[geos-devel] GEOS Usage

[Howard Butler]

On Oct 11, 2013, at 8:09 AM, Malcolm Toon <malcolm at foreflight.com> wrote:

> Good morning!
> 
> We are evaluating spatialite with GEOS extensions for inclusion in an iOS project we're working on. Because we're targeting iOS, we cannot dynamically link the library in, so we'd have to statically link it in (unmodified though).  We've reviewed the license for GEOS and due to the LGPL and its static linking clause, it appears we'd have to provide the open source community access to any artifacts needed to build our project.  This isn't something we can do.  So, we're reaching out directly to see what advice you're giving other iOS developers in the same situation.  
> 
> We are certainly excited about the ability to use GEOS and would probably be in a position to contribute to the project in the long term, but cannot provide the community any access to our project.
> 
> Please advise.
> 
> Thank you!

Malcolm,

Unlike the others, I'm a bit more sympathetic, mostly because GEOS' puritanical distain leaves a void in the ecosystem in terms of commercial contribution.  GEOS (and its parent, JTS) casts a big shadow on the space that sucks up a lot of the drive-by contribution and the majority of open source project integration due to its wide-rangine featureful- and awesome-ness.  Its license, however, causes some organizations such as yours to actively avoid contributing. 

As Sandro said, those who wish for this are welcome to start bootstrapping. GEOS isn't designed to take advantage of all of the recent trends in compiler development toward compile-time execution and optimization of data structures. A native-language design and a liberal license are good reasons to embark on development of an alternative open source geometry engine. boost::geometry is a good start on both of these, but it hasn't quite gotten over the hump.  It needs more contribution from commercial organizations such as yours that are willing to put skin in the game to make the desire for a liberally licensed, feature-complete geometry engine happen. For most, it isn't worth it, and it's easier to buy LEDA or something else and be done with it. And the void continues...

Those who've already spoken say good riddance, but I think great developers are everywhere, and the story of geometry algebra engines in 2013 is one of plumbing rather than magic sauce. That this community is hostile to contribution from organizations such as yours is an opportunity. Enough organizations need to care enough to actually act upon it, however.

Maybe you could get ESRI to release the C++ library on which https://github.com/Esri/geometry-api-java is ported from under the same Apache license :)

Howard


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://lists.osgeo.org/pipermail/geos-devel/attachments/20131014/a913d55f/attachment.pgp>