Minutes
Opening
Roll - quorate with 20 of 31 voting members
Agenda - agreed with the addition of discussion of the face to face meeting
Resolution: Minutes of 2008-04-01 approved w/o
Action Items
Action: id=2008-03-18-2 status=done Dave Booz to write up a modified version of the proposal for Issue 16 based on his thoughts expressed
during the call
Action: id=2008-04-01-1 status=pending
Action: id=2008-04-01-2 status=pending
Face to Face Meeting
tentative resolution: Face to face meeting date June 2-3
Resolution: m:Patil Face to face location will be the SAP location in Walldorf Germany
Current Issues
Koenig:
Agrees that we need some language on the server side
<Martin C>
lets not confuse thread blocking with transport blocking
<Martin C>
invoke in bpel blocks the thread
<Michael Rowley>
It blocks BPEL's concept of a thread. It does not necessarily block an operating system thread.
<Martin C>
for bpel it makes no difference
<Mike Edwards>
So the intent would primarily influence the binding
<Mike Edwards>
in SCA terms
<Martin C>
yes i beleive so
<Mike Edwards>
...there is then a separate issue of what to do for languages like Java
<Martin C>
its is very much a binding issue if you dont want to change client and server app code
<Michael Rowley>
The point is that computing resources (OS threads, sockets, etc) should not be held while waiting for the response.
<Mike Edwards>
...Dieter does propose to change the (client) code I think
<Martin C>
some computing resources have to be held by defintion;)
<Bryan Aupperle>
I think so as well. The client sees the async./callback interface.
<Michael Rowley>
Some BPEL implementations hold _no_ computing resources, except disk space.
<Martin C>
we did this all in corba...you have to separate the local threading models from the interaction model
<Martin C>
disk space is a reosource
<Michael Rowley>
Yes, that is the point of this proposal. Essentially, only disk space (which can be held for a long time) should be tied
up.
<Sanjay>
looooooooooooooooooooooooong running
<Mike Edwards>
how many "o"s is that ??
<Sanjay>
don't block on counting the 'o's. that's the point.
<BobF>
Long running is a useless concept since it is relative to the perceptions of either side of the connection as well as the
media itself
<BobF>
What is important is to make it possible an asynchronous or time-independent model
<BobF>
Or maybe better described as a time-expectation independent interaction
<Sanjay>
Bob, the question then becomes - what is short running?
<BobF>
No, the point is that it all should be async unless otherwise possible
<Mike Edwards>
It is up to the writer of the interface to decide what is "long running" - they then mark the intent in the interface and
the system behaves accordingly
<Sanjay>
+1 to MikE. And I think this proposal could be limited to just that - define the intent and its semantics.
<Michael Rowley>
If anyone uses that intent, the policy algorithm will cause an error to be generated if nothing "provides" that intent.
<Sanjay>
that is for the bindings to define
<Sanjay>
one binding may use 'async interaction form' and the other may use polling
<Mike Edwards>
This is one of these thankfully rare issues that touches a lot of the specs at once
<BobF>
Synchronous is an unnatural artifact that assumes the existence of a temporily active response channel which is transport
dependent
<BobF>
Assuming async is more robust with systems near saturation during peak load
<Michael Rowley>
I think this longRunning intent would often replace the need for callbacks. I don't see that it would replace any uses of
conversational interfaces.
<Sanjay>
sorry. I meant callbacks.
Discussion further postponed until Dave Booz is present on the call
<Bryan Aupperle>
What does GED stand for?
<Martin C>
global element declaration
<Martin C>
or definition
anish didn't quite understand why xsi:type won't occur on <property>?
Schreiber diagnostics output
[Delete this section before publishing the minutes]
statistics: Schreiber found 137 input lines
edits: Schreiber found the following text-edit commands:
edits: Line 71: Bob: s/Oo/O
command-scribe: Line 4: Bob Freund recognized
command-scribe: Schreiber detected that this section was scribed online
citation-detection-irc1: Line 13: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 15: Check for possible unrecognized nick '3. Action Items'
citation-detection-irc1: Line 17: Check for possible unrecognized nick '2008-03-18-2'
citation-detection-irc1: Line 20: Check for possible unrecognized nick '2008-04-01-1'
citation-detection-irc1: Line 22: Check for possible unrecognized nick '2008-04-01-2'
citation-detection-irc1: Line 31: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 36: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 39: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 41: Check for possible unrecognized nick 'ASSEMBLY-14'
citation-detection-irc1: Line 42: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 45: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 46: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 50: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 51: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 54: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 57: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 58: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 61: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 64: Check for possible unrecognized nick 'http'
citation-detection-irc1: Line 65: Check for possible unrecognized nick 'http'
edit-substitute: command on line 71 succeeded, changed line 70 from 'Oo' to 'O'
edit-delete: Line 71 was deleted
system: Transformer: SAXON 8.9
[End of Schreiber diagnostic output]