Second teleconference re THREDDS integration, 28th February 2008

Attendees

• JonBlower
• PaulineMak
• JohnCaron
• EthanDavis
• RussRew

Agenda (draft)

1. Progress so far
   • Improved Javadoc for ncWMS released - any questions?
   • Does everyone understand how the GetMap operation is implemented?
2. Which version of Java-NetCDF and THREDDS should we target?
3. Any large-scale architectural issues?
   • Should we retain Spring or refactor code to remove this?
4. Important parts of ncWMS that could be integrated with THREDDS
   • WMS interface code and exception handling
   • Layer: simple 4-D data object representing a Layer in the WMS: maps to a NetCDF Variable
   • DataReader.read(): Reads data from a NetCDF dataset onto a 2-D horizontal grid efficiently
   • Regular1DCoordAxis.getIndex() and Irregular1DCoordAxis.getIndex() - efficient methods for getting the index along a 1-D coordinate axis
     • Maybe these belong in the Java NetCDF libs?
     • Both need more thorough checking
   • ...
5. How much code can we share with THREDDS-WCS?
   • Harmonize Layer and Coverage classes?
   • Harmonize data-reading code?
   • Harmonize approach to exception-handling?
6. Can we converge with GeoServer and GeoTools?
   • GeoTools contains potentially-useful methods and classes
7. Broad work plan
   • Quick and dirty integration: 1 March - 31 May 2008
   • Tidying-up and producing better version: 1 August - 30 September
   • Any problems with this?
8. Collaboration
   • Do we need a mailing list/Google Group?
   • Shared Subversion/CVS server?
   • How often do we need telecons from now on?
9. More future customers
   • MyOcean (European Marine Core Services)
10. AOB
11. Next meeting

Notes

1. Config and metadata in Javadocs need a bit more work
   • But these may be the bits that aren't needed by THREDDS
   • GetMap operation seems to be well-enough documented for now
2. Best to work on TDS version 4
   • This is the active branch of development
   • But watch out for instability
   • Also experimenting with Spring but maybe that's not relevant to this project
3. Main difference is in configuration
   • Move to one WMS endpoint per dataset
   • Don't load all of metadata in memory (scales better) - use a MetadataStore that wraps a THREDDS catalog? Anyway we probably need to read metadata from NetcdfDatasets on-demand.
   • NetcdfDatasetCache is enabled by default in THREDDS (can set size and disable)
4. (no notes)
5. TDS 3.16 has WCS1.0, TDS 4 has a WMS1.0, WCS1.0plus, WCS1.1(incomplete)
   • Hence lots of Coverage classes
6. GeoTools integration may be out of scope but good to keep on radar
7. 1st half of March Pauline's working on another project, otherwise fine.
8. Unidata will look at exposing and password-protecting their Subversion server?
   • Need to decide whether to work on trunk or a branch - preferably the former if possible
   • Will talk to person responsible and get back within a week
   • Unidata will set up a mailing list (today or tomorrow)
   • Monthly telecons from now on (see how discussions go on mailing list)
9. Hope to have stable TDS4.0 by summer 2008 - good timing for MyOcean

-- JonBlower - 27 Feb 2008