[byteio-wg] Notes from telcon
neil p chue hong
N.ChueHong at epcc.ed.ac.uk
Thu Jul 6 11:04:04 CDT 2006
Notes from ByteIO telcon 06/07/2006
Present:
Mark Morgan,
Neil Chue Hong,
Michel Drescher,
Amy Krause
The next call will be on 25/7/06 at 10am Eastern / 3pm UK Summer Time.
GGF18/Washington DC Planning
============================
- Mark and Michel both going along
Neil will book 1 session of 90 mins avoiding OGSA-BES, WS- Naming, JSDL,
DMIS, OGSA
Documents
=========
Documents have gone through public comment, Neil will post group replies to
documents on GGF Editor forums:
http://forge.ggf.org/sf/discussion/do/listTopics/projects.g
gf-editor/discussion.rec_byteio_specification_1_0
http://forge.ggf.org/sf/discussion/do/listTopics/projects.g
gf-editor/discussion.rec_byteio_wsrf_profile_1_0
Mark will take the editor token first, and change-track the documents.
* Protocols - would like to have GridFTP
We agreed to add a unique name and protocol for GridFTP as a MAY to the
list, as it is stable and defined in GGF.
We need to identify someone who will commit to doing a byteio
implmentation which uses GridFTP.
* WS-Addressing version
We agreed to go to a "final" version of ws-addressing
There was a change from ReferenceProperties to ReferenceParameters in an
earlier draft
We agreed to consider the version of "Web Services Addressing 1.0" dated
9th May 2006 with namespace wsa
http://www.w3.org/2005/08/addressing
and "Web Services Addressing 1.0 - SOAP Binding" version dated 9th May 2006
with name space
wsa http://www.w3.org/2005/08/addressing
wsaw http://www.w3.org/2006/02/addressing/wsdl
Documents can be found at:
http://www.w3.org/TR/2006/REC-ws-addr-core-20060509/
http://www.w3.org/TR/2006/REC-ws-addr-soap-20060509/
* WS-RF rendering - schema is invalid
We agreed to move to WS-RF v2 to fix the WS-BaseFaults v1 flaw.
* Nillable elements
We agreed to use the suggested fix from Michel.
* WSRF rendering of transfer mech
A faulty understanding of WS-RF contributed to some confusion here.
We agreed that all that is necessary is to rename the "TransferMechanisms"
ResourceProperty to "TransferMechanism" to make the singular nature clear.
* Structure of byte array passed back and forth in RandomByteIOWrite
operations
Parameters are now a bit clearer for read, but still need clarification for
write.
Basically if stride > bytes-per-clock are there holes or are bytes
concatenated?
Mark stated that he assumed that blocks were concatenated end to end when
strides>bytes-per-block and when strides<bytes-per-block then again blocks
were concatenated end-to-end again.
We will need to provide better examples of this behaviour in the documents.
NB: We will need to change the examples in the documents as these will be
affected by a lot of the proposed changes to the documents.
Interop Fests
-------------
Suggested that we broadly follow the model which WS-RF used.
Factory ops are out of scope in OGSA, so file creation is difficult.
Interop tests should give the SOAP for the operation to create a file.
Likewise the client has to get an epr, and so far eprs have always come
from server. This is again something for the bootstrap soap message.
Michel will come up with a first draft of a document which contains a
script of how every interaction goes in an Interop scenario to the level of
the bytes you expect to get back.
We should ensure we exercise striding in scenarios.
We should stick to only the simple transfer mecahnism for this interop
round.
We should have a simple sensor (e.g repeating stream of 1- 255) scenario for
streamable but only UVa has implemented the streamable interface so far.
FLE have a implementation of ByteIO using simple transfer mechanism,
without RPs.
UVa have Java and C# implementations. Java uses axis (like everyone else)
--
Neil P Chue Hong | T: [+44] (0)131 650 5957
Project Manager, EPCC | F: [+44] (0)131 650 6555
Rm 2409, JCMB, Mayfield Rd. | E: N.ChueHong at epcc.ed.ac.uk
Edinburgh, EH9 3JZ, UK | W: http://www.epcc.ed.ac.uk
BT MeetMe: http://tinyurl.com/8mwhd - Code: 14712935#
"A film is like a battleground. It's love, hate, action,
violence, death - in a word, emotion." - Sam Fuller
More information about the byteio-wg
mailing list