[geotk] Exploring the possibility to join Apache SIS project

[Martin Desruisseaux]

Hello all

This email is for letting peoples know a possibility that we are 
exploring. In an effort to build a community, we had a recent discussion 
with members of the Apache Spatial Information System project 
(http://incubator.apache.org/sis/), which is in incubation. A small 
email exchange gave me the feeling that the design goals of this SIS 
project could be in phase with the Geotk ones.

We would like to offer them the Geotk "core" code. This would imply a 
change of license from LGPL 2.1 to the Apache license. Given that the 
Apache license is more permissive than LGPL, we think that this 
re-licensing would only be a win for all users. We need OSGeo permission 
for this re-licensing, but given that Geomatys and IRD wrote 95% of the 
GeoTools 2.6 "core" code which was the basis of "Geotidy" in 2008, and 
given that we are willing to rewrite the remaining 5% if necessary, I 
feel confident that we can find an agreement. I would like to emphase 
that it can be a win for the GeoTools project too, since they could (if 
they wish) gain a referencing library maintained by a trusted 
foundation, which could replace their current referencing module. The 
Geotk referencing module got 4 years of extensive development since we 
left GeoTools, and features capabilities which are - to my knowledge - 
unique in the open source world.

The plan is *not* to transfer Geotk to Apache SIS as one big block. The 
plan is to re-license Geotk to the Apache license, then help SIS to 
pick-up whatever code they want from Geotk at a speed that allow 
consensus. I think this would avoid any pressure, because in case of 
disagreement, the controversial code can remain on Geotk source code 
repository (or any other repository) until a consensus is found. Geotk 
could be a kind of "laboratory" for SIS where some development happen, 
and migrate to SIS only when considered mature. The migrated Geotk code 
would be progressively replaced by the corresponding SIS dependency.

Does anyone has comment or concerns with this proposed approach?

         Martin