[Inspire-data] State of INSPIRE with FOSS (long)

[Just van den Broecke]

Hi,

This mail (sorry for cross-posting) is to provoke some thinking with 
regards to INSPIRE with FOSS, its products, the INSPIRE timeschedule and 
the competition with proprietary INSPIRE solutions. Also since I have no 
immediate answers I am seeking some wisdom from you clever FOSS folks.

The issues below are somewhat colored for the Dutch situation, the 
questions I get, and focus on Download and View Services (viz WFS/WMS). 
Metadata/CSW for INSPIRE through GeoNetwork is in quite good shape IMO.

- FOSS products: deegree v3, GeoServer (GS) and recently MapServer are 
all providing INSPIRE extensions. These are however in various stages of 
development and priorities (e.g. GS WMS first and deegree WFS first), 
using very different architectures (deegree Blob storage upcoming 
hybrid, GS on-the-fly relational mapping) and different focus.
I think this diversity is good (where is Python?) but the situation is 
currently that neither is INSPIRE-ready for both WMS (1.3) and WFS 
(2.0?) with GML 3.2.1.

- INSPIRE timeline: Discovery (CSW) and View (WMS) services (Annex I) 
have to be operational by november 2011, Download (and Transformation) 
Services (WFS) by dec 2012. See also
http://inspire.jrc.ec.europa.eu/index.cfm/pageid/44

- Proprietary solutions: apart from some smaller vendors like Snowflake, 
and larger ones for FME, also ESRI has entered the INSPIRE market. 
ESRI's solution is provided by the German firm Conterra:
http://www.conterra.de/en/products/esri-software/afi.shtm
As usual, dressing is sharp, brochures glossy and maybe their solution 
actually works.

- FOSS success: in the past two years National Mapping and Cadastral 
Agencies (NMCA's) have been prototyping/testing various FOSS and 
Proprietary solutions and architectures, mainly for the entire "INSPIRE 
chain", i.e. Transformation (ETL), services (WMS/WFS/CSW). Like for 
example within ESDIN (www.esdin.eu) where the focus was on 
Transformation and WFS, about 13 out 15 NMCAs finally used FOSS tools 
like PostGIS, GDAL, deegree3 because these actually worked.
See also http://2010.foss4g.org/presentations_show.php?id=3416 ;-)

- More FOSS success: Planetek with lat/lon (deegree) have been chosen to 
build the European INSPIRE Geoportal. See 
http://www.planetek.it/eng/projects/inspire_geoportal and GeoServer 
is/has contracts with various NMCAs (UK OS and IGN FR?) to develop 
INSPIRE extensions.

- The problem: currently the situation has come that NMCAs are actually 
choosing "an INSPIRE solution" and the ESRI's of this world are entering 
the game at the same time. The smart NMCAs like IGN and OS may just 
acknowledge that FOSS will bring them long term benefits. The problem 
IMHO at this moment is that there is no complete "FOSS INSPIRE Stack". 
For example, the solution used in the inspire-foss project 
(http://code.google.com/p/inspire-foss/) with 
GDAL/XSLT/PostGIS/deegree3/GeoExt lacks a compliant WMS 1.3. GeoServer 
is also still not there for INSPIRE WMS 1.3 and lacks a robust WFS/data 
mapping solution (and as a warning to GS-devs, getting this right, 
INSPIRE-compliant and performant is much much harder than you think). 
Diversity is good, but for the NMCAs to which I have been presenting the 
situation is unclear. Maybe we are too honest in the FOSS world as many 
proprietary products may not be ready either.

- The way out: the inspire-foss project has always been geared on 
multiple INSPIRE solutions, e.g. not just deegree-only, though this is 
still a very strong solution for ETL+fast WFS serving. As said, part of 
this mail is to clarify what I think is the current situation but also 
to possibly get ideas from you. Others may think that INSPIRE is just a 
nasty heavy beast designed top-down without any rational software 
engineering constraints and will just fade away as so many 
all-encompassing standards' frameworks that had no prior proven 
implementation (remember the ISO 7-Layer OSI stack?). We may find 
INSPIRE back in another format (OSM?). Basically this mail comes back to 
what Jeroen Ticheler, Arnulf, Jo Walsh and others had already formulated 
2/3 years ago:
http://wiki.osgeo.org/wiki/INSPIRE and 
http://wiki.osgeo.org/wiki/INSPIRE_data_experiments

I think we have come a long way since then and have proven/diverse 
INSPIRE FOSS solutions though in various stages. I realize I have maybe 
been too much focused on technique/software, loosing the bigger picture, 
and that we need to have something more integrated both in organization 
(OSGeo?) and in the software ("INSPIRE Stack" with component options). 
If you got as far as here thanks for reading !

best regards,

--Just

Just van den Broecke
The Netherlands
www.justobjects.nl