Hi Dom,

This with Ubuntu python too? There's no problem with GEOS 2.2.3 and
Python 2.5.2 built from source on Ubuntu.

Could you try commenting out the lines involving
GEOSCoordSeq_getOrdinate in ctypes_declarations.py? Maybe that's the
fix. We're not using the function in Shapely, so no loss.

Sean


Dominic Lowe wrote:
> Hi Sean,
> 
> I encountered this problem too on Ubuntu and had to roll back to 1.0.7 
> (probably should have reported it...)
> 
> The /usr/include/geos_c.h file does have the following in it:
> extern int GEOS_DLL GEOSCoordSeq_getOrdinate(const GEOSCoordSeq s,
>         unsigned int idx, unsigned int dim, double *val);
> 
> However I noticed that this file is actually installed with the package 
> libgeos-dev. If you remove libgeos-dev (and hence the geos_c.h file) you 
> still get the same error.
> 
> You can recreate this problem quite easily using vmplayer 
> (http://www.vmware.com/products/player/) and this virtual ubuntu: 
> http://www.gisvm.com/
> 
> Cheers,
> Dom
> 
> On Wednesday 19 November 2008 02:16:53 Sean Gillies wrote:
>> Alessandro Amici wrote:
>>> I just updated to Shapely version 1.0.10 with easy_install on two Ubuntu
>>> systems, compile is fine, but I get the following error at runtime:
>>>
>>> AttributeError: /usr/lib/libgeos_c.so.1: undefined symbol:
>>> GEOSCoordSeq_getOrdinate
>>>
>>> GEOS is version 2:
>>>
>>> ii  libgeos-c1                   2.2.3-4                      Geometry
>>> engine for Geographic Information Systems - C Library
>>>
>>> I worked around the problem downgrading to Shapely 1.0.7 (after seeing
>>> Sean post about bad GEOS 2 compatibility for the 1.0.8 release).
>>>
>>> I'd post it on trac, but there's no info on where and how to post bugs
>>> in the wiki!
>>>
>>> Keep up the good work,
>>> Alessandro
>> Alessandro,
>>
>> Can you show me the 2.2.3-4 geos_c.h file? GEOSCoordSeq_getOrdinate is
>> definitely present in the code I checked out from the GEOS repository
>>
>> http://trac.osgeo.org/geos/browser/tags/2.2.3/source/capi/geos_c.h.in
>>
>> and tests with it are passing. Perhaps there was a mix-up, with the
>> Debian package missing a function?
>>
>> Sean
>>
>> _______________________________________________
>> Community mailing list
>> [email protected]
>> http://lists.gispython.org/mailman/listinfo/community
> 
> 
> _______________________________________________
> Community mailing list
> [email protected]
> http://lists.gispython.org/mailman/listinfo/community
> 

_______________________________________________
Community mailing list
[email protected]
http://lists.gispython.org/mailman/listinfo/community

Reply via email to