Posted by jive
on September 14, 2006 at 4:18 PM PDT
Mostly a day of presentations, interesting contrasts between development communities and the social effects of license choice between uDig and gvSig.
FOSS4G Day Four
I was able to attend this one, but only caught snatches of material as
I reviewed my presentations for the day. Nice to see some new catalog
profiles supported, I would really like to work against a public
catalog that actually mattered to people.
I am looking forward to seeing GeoNetwork step up to OSGeo involvement
- it is one aspect of the software stack that OSGeo has been lax on,
although producing a useful understood profile may be another challange.
I had a couple of presentations, rather then talk about them much I
will simply link to them.
Getting Standards to Work (with
Working on Standards (with
The first presentation was kind of like GeoTools 101 (with working code
examples - something our website lacks) and is recommend for anyone
wanting to start using the library.
The second was a little less optimisitic (although subsequently
generated the most positive feedback) and talked about some of the
leasons learned through success and failure with GeoTools over the
years. I need to produce an update of this one capturing Simone's
experience of 15-20% synchornization overhead when working on branch
(yes developing with open source has a certaint expense to it).
Once again Paul's slides speak better then what I can manage here:
Desktop Application Framework (with
The highlight of the talk for me occured during the break where I got
to see some of the uDig developers from around the world comparing
notes (in this case it was talking about upgrading to uDig 1.1).
Paul did answer some very difficult questions on why uDig and gvSig
both exist, very similar to the questions I was asked last year as part
of the MILES
online conference thing. The answer amount to:
- Both projects started at the same time
- They have different licenses (reflecting the different
roles they wish to play)
- Differences in design goals
We do try to collaborate with gvSig on matters such as network graph
analysis or GML handling. We tend to collaborate as a matter of course
through involvement with other projects like GeoTools, GeoAPI and
Very fun to see a successful project based on the uDig &
platform. Interesting to see how things that are often abstract to me
(wms layers), leap into focus as a climate prediction is compared to
DivaGIS Highlights: workflow from GIS using a Grid Analysis tool to
generate out a "grid", and thend sending this grid off to there
existing analysis tools (based in R, etc...). Serving as a
friendly intergration platform. Fun to see custom "Style Views" that
are explicitly focused on the target data sets. Same deal with a
preference page set up to define database connections for the analysis
Someone was else talking about uDig intergration with R (Adrian
Custer), interesting to
see it already done :-). Diva GIS made use of R "serve" (once
again defined with a preference page), and constructed an R bridge.
Seems to be outed as a uDig operation, this particular one
interpolation based on precanned settings.
Next Steps for the DivaGIS project:
- migration to grid manipulation
- interoptability ICIS and BioCase by BioMoby protocol for bio
diversity data sets
- connectivity to R for geo-stats analysis
- GeoServer planned
Nice talk, with
presentation of features. A few notes scattered notes were
all I had time for ...
- deeply modular
- table joins
Interesting array of data sources:
- WMS 1.1.0 - 1.3.0
- WFS 1.0.0 + GML import/export 2.1.2
- WCS 1.0.0
- Catalog search tool ( OGC CSW 2.0, IDEC)
- Gazetter search tool (WFS 1.0.0, WFS-G 0.9, ADL)
GeoProcessing, made available as tree of "buffer" and so on (always
like to see how people end up sticking a user interface on these things
- I am more a circles and arrows kind of man myself - ie like FME).
Image georeferencing, use of control points etc... ArcIM Map and
Feature services. Image enhancing and pan-sharpening controls. Jython
scripting to simple for gui-based tools.
And what is next for gvSig ...
- initial needs are now met :-)
- network and topology analysis
- raster analysis
- 3D and temporal
- SDI authoring: OGC service plugins (generate of mapserver
Ended up talking to one
of their developers about GeoServer export as well (you can produce raw
xml files or make use of java DTO objects directly).
- raster formats
- image filters, historgrams, lookup tables
- topology construction
Network topology analysis
- optimal path, service tree
- raster analysis, reprojection, mosiacing and fushion tools,
classification / vectorization
It would be very interesting to see what gvSig is doing (I have offered
them the GeoTools graph / network package before as a point of
collaboration but so far they have been quiet). I do have some hopes on
collaboration on GML parsing technology (we will see if they attend a