On 6/2/2011 11:18 PM, Daniel Kastl wrote:
2011/6/3 Anton Patrushev <anton.patrushev@georepublic.de
<mailto:anton.patrushev@georepublic.de>>
> There was some talk about merging with PostGIS. Is anything
happening on that front or is that dead?
Looks more like it's dead so far.
In my opinion it is not. Because it wasn't more than a first idea and
the earliest time would have been PostGIS 2.1 anyway. So I guess there
is still plenty of time and it's more up to pgRouting to do the
necessary work.
If integration into PostGIS is a goal, then changing or adding a license
isn't a topic, right?
I think the bigger question is is the purpose of integrating with PostGIS? This is not to challenge whether not not we should do this, only to get a better understanding of the reasons and to validate them.
Do we think it will get us more development? How and Why do we believe this would happen?
Do we think it would improve the product from a technical, marketing, management, funding or other point of view? How and Why do we believe this would happen?
Some other assumptions that we might have?
So there is a good president for project consoldation, in that mod_geocache and tinyOWS are small projects that are in the process of moving their provenance and management under the mapserver PSC. Both of these projects provide functionality that are commonly needed by mapserver users and smoothing the integration and coordinating releases adds some value to the mapserver users and hopefully some of the mapserver developers will get involved with these projects.
So back to your question of licensing ... As long as the license is not in conflict with the PostGIS licensing then it should not be a problem. For example today you use boost graph which has a permissive license like MIT-X and this is not a problem. If we moved all the core algorithm development into something like opengraphrouter and then wrote pgRouting wrappers to integrate it it would be the same as what you have now.
But there would be a huge advantage to the permissive licensing if you remember a while back there was some discussion with Ingres database, or we might want to integrate into mysql, sqlite, spatialite or just write standalone routing daemons. Having license flexibility makes this much easier.
Ok, so we don't want to hassle with re-licensing pgRouting, then if we can prove that opengraphrouter code can be integrated into pgrouting then we could look at what to would take to replicate the existing functionality in opengraphrouter and wrap that back into pgrouting and then continue core algorithm development there and use the pgrouting PSC as the governing body for this work.
We have a way to go to prove this, but Ashraf is willing to work on this to prove that we can do it.
-Steve
