[OSGeo-Discuss] Software Standards - The Ugly Truth

[SteveC]

On 23 Jun 2007, at 21:55, Cameron Shorter wrote:

> Landon,
> You have made some accusations about poor OGC standards and the  
> financial barrier to joining the OGC.
>
> 1. I'm with you on the financial barrier to joining the OGC. In  
> particular, I don't like OS developers being locked out of OGC's  
> testbeds (which is where the OGC tries out proposed standards).
> Consequently, standards are being built outside of the OGC and  
> "tested in the wild" before they are adopted by the OGC. In  
> particular, the GeoRSS standard was developed outside the OGC then  
> adopted. There is currently work on a Tiled WMS spec and a GeoJSON  
> spec which I expect the OGC will adopt soon. These specs are being  
> built by Open Source developers.
> To me, this shows limitations in the OGC's membership criteria  
> which is locking out Open Source developers.
>
> 2. You make accusations that the standards don't work, and that you  
> have something better. I'd like to hear examples of this before I'd  
> be ready to agree with you.

openstreetmap. We don't care about standards (in this OGC context),  
we care about making maps.

Amazingly if you talk to Real Map Data Companies our data model and  
API are very close to what they do, but nothing like...

> What standards don't work?

WFS(-T)

> What standards stifle innovation?

WFS(-T)

If you look at the things that do work (and I'll define work as  
things that large numbers of people use), they're either the simplest  
things you could possibly do (viz georss, all you're doing is adding  
some numbers (lat/lng) to it before the GML nuts got a handle on it)  
or some people wrote something in a black box and everyone adopted it  
because it just worked (googles tile spec) or you had no choice to be  
with the cool kids (KML).

WMS vs. googles tiles is, I think, a great example of what the  
original poster was trying to get at. On the one hand you have a  
Specification which does everything, for everyone all the time in any  
way, and on the other you have something which is approximately  
trivial, does one thing which happens to be what 5 9's of the  
population want. Tiled WMS is this big band aid. The only real  
innovation on top of gtiles was MSFTs quadtree and even that was a no- 
brainer.

Standards, like ideas, are cheap. It's the implementation that's  
expensive.

have fun,

SteveC | steve at asklater.com | http://www.asklater.com/steve/


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/discuss/attachments/20070624/488e1cad/attachment-0001.html>