Re: [provision] Non-batchable operations.

[Gary P Cole <Gary.P.Cole@Sun.COM>]

Darran Rolls wrote:

> With many on the road this week, we should let this float on the list 
> for a week before drawing a close.
> For my part, I agree with your assessment. 1-6 and say no to batchable 
> bulks (now there's a mouth full :-)
>
> One follow up question.  If one defines a new capability and one knows 
> that it will be problematic to handle bulk and batch, is there a 
> universal model for specifying "no batch" and "no bulk" on the 
> capability?

No, as it stands there is not. 
- We lost the ability to say "batchable" when we dropped 
BatchableRequestType. 
- We never really *had* a way to say "mustBeSynchronous".
 
1) The former (BatchableRequestType) disappeared in some version of the 
XSD.  I'm not sure whether there was a technical requirement that it 
disappear, or whether the XSD author(s) just thought that the model was 
cleaner without it.  However, Mr. Bohren has pointed out more than once 
that:
- the XSD is *not* the specification; and
- validation in many parsers is so weak that any restriction in the XSD
   is likely not be enforced by the parser.

Jeff, why did we get rid of BatchableRequestType?  Would anything break 
if we brought it back?

2) The latter ("mustBeSynchronous") was not really on our radar in SPML 
1.0.  The choice of execution mode was always at the requestor's 
initiative, and the provider was free to say unsupportedOperation.

Synchronous and asynchronous execution modes suddenly became a lot more 
important in SPMLv2 when we realized that (the possibility that 
provisioning operations require workflow--which in turn might require 
approvals--implies that) many operations could become asynchronous *at 
the provider's initiative*.

gpc