OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

ws-rx message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: Fw: [ws-rx] Issue 129: Define "discard"


Hi all,

2 points. Firstly, as there are now 2 IncompleteSequenceBehavior elements 
in the spec (one in the Offer, one in createSeqResponse), we should make 
sure that the edits to the wording apply to both.

Secondly, I think that the text below rules out a potentially useful 
implementation. You could imagine an RMD that intends to process entire 
sequences under a single unit of work. They could process each message as 
it arrives (in order, or otherwise) and only commit the transaction when 
they are happy that the sequence ended cleanly. If the sequence ends with 
a gap then they can rollback, and all the work is magically undone and the 
messages are discarded. That behaviour sounds like 
"DiscardEntireSequence", except that the messages _were_ processed - and 
then rolled back. I think this is what Bob was getting at in his note, if 
i understood him correctly.

So I think we need some wider term to describe the effects of processing, 
rather than to narrowly rule out processing. The first sentence of the 
text leads that way (talking about _behaviour exhibited_ by the RMD), but 
it seems to tighten up later.

How about the following (additions surrounded by *):

/wsrm:CreateSequenceResponse/wsrm:IncompleteSequenceBehavior
This element, if present, specifies the behavior that the RM 
Destination will exhibit upon the closure, or termination, of an 
incomplete sequence. For the purposes of defining the values used, the 
term 'discard' refers to the *behavior exhibited by the* Application 
Destination never processing 
a particular message.

A value of ?DiscardEntireSequence? indicates that the entire Sequence 
MUST be discarded by the RM Destination if the Sequence is closed, or 
terminated, when there are one or more gaps in the final 
SequenceAcknowledgement.

A value of ?DiscardFollowingFirstGap? indicates that the messages in 
the Sequence beyond the first gap MUST be discarded by the RM 
Destination if the Sequence is closed, or terminated, when there are 
one or more gaps in the final SequenceAcknowledgement.

The default value of ?NoDiscard? indicates that no acknowledged 
messages in the Sequence will be discarded by the RM Destination. 

An alternative delta would be to talk about "...behaviour equivalent to 
the..." (at the same insertion point).

Ultimately, I'm happy if we stick with the text that others have agreed 
to, so long as we fix it in both places.

Thanks,

Matt

Doug Davis <dug@us.ibm.com> wrote on 16/06/2006 02:12:59:

> 
> Ok, I can live with that text, how about: 
> 
> /wsrm:CreateSequenceResponse/wsrm:IncompleteSequenceBehavior
> This optional element, if present, specifies the behavior that the RM 
> Destination will exhibit upon the closure, or termination, of an 
> incomplete sequence. For the purposes of defining the values used, the 
> term 'discard' refers to the Application Destination never processing 
> a particular message. 
> 
> A value of ?DiscardEntireSequence? indicates that the entire Sequence 
> MUST be discarded by the RM Destination if the Sequence is closed, or 
> terminated, when there are one or more gaps in the final 
> SequenceAcknowledgement.
> 
> A value of ?DiscardFollowingFirstGap? indicates that the messages in 
> the Sequence beyond the first gap MUST be discarded by the RM 
> Destination if the Sequence is closed, or terminated, when there are 
> one or more gaps in the final SequenceAcknowledgement.
> 
> The default value of ?NoDiscard? indicates that no acknowledged 
> messages in the Sequence will be discarded by the RM Destination. 
> 
> 
> Bob? DougB? 
> 
> thanks, 
> -Doug 
> 
> 

> 
> Doug Bunting <Doug.Bunting@Sun.COM> 
> Sent by: Doug.Bunting@Sun.COM 
> 06/15/2006 05:30 PM 
> 
> To
> 
> Doug Davis/Raleigh/IBM@IBMUS 
> 
> cc
> 
> ws-rx@lists.oasis-open.org 
> 
> Subject
> 
> Re: [ws-rx] Issue 129: Define "discard"
> 
> 
> 
> 
> Well, this mostly addresses my confused objection.  I the 'discard' 
> definition still separates the RMD and AD more than necessary.  How 
about:
> 
>    ... refers to the Application Destination never processing a
>    particular message. ...
> 
> If you wish, could add "(or perhaps never receiving)" or "(or perhaps 
> never seeing)" or some such after "processing".
> 
> thanx,
>    doug
> 
> On 15/06/06 14:26, Doug Davis wrote:
> >
> > DougB,
> >   here's the latest proposal:
> >
> > /wsrm:CreateSequenceResponse/wsrm:IncompleteSequenceBehavior
> > This optional element, if present, specifies the behavior that the RM 
> > Destination will exhibit upon the closure, or termination, of an 
> > incomplete sequence. For the purposes of defining the values used, the 

> > term 'discard' refers to the RM Destination never allowing a 
> > particular message to be processed by the Application Destination.
> >
> > A value of ?DiscardEntireSequence? indicates that the entire Sequence 
> > MUST be discarded by the RM Destination if the Sequence is closed, or 
> > terminated, when there are one or more gaps in the final 
> > SequenceAcknowledgement.
> >
> > A value of ?DiscardFollowingFirstGap? indicates that the messages in 
> > the Sequence beyond the first gap MUST be discarded by the RM 
> > Destination if the Sequence is closed, or terminated, when there are 
> > one or more gaps in the final SequenceAcknowledgement.
> >
> > The default value of ?NoDiscard? indicates that no acknowledged 
> > messages in the Sequence will be discarded by the RM Destination.
> >
> >
> > thanks,
> > -DougD 



[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]