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: Re: [ws-rx] i089 joint proposal



Doug.Bunting@Sun.COM wrote on 06/01/2006 12:39:57 AM:
> Doug,
>
> A joint proposal from whom?  Which options are now off the table from
> the submitting group's perspective?

Sorry about that - it was late and I was tired :-)  This joint proposal
replaces all previous proposals put forward by BEA, IBM and/or WSO2.

> Apart from a number of typographic and grammatic errors I'll leave to
> out esteemed monkeys and / or editors (assuming for a millisecond this
> gets accepted), a few questions:


The monkeys thank you for your support

>     * Could you please explain (I know, one more time) how you see this
>       alternative SOAP binding to be in scope for the WS-RX TC?  I am
>       most concerned about the
>
>       [ <wsrm:Address> /xs:anyURI/ </wsrm:Address> ] ?
>
>     portion and its implications.  I am having trouble seeing this
>     aspect from your perspective.  Why is it our responsibility to solve
>     the one-direction-works-better transfer protocol issue for every
>     WS-RM exchange, let alone everyone?

I don't understand what you mean by "one-direction-works-better transfer
protocol issue".  If you mean anon/unreachable endpoints then I suspect
I couldn't say anything to convince you that its is a valid thing for
us to look at (aside from what has already been stated over and over)
but the TC has voted several times to head down this path.
Since you pulled out the getByAddress variant and not the getBySeqID
one let me state, yet again, why this is needed - without this form
new protocol messages (like CreateSeq) would not be able to flow.

>     * The text remains unclear about piggy-backing response messages on
>       the underlying protocol back channel.  If you and I are using HTTP
>       to communicate, may I place response messages in the HTTP response
>       to messages from the same sequence?  Can you switch to (and from)
>       using MakeConnection at any point, say when you haven't sent an
>       outbound message in the sequence for some interval?


As long as you (the server) have the proper correlation info to determine
that the backchannel is the 'right' backchannel then yes.  Normally this
would mean just the same connection that carried the request message
and the connection with the appropriate MakeConnection message.

>     * "this URI is semantically equivalent to the WS-Addressing
>       anonymous URI if a protocol-specific back-channel is available"
>       seem to answer some of the above bullet in the negative,
>       disallowing one of the more interesting use cases.  That is, (as a
>       corner case perhaps worth solving in this TC), I am interested in
>       getting additional information back to you after you are done
>       sending me messages.  I don't however know in advance when you'll
>       be done.  I'd certainly rather we didn't have to waste time with
>       extra MakeConnection round trips while you are still sending me
>       regular messages in the sequence (over HTTP).  It seems if a back
>       channel is available we must always use it.


Are you talking about sending multiple messages over the same connection
(ie. pipelining)?  If so, I don't think this proposal stops that at all,
in fact it allows it by saying things like "mechanisms such as MakeConnection".
It was realized that there may be other ways for people to determine
which backchannel is the "right one" and we didn't to stop those from
working.

>     * If we get this far: Why send the whole URI in the wsrm:Address
>       element?  Why not just provide the supposedly-sufficient UUID?


We talked about this and it would be a valid choice but people thought
it would be less complicated if we just passed-in the entire URI we
were interested in and not worry about concatenating things.

>     * Minor:
>
> [ <wsrm:Identifier> /xs:anyURI/ </wsrm:Identifier> ] ?
>
> [ <wsrm:Address> /xs:anyURI/ </wsrm:Address> ] ?
>
>     should probably be
>
> ([ <wsrm:Identifier> /xs:anyURI/ </wsrm:Identifier> ]
> [ <wsrm:Address> /xs:anyURI/ </wsrm:Address> ] ?)
> | [ <wsrm:Address> /xs:anyURI/ </wsrm:Address> ]
>
>     or some such.

Well, need to work on this :-)  There will be a trade-off between
being BNF-correct and scaring people  :-)

> thanx,
>    doug
>
> On 31/05/06 19:12, Doug Davis wrote:
> > All,
> >   the planets have aligned and we managed to come up with a joint
> > proposal for the TC to discuss on tomorrow's call.  Enjoy!


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