Quoting [Tom Goodale] (Dec 08 2004):
Hi,
so for the late reminder, we are going to have a telecon to discuss the design team meeting tomorrow (Wednesday 8th of Dec) at 8AM Pacific = 10AM Central = 16:00 UTC = 17:00 central europe.
Hartmut, Gab and I are at the GridLab workshop in Poland - we don't know if we manage to join, but will try to. In case we can't, here are some comments for the listed items:
0) Agenda bashing 1) Discussion of previous meeting (notes below) 2) Coordination of Postdam and CCT design tream fragments
Hartmut organised the meeting in Golm, Potsdam, but the participant rate is low (may be a good thing, though) - we are GridLabbers only: Hartmut, Andrei, Michael, me. Do you know of anybody else we did not hear from? We have the AG node daily from 4pm CET.
3) Coordination of CCT and other USA design team participants 4) Finalise participants lists 4) Discuss focus areas
see comments below.
5) Preliminary discussion of security issues (if sufficient interest)
Design team meeting must
o identify and review existing APIS - what is the status of the API list document Andre merzky and Gabrielle Allen are working on ?
The result of that is attached, and on http://cs.cct.lsu.edu/saga/space/Related+Grid+APIs The atteched text is obviously incomplete and not polished - thats why we have been hesitant to send it earlier. Any help in getting that done with soon would be appreciated.
o identify where we can coordinate with these APIs
From a first glance, it seems not to make sense to just create a mixture/superset of these APIs - the Look and Feel is too diverse - the same problem space is solved in different (sometimes inconsistent) ways But of course the are common paradigms. The most strong ones are: - notion of directories for files/replicas - provide remote data accesss via 'files' - good, verbose error handling (different approaches) I personally would focus on those points first, since they also seem the most well understood ones, but if I understood correctly, the F2F at SC04 already provided a (different?) list? (Can you send the notes, please?). Also, it seems useful to discuss general paradigms as Tasks, Bulk Requests, Workflow, etc as general design points.
o identify where the existing APIs don't interact well with each other
o come up and work on an 'atomic core set' of APIs.
That is something you could derive from the attached doc as it is now, that's why I do attach it. Hope that helps for an initial set of APIs.
We need to document what we find out and make sure it is widely disseminated. In particular we should have a two page summary document listing the use cases we looked at and the other APIs we reviewed.
There was worry about how to coordinate the two design teams.
IIRC, the original proposal was to work semi-independently: start the work in Europe, have a 2-4 hour overlap with the US and hand over results/proposals/documents, and then vice versa. Problem will be the handover back to Europe - we are missing an Asian proxy *sigh*... Best regards, Andre.
I am working on getting notes from the face to face out as well, however a lot of that discussion focussed on the use cases and focus areas, which I have previously summarised in a mail, and then went into a discussion of how to deal with security aspects of the API, for which we decided that we should actively engage security experts to work with us.
I hope to be able to speak to many of you on the call tomorrow.
Cheers,
Tom
-- +-----------------------------------------------------------------+ | Andre Merzky | phon: +31 - 20 - 444 - 7759 | | Vrije Universiteit Amsterdam (VU) | fax : +31 - 20 - 444 - 7653 | | Dept. of Computer Science | mail: merzky@cs.vu.nl | | De Boelelaan 1083a | www: http://www.merzky.net | | 1081 HV Amsterdam, Netherlands | | +-----------------------------------------------------------------+