Geos library

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

Geos library

Barry Rowlingson
Has anyone considered an R interface to the geos library:

http://geos.refractions.net/

GEOS (Geometry Engine - Open Source) is a C++ port of the Java Topology
Suite (JTS). As such, it aims to contain the complete functionality of
JTS in C++. This includes all the OpenGIS "Simple Features for SQL"
spatial predicate functions and spatial operators, as well as specific
JTS topology functions such as IsValid().


  - now if we can link R spatial classes to this, we can build a
complete set of spatial predicate functions and spatial operators in R....

Barry



Reply | Threaded
Open this post in threaded view
|

Geos library

Tim Keitt
Its a good thought. Not sure the best way to proceed. The OGR part of
GDAL now uses GEOS for spatial predicates, so wrapping OGR (which I
think is done?) would indirectly give you GEOS functions. The other
route is to give entry points to the geotools java libs. There seems to
be a lot of momentum behind geotools (used in JUMP and uDIG clients). Or
as you say, one could use R storage and wrap GEOS algorithms to work
directly on R data.

Tim

On Tue, 2004-10-05 at 10:33, Barry Rowlingson wrote:

> Has anyone considered an R interface to the geos library:
>
> http://geos.refractions.net/
>
> GEOS (Geometry Engine - Open Source) is a C++ port of the Java Topology
> Suite (JTS). As such, it aims to contain the complete functionality of
> JTS in C++. This includes all the OpenGIS "Simple Features for SQL"
> spatial predicate functions and spatial operators, as well as specific
> JTS topology functions such as IsValid().
>
>
>   - now if we can link R spatial classes to this, we can build a
> complete set of spatial predicate functions and spatial operators in R....
>
> Barry
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
--
Timothy H. Keitt
Section of Integrative Biology
University of Texas at Austin
http://www.keittlab.org/



Reply | Threaded
Open this post in threaded view
|

Geos library

hi_ono2001
In reply to this post by Barry Rowlingson
 How about using PostGIS which has an interface to GEOS?

 And PostGIS's spatial indices has good performance for overlay analysises
and spatial queries.

 And R has a few interfaces to PostgreSQL.


> Has anyone considered an R interface to the geos library:
>
> http://geos.refractions.net/
>
> GEOS (Geometry Engine - Open Source) is a C++ port of the Java Topology
> Suite (JTS). As such, it aims to contain the complete functionality of
> JTS in C++. This includes all the OpenGIS "Simple Features for SQL"
> spatial predicate functions and spatial operators, as well as specific
> JTS topology functions such as IsValid().
>
>
>   - now if we can link R spatial classes to this, we can build a
> complete set of spatial predicate functions and spatial operators in R....
>
> Barry



Reply | Threaded
Open this post in threaded view
|

Geos library

Edzer J. Pebesma
Hisaji ONO wrote:

> How about using PostGIS which has an interface to GEOS?
>
> And PostGIS's spatial indices has good performance for overlay analysises
>and spatial queries.
>
> And R has a few interfaces to PostgreSQL.
>  
>
There's much to say for this. Maybe the question is whether we want to
implement a GIS or interface to GIS, using R. Of course both is possible,
but who would prefer using R over existing GIS for GIS functionality?
This question needs zooming in wrt different types of functionality.

I've been looking at GEOS in the past, and found it huge and complex.
Interfacing a major part of its functionality will be a large task, and then
running all the tests...

I think one of the first goals -- and we're working on this [1,2,3] --
must be
to provide suitable spatial classes, write package support for them so
that packages can use a common spatial data base, and provide convenient
GIS interface and visualisation functions. Of course, visualisation is part
of GIS as well, but then GIS don't have lattice and ... so on.
--
Edzer

[1] http://sourceforge.net/projects/r-spatial/
[2] http://elearning.maths.lancs.ac.uk:8080/RSpatial/
[3] any help is welcome!



Reply | Threaded
Open this post in threaded view
|

Geos library

Tim Keitt
On Tue, 2004-10-05 at 12:54, Edzer J. Pebesma wrote:

> Hisaji ONO wrote:
>
> > How about using PostGIS which has an interface to GEOS?
> >
> > And PostGIS's spatial indices has good performance for overlay analysises
> >and spatial queries.
> >
> > And R has a few interfaces to PostgreSQL.
> >  
> >
> There's much to say for this. Maybe the question is whether we want to
> implement a GIS or interface to GIS, using R. Of course both is possible,
> but who would prefer using R over existing GIS for GIS functionality?
> This question needs zooming in wrt different types of functionality.
>

Yes, this is a key point. R could be converted into a pretty good GIS
platform, but how many would use it? Perhaps better is to hook R into a
client like UDIG (http://udig.refractions.net/). I'm imagining opening a
UDIG session for spatial preprocessing and then attaching an R session
with access to the UDIG data structures for model fitting. Of course,
one would still need proxy classes covering the geotools api.

T.

> I've been looking at GEOS in the past, and found it huge and complex.
> Interfacing a major part of its functionality will be a large task, and then
> running all the tests...
>
> I think one of the first goals -- and we're working on this [1,2,3] --
> must be
> to provide suitable spatial classes, write package support for them so
> that packages can use a common spatial data base, and provide convenient
> GIS interface and visualisation functions. Of course, visualisation is part
> of GIS as well, but then GIS don't have lattice and ... so on.
> --
> Edzer
>
> [1] http://sourceforge.net/projects/r-spatial/
> [2] http://elearning.maths.lancs.ac.uk:8080/RSpatial/
> [3] any help is welcome!
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
--
Timothy H. Keitt
Section of Integrative Biology
University of Texas at Austin
http://www.keittlab.org/



Reply | Threaded
Open this post in threaded view
|

Geos library

Barry Rowlingson
Timothy H. Keitt wrote:

> Yes, this is a key point. R could be converted into a pretty good GIS
> platform, but how many would use it? Perhaps better is to hook R into a
> client like UDIG (http://udig.refractions.net/). I'm imagining opening a
> UDIG session for spatial preprocessing and then attaching an R session
> with access to the UDIG data structures for model fitting. Of course,
> one would still need proxy classes covering the geotools api.

  Have you got UDIG working? I've installed (I think) the java prereqs
and other packages, including the Eclipse stuff referred to as the RCP
Runtime Library. However the docs then go:

  * Unzip the RCP Runtime Binary where you would like to install UDIG 0.3
  * Download UDIG 0.3 is and unzip it into your RCP Runtime Binary folder
  * Run udig and enjoy

  Yet the precise meaning of 'run udig' isn't given. There's a few jars
here and there, but no startup shell script....

Baz



Reply | Threaded
Open this post in threaded view
|

Geos library

Tim Keitt
Yes, its pretty rough now. I haven't run it myself, but have studied the
code a bit. I'm encouraged by the integration with eclipse and large
amount of code sharing across projects because of geotools2/geoAPI. I
have gotten JUMP (http://www.jump-project.org/) working and it looks OK.
I'm not sold though on it being swing based and having a relatively poor
architecture. My money is on UDIG in about a years time. I think we'll
see lots of momentum there. That motivated my suggestion of trying to be
compatible with geotools & UDIG.

T.

On Wed, 2004-10-06 at 09:15, Barry Rowlingson wrote:

> Timothy H. Keitt wrote:
>
> > Yes, this is a key point. R could be converted into a pretty good GIS
> > platform, but how many would use it? Perhaps better is to hook R into a
> > client like UDIG (http://udig.refractions.net/). I'm imagining opening a
> > UDIG session for spatial preprocessing and then attaching an R session
> > with access to the UDIG data structures for model fitting. Of course,
> > one would still need proxy classes covering the geotools api.
>
>   Have you got UDIG working? I've installed (I think) the java prereqs
> and other packages, including the Eclipse stuff referred to as the RCP
> Runtime Library. However the docs then go:
>
>   * Unzip the RCP Runtime Binary where you would like to install UDIG 0.3
>   * Download UDIG 0.3 is and unzip it into your RCP Runtime Binary folder
>   * Run udig and enjoy
>
>   Yet the precise meaning of 'run udig' isn't given. There's a few jars
> here and there, but no startup shell script....
>
> Baz
--
Timothy H. Keitt
Section of Integrative Biology
University of Texas at Austin
http://www.keittlab.org/