corresponding index must not be skipped by <foreach>.

			David

Received on Wednesday, 10 April 2013 13:56:58 UTC