OME Madison Meeting May 3-6, 2007
Thursday attendance
|
Dundee:
|
|
|
Madison:
|
|
|
Baltimore:
|
|
Thursday AM notes: Internal discussion
Thursday PM notes: we discussed bioformats (ask for details) and walked through changes to the OME-XML standard.
took notes and posted to the OME-XML Evolution page.
Friday attendance and presentations
|
Dundee:
|
|
|
Madison:
|
Teresa (AM only)
|
|
Baltimore:
|
|
|
UCSB:
|
|
Friday AM presentations: see table above
Friday PM notes: we finished discussing changes to the OME-XML standard.
took notes.
Saturday attendance
|
Dundee:
|
|
|
Madison:
|
Eric
|
|
Baltimore:
|
|
|
UCSB:
|
|
Saturday AM notes: Josiah's Notes
Saturday PM notes: Discussed differences between OME and OMERO for some time. Primary differences in server-side functionality brought up in discussion were:
- no support for the STD portion of the OME-XML specification in OMERO
- access control and data locking explicitly performed on a row-level in OMERO rather than tacitly on a set-of-rows level in OME
- no support for batch analysis in OMERO
- no web-interfaced in OMERO
We decided to expand functionality in OMERO then migrate toolsets from OME to OMERO.
- OMERO will never dynamically upgrade for new semantic types, but facilities will be put into place for importing new STs with a server restart.
- A new "Event" subclass will be defined that will have set-of-rows access-control and locking.
- OMERO developers will survey open source workflow environments that could be integrated and have comparable functionality to OME's Analysis Engine. Many bits of functionality managed by the server side Handler classes will be migrated to thin wrappers written in the same language as the algorithm. Major portions of the server API will be migrated to several popular algorithm and "glue" languages. We hope this will allow more direct levels of debugging for algorithm developers, and provide more support for integrating manual and semi-automated analysis tools. Also, we will provide several examples of writing these thin wrappers.
- A general-purpose web infrastructure will be written by OMERO developers, likely in python using turbo-gears. This infrastructure will use automated code generation to have template-controlled search, display, and object-creation functionality for any defined object-type. OME developers will gradually migrate other functionality from the existing OME web interface.
