Pages

Monday, 25 February 2013

OGC and lowering the bar of implementation

I have a couple blog posts up on the LISAsoft website of an OGC nature:

  • OWS 9 and WMS 1.3.0 Adoption: OWS9 is now complete (with an impressive $2.65 million from sponsors answered with $5 million in-kind from industry). For my part I was able to look at Daniel Morissette's excellent advise not to upgrade to WMS 1.3.0 and start attacking the source of the problem - restoring trust in WMS clients.
  • OGC Struggling to Reach out to Implementors: Last week saw a great example of tough love for the GeoPackage standard. Carl Reed is doing the right thing and taking work like OWS Context to the standards@osgeo.org list for review. I had a look and while I am impressed with the level of "reuse" shown, I am concerned with the amount of work (and risk of mistakes) being fostered on client authors.
The common theme here is that OGC takes care of services pretty darn well (this makes sense as it is often what sponsors of events such as OWS9 are willing to pay for). In reaching for a larger audience of client applications the standards body really needs to step up its game and work hard on communication, simplification and lowering the bar of implementation.

I am heading to the AusNZ OGC meeting this week to discuss OWS10 - it will be interesting to see where the priorities are.

1 comment:

Brad Hards said...

In defence of the GeoPackage draft standard, much of the "its all too much" and "its too long too read" ignores that its how OGC does business.

That is partly driven by trying to get ISO standards format compatibility. I'd personally be happy for that to go away, but there is a lot of customer and big business demand for that kind of stuff.

The comparison between mbtiles and geopackage isn't all that reasonable, because geopackage supports a lot more than one tile set, web-mercator, power-2 zoom level. If all you need is mbtiles, then use it. If you want to hold vector data (e.g. POI, boundaries, lines), photos, multiple tile sets, polar data, or anything else that might come out of a WMS/WCS/WFS/WMTS server, then geopackage is probably a better bet.

I'd like to remove stuff than no-one will use. However non-specific "make it simpler" isn't going to help. If there is stuff that could come out to make it simpler, please help us to identify it.