Suggested agenda

[Ray Osborn]

Here are some questions that I think we should be addressing during this
meeting.  In effect, they define a draft agenda for the workshop although I
haven't put down suggested discussion times yet.  Please let me know if you
think I have left out any important topics, or if you disagree with some
I've included.  

While I think we should all have some informed opinions about each of these
issues, I would like someone to lead some of these sessions with a brief
presentation (~20 min) of the issues and suggested answers to some of the
questions.  I assume that Chris will want to lead the XML session.  Since
Mark has already done a lot of work in Java, and Przemek has expressed
interest in doing some advance work on HDF5, I would like to suggest that
they deal with the Java and HDF5 sessions respectively.  Unfortunately,
Freddie doesn't think he can be with us, although he is probably best
qualified to discuss the CVS and installation issues.  If anyone would like
to volunteer to address these or any other issues, please let me know.

Topics for NeXus'2000
=====================

1) Review of current API
    What¹s working and not working?
    What is the feedback from the user community?
    Is there a need for extra features (e.g.  NXdelete)?
    Are existing compression schemes adequate?
    Do we need to improve multi-file support (appending files, linking etc)?
    What design changes might be necessary (e.g. axis definitions)?
    Is there a need for an additional utility API (cf NXdict, F90 routines)?
    How do we improve the installation procedures (e.g. configure scripts)?
2) Policy on conversion to HDF5
    Is HDF5 sufficiently mature?
    Does it run on enough platforms (e.g. VMS)?
    Does it have enough functionality and performance?
    Should we change NeXus to take advantage of its new features?
    What are the design issues in moving to HDF5 (e.g. NXlink)?
    How do we handle backward-compatibility?
    How do we address lack of third-party support (e.g. IDL, Matlab)?
3) Java support
    Is it advisable to have a NeXus version of java?
    What issues does it raise concerning the NeXus design (i.e. OOP)?
    Who should write it and maintain it?
    Should we aim to make the utilities (e.g. NXbrowse) Java applications?
4) XML
    Is it useful to have NeXus defined in XML?
    What are the advantages and disadvantages?
    How do we exploit the advantages?
    Do we need XML to NeXus converters?
    Should we define instruments in XML?
5) NeXus utilities
    What utilities currently exist (with demos)?
    Should we be providing central support for NeXus utilities?
    What utilities are these (browsers, plotters, analysis)?
    What is the best language to produce them in?
6) Future maintenance
    Is the current CVS process working OK?
    Do we need to expand the number of people contributing?
    If so, do we need levels of access (e.g. submissions/authorization)?
    Do we need hard documentation or is the web sufficient?
7) Organization of major NeXus workshop

Best regards,
Ray
-- 
Dr Ray Osborn                Tel: +1 (630) 252-9011
Materials Science Division   Fax: +1 (630) 252-7777
Argonne National Laboratory  E-mail: ROsborn at anl.gov
Argonne, IL 60439-4845