Changes between Initial Version and Version 1 of ServerMtg081001

20/10/08 14:00:38 (12 years ago)



  • ServerMtg081001

    v1 v1  
     2=== NDG Software Meeting October 1, 2008 === 
     4Need to split old NDG stack into two clear stacks: a server stack, and a client stack. 
     5The former will be COWS, the latter needs a new name. ''What should we call it?'' 
     7Both clients and servers need securing. 
     9'''Proposed Resources and Timeline''' 
     11Stephen: 40 days. From the new year. 
     12Dom: 40 days. Mainly from the new year, although Dom could move the WCS as described below, before Christmas. 
     14''' Server Activities (COWS) ''' 
     16Issues and Actions 
     17 * Observe that the parameter names exposed by WMS for HiGEM are meaningless.  
     18   * This is in part because they stack currently uses gml:id rather than the parameter name.  
     19      * This is a csml issue for Dom to fix. 
     20   * It is also in part because the original NetCDF files have rather obscure parameter names in them, and since what we *already* have is not CF compliant, we need to deal with it. The NCML solution is to allow NCML to provide alternative names for actual attribute names within files.  
     21      * This could be fixed by modifying the storage descripter to allow an actual netcdf attribute name and a CSML understood name, this would need to be supported by all the code stacks. Dom Issue. Andrew to understand and approve. 
     22      * ''Dom's note:''  Actually it's simpler than that - the storage descriptor doesn't need to change. All that is necessary is that the gml id,description and name values in the csml documents change. The links between csml features and storage descriptors are managed by xlinks with  gml:ids that aren't exposed - they aren't linked by variable name.  
     24 * The existing WCS server stuff in the NDG stack needs to move to the COWS stack. Dom. 
     25 * Does the WMS stack needs to support reprojections? Stephen. See polar stereographic below. 
     26 * Configuration needs to be significantly improved so WMS and WCS can be more easily deployed starting from a global configuration file and CSML files.  
     27   * Action: Stephen and Dom. 
     28 * Need to add unit tests for the WMS and WCS servers using OWSlib.  
     29   * Support for levels and duration should be deployed and tested in both WCS and WMS. Dom and Stephen. 
     30 * WFS support in general. We now the existing geoserver work which is designed for providing CSML access to databases. The current work was limited in terms of what could be exposed (temporal duration, parameter picking, midas snaphot, scaling issues). It is unlikely that we can make significant progress on this work during this phase of the project, but future work should be properly scoped and identified and relation to C-SEKT clarified. 
     31   * Action: Stephen and Dom to produce a paper describing what was achieved and what needs doing, and what resources it would take. 
     33'''Client Stack (MILK)''' 
     35Current state is that we have the work that Calum did running on a server not much 
     36plumbed in anywhere. We need to improve this and the client stack needs to be refactored to be standalone. It should include the WMS client, WCS client, the old browse client, the new browse client and a discovery client, and it should be deployed at BADC and existing NDG sites.  
     38A key requirement for the visualisation is to support polar stereographic as well as standard map projections.  A future requirement would be to include an interface to DCIP-type WPS methods to provide more complicated plots. 
     40Presumably the polar stereographic reprojection would be done in the server stack not in openlayers per se? Need some clarity. 
     42Specific obvious actions: 
     43 * Evaluate and potentially improve what Calum has done. Make sure it's understood. Stephen. 
     44Download button for maps on the WMS page, which encode a precanned WCS call. Return NetCDF. Dom. 
     45 * Build a standalone pylons WCS client (which could be deployable via easy_install and localhost). Ideally get someone from outside the exsting team to do this. PersonX. 
     47Generic CEDA issues: 
     48 * The last brings us to a point to consider CEDA wide: we don't have a web design expert with javascript and layout skills. We need to work out who that might be. 
     49 * Do we want to consider building a webhelpers equivalent for the Yahoo Javascript library (YUI?). 
     52Software: [wiki:Software], [wiki:ServerMtg081020 Mtg 08/10/20] 
     54NDG3: [wiki:Capability], [wiki:Discovery], [wiki:Vocab], [wiki:Software], [wiki:MOLES], [wiki:Security], [wiki:Community], [wiki:Roadmap]