OGSA Teleconference - 19 January 2005 ===================================== * Participants Abdeslem Djaoui (RL) Ian Foster (ANL) Andrew Grimshaw (UVa) Hiro Kishimoto (Fujitsu) Tom Maguire (IBM) Mark Morgan (UVa) Takuya Mori (NEC, ANL) Steve Newhouse (OMII) Andreas Savva (Fujitsu) Dave Snelling (Fujitsu) Latha Srinivasan (HP) Ravi Subramaniam (Intel) Jem Treadwell (HP) Minutes: Andreas Savva * Jan 10 and 12 minutes approved with no changes * F2F logistics - Place is Westin at O'Hare airport. - Logistics are not completely worked through yet. Action: Andrew to - Book conference room for the OGSA-WG meeting (3 days: Feb 14-16) - [And an additional one day for the non-OGSA meeting] - And let the group know of the room rate (for staying at the hotel) Action: Hiro will update F2F agenda, etc to reflect discussions and re-post. * GGF13 preparation - Hiro added InfoD to the RM session (now a shared session) and submitted the OGSA scheduling requests to Stacey. - There is a plan to do an EM-related BoF. Stacey should know about it but it is not the job of the OGSA-WG to tell her---the proposed-chairs should do it. Preparation has not progressed to that point yet so Andrew as Area Director will take care of it. Action: Andrew as Area director to send an email to the directors of the relevant Area (SRM: Bill, Stephen) to let them know of the proposed BoF; also to Stacey to reserve a slot. - Andrew mentioned that on the Data call there was a discussion for a similar BoF for Data. * Basic profile review Reference: https://forge.gridforum.org/projects/ogsa-wg/document/draft-ggf-ogsa-basic-profile/en/2 ** Introduction/Front matter - Many changes since the last review. Because the document was still taking form these were not change tracked. The authors will do change tracking for future versions. - Ian re-wrote the Introduction. - Security section by Dave S. - Basic security addition is not reflected in the front matter yet. - Tom fleshed out many compliance statements. - Agreed to track issues via trackers. All comments should be deleted for the next version: either address in the text or move to the tracker, if substantial enough. Action: Andreas to create issues tracker. [Done. https://forge.gridforum.org/tracker/index.php?func=browse&group_id=42&atid=780] - Submit issues to the tracker directly or to the ogsa-wg list (and someone (probably Andreas) will put in the tracker.) The tracker is set to send mail to the list on addition of new artifacts only. ** Profile Conformance - Conformance targets need update. Probably some of these will be deleted. - Conformance Claim URI: Not defined yet. Reuse the WS-I approach. - There are two different types of statements marked with different prefix letters: {R, E} - [R]: Requirements statement. These are fairly consise. They may restrict but not extend an existing specification. (They should not be used to try and write new specifications.) - [E]: Extensibility points. Call out points where extensions may be made. E.g., where open content might be required. These highlight areas where behaviour may vary. (In a sense these are "warning flags.") ** Addressing - Using submission drafts as base; not trying to actively follow the latest ones. - R1002: Highlights an issue with the current spec.; the use of policy elements. Needs follow up. - R1003: It might be the case that the URL contains the ServiceName but this is not required. This statement is trying to codify the desired behaviour. *** 3.1.5 Resilient References - The intention is to describe ogsi-similar renewable references. - Focus is on capturing the renewable reference aspect and NOT on how to register a name so that it can be resolved. - 'Speculative' addition of rns:resolve since the RNS spec. may provide this operation. - There was a proposal for standardization of renewable references (RR). But it seems to be falling between the cracks of various groups: W3C Addressing group has placed it out of scope; WSRF does not seem to be the right place for it. *** 3.1.6 Reference properties - R1005: This statement will probably disappear into a tracker artifact. It is not easy to test or check. - In terms of terminology note that the profile deals with a particular message exchange. So terms like SENDER and CONSUMER are used instead of CLIENT/SERVER since the latter would describe a relation beyond a specific message exchange. - "Encoding scheme" might be work for a higher level profile. ("Correlateable properties?") *** 3.2 Addressing MIH Includes just boilerplate text for now. Need to think about it and revisit as WS-Addressing progresses. (It is expected that by March the W3C Addressing WG wil have a committee draft for this.) ** Resource embodiment - One major issue here is how far Data requirements can be accomodated while promoting interoperability. It might be good idea to schedule a joint-session with DAIS on this topic at GGF13. *** Resource properties **** 5.1.1 Property names - Qnames to point to the type of a resource property. And assume that the entity retrieving that information knows or can find out what they mean. - (There was a side-argument on why type information cannot be returned in general. Many implications if dynamic resource property elements are allowed and how they should be supported. Might want to record this as an issue but it is not likely to be something that can be solved in this profile.) - (DTDs were also mentioned as an alternative.) **** 5.1.2 Metadata descriptor WSRF has made this optional so it should not be made mandatory here. But it is probably a good idea to make it a separate section to draw out the relevant issues. *** 5.3 Query resource properties - (Issue) Requires XPath 1.0; but there is no standard serialization for an XPath response; so it ends up being implementation dependent. (XPath 2.0 on the other hand defines the response as well.) *** 5.4 ResourceProperty change notification - Might not want to require Topics. (Make no statements about their usage.) - Then they could not be used in a manner compliant to this profile but people could still experiment with them. And the profile could be kept simple for now. - Adding them in a later version would not affect other parts. *** 5.5 Set resource property - Another problematic function. Ordering issues are one typical issue (but there are other related issues discussed in the WSRF TC). - Revisit after checking with likely users: WSDM and possibly GT - Proposal at the moment is to give them the same treatement as Topics: say nothing about them and leave the field open for experimentation. ** 6 Resource lifetime - Not sure how far to mandate support for immediate destruction. - Proposal: A resource that can be created by message exchange should support immediate destruction - Might want to require response by all (even if a resource was statically created). - The semantics are defined by WSRF; not in the profile. In particular there was some discussion on what a destruction message means (in particular a successful reply). - A long discussion on whether to allow other behaviours, e.g., not mandate that a resource has to support this behaviour. It seemed to be due to a misunderstanding and it was agreed that the discussion should be taken offline; revisit in a later call if still an issue. *** 6.4 Notification of pending destruction - Might be something to propose to the WSRF group since it requires functionality not currently specified. ** Security - A first attempt to do a security profile segment. It uses the WS-I basic security profile, HTTP over TLS, and TLS - Note terminology: SENDER/RECEIVER not CLIENT/SERVER since the statements are with respect to message exchange roles as noted earlier. - It is transport-based security: not necessarily session-based--- it could be message-based. This seemed not to come out clearly enough from the text. - Contrast with WS-I basic security profile - WS-I profile statements are that "if something needs to be secured this is how it should be done" while this profile makes statemens on what functionality is required - Restricted authentication to just x509 certs. (And therefore disallowing username/passwd authentication.) This should probably be stated more clearly. ** Other business - Tom has the pen and will produce the next version of the basic profile document.