Minutes of Face-to-face meeting 2005-09-30 P2D
The third face-to-face meeting of the Network was held at
Euroforum, Palacio de los Infantes, San Lorenzo de El Escorial, Spain, Sep.30-Oct.1, 2005. The minutes of the meeting are give below.
Participants:
KlausBanse,
NicolaCaon,
BiancaGarilli (BIG),
PrebenGrosbol (PRG),
DanielPonz,
PhilippePrugniel,
KlausReinsch,
ChristianSurace,
PerryGreenfield (PEG),
MarkoUllgren,
DougTody, AlbretoConti, MartinVogelaar, HansTerlouw, BrianGlendenning (BRG), DavidDeBonis, JoeMcMullin
Friday, September 30, 2005 (Chair: PrebenGrosbol)
14:00-14:15 Welcome
- After a short welcome by PRG, DP went through the logistic arrangement.
14:15-16:00 Discussion of revised requirements
- During the previous face-to-face meeting, June 2005, the DetailedHighLevelRequirements were discussed at length. The changes were applied to the TWiki page and the converted into a LaTeX document (v0.5) which also is available at in PDF format. The discussion was based on the latter as a few editorial changes had been applied to that but not to the TWiki page.
- PEG found that priorities should be further detailed. At this moment the requirements read more as a wish list.
- The updated list of requirements were only announced on the
on36 e-mail list. Such announcements should in the future be send to the fase list.
- DT emphasized the importance of supporting clients to run on Windows systems.
- During the discussions many detailed changes were accepted and implemented in the LaTeX version (v0.6).
16:00-16:30 Coffee break
16:30-17:00 Formal version of requirements
- The LaTeX version of the requirements document will constitute the official version. After the changes agree here have been implemented some additional text should be added to each requirement. This version should then be reviewed internally in the group. This should be done in a semi-formal manner using a written procedure. It would be good if a final discussion of comments could be done during our next fate-to-fate meeting in the late spring of 2006. After that a wide public review should be started around ADASS 2006.
17:00-17:30 Standard task and astronomical applications
- The requirements for GUI's were moved to StandardAppsAndLibs. It was felt that we should postpone the discussion on this topic until the high-level requirements was fully finished.
17:30-18:00 Review of architectural concept (
DougTody)
18:00-18:30 VO clients and publication of Web services (
DougTody)
- DT reported on recent work in related VO areas which included WEB service portals and applications as WEB services.
- In the context of the architecture, one could provide both. A WEB service interface could be implemented at the Presentation Layer while VO services could be used through special Components such as data access to VOstore.
18:30-19:00 Summary of parameter model (
DougTody)
- The ParameterModel was discussed briefly as no major changes were made since last face-to-face meeting. Everybody was encouraged to re-read it and given comments. There is already a basic proto-type implementation done in relation with the AIPS++ project.
Saturday, October 1, 2005 (Chair: PrebenGrosbol)
09:00-09:15 General information
- DP and PRG gave short update on the general arrangements.
09:15-10:00 ALMA/AIPS++ presentation (JoeMcMullin)
10:00-10:30 Discussion on Architecture and ALMA/AIPS++ concept
- There were a short discussion on different Distributed Object technologies (e.g. CORBA, ICE, DBus). PRG mentioned the C++, Python and Java containers developed by the ALMA ACS.
10:30-11:00 Coffee break
11:00-11:30 Python infrastructure (
PerryGreenfield)
- OpticonPython.pdf: Presentation on Python infrastructure
- All agreed that Python seems to be the best candidate for a high-level scripting language for a future environment.
- PRG asked concerning the differences between Numeric and numarray. PEG reported that there were ongoing efforts to join them.
11:30-12:00 PyRAF (
PerryGreenfield)
12:00-12:15 PyMIDAS (
MarkoUllgren)
12:15-12:45 Subsystems
- The TWiki page SubSystemDefinition was discussed. DT proposed some modifications such as
- Replace
Manager with Execution framework possibly including other related services.
- Add components for container-component API, general utility class libraries, Python binding and CLI, and VO related interfaces.
- With respect to CLI one would have to evaluate if Python would be sufficient or one may want IPython which provides more interactive features.
12:45-13:00 Strategy for defining interfaces
- It was agreed that an important next step, once the basic system components were defined, was to define the interfaces between them.
- Instead of going through a lengthy formal design phase a more direct definition through proto-typing was preferred.
- The container-component interface has the highest priority followed by container-framework interface. With this one could start making simple implementations of actual component to gain experience.
13:00-14:30 Lunch
14:30-15:30 Proto-typing efforts
- Sampo_Opticonf2f_011005.pdf: Presentation on SAMPO
- Some proto-typing is also expected associated to the CASE project of the AIPS++ group.
15:30-16:00 Reference implementation
- It will be important to provide reference implementations for the environment to show the community how to use it. After a set of proto-types has been made one may refine/refactor them in a reference. This concept has proved useful in for VO.
16:00-16:30 Coffee break
16:30-17:00 Towards a real environment
- Our current work in ON3.6 is aimed to define a future environment by establish requirements, layout the architectural concept and specify interfaces. This emphasizes design and does not include any efforts towards a real implementation.
- It would be natural after our design work has been completed to aim for a real implementation as a follow-up project. The current ON3.6 network is scheduled to finish around the end of 2007. Thus, we would need to work on a possible proposal for an actual implementation project already in the fall of 2006.
17:00-17:30 Long term maintenance and support
- It was recognized that this would become an issues. A model should be developed to ensure that the future environment would be supported properly and not suffer the same problems as the current generation of systems.
17:00-17:30 Priorities for next 6-9 months
- The following tasks should be done in the next 6-9 month time frame:
- Finalizing the high-level requirements that is
- revise the formal version according the our discussions
- add detailed comments to explain individual requirements
- make formal, written review before next face-to-face meeting
- Defining the general system structure and interfaces
- container-component and container-framework interfaces
- proto-type container implementation
- proto-type parameter model
17:45-18:00 Conclusions
- The two major objectives of the meeting were achieved namely
- to discuss high-level requirements and achitecutal concept within a wider group also including many who only were able to participate due to the ADASS'05 meeting (e.g. people from North America)
- to start the process of defining subsystems and interfaces between them
- If we would like to make a proposal for the actual implementation of the environment, we need to start to write an application in the fall of 2006 as it may need to be submitted in 2007. At the same time institutes should indicate if they would like to join.
- The most urgent tasks for the next 6-9 month period are:
- definition of container-component interface
- definition of container-framework interface
- proto-type implementation ofa container
- proto-type of parameter model (preliminary work already done by AIPS++ group)
- finalize formal version of high-level requirements and subject it to a formal written review
Thanks to
ChristianSurace for his notes which help me finishing the minutes.
Please participants, check these minutes, add what I forgot and correct any errors.
--
PrebenGrosbol - 19 Dec 2005
| Assigned to |
Due date |
Description |
State |
Notify |
|
| PrebenGrosbol | Sun, 30 Oct 2005 | Update high-level requirements | closed | | edit |
to top