[Ubuntu] Python mapscript Segmentation fault

[Alan Boudreault]

Ok, 

Please fill a ticket in the project trac: http://trac.osgeo.org/ubuntugis/
Unfortunately, I'm afraid I'm busy with a few projects with deadlines. I hope 
to get a moment to take a look soon.

Thanks,
Alan
 
On February 16, 2011 05:50:44 am Frank Broniewski wrote:
> Hi Alan,
> 
> I did a
> 
> apt-get build-dep python-mapscript
> 
> and downloaded the latest mapserver source from mapserver.org (5.6.6).
> In the mapserver source directory:
> 
> ./configure --enable-debug --with-ogr --with-gdal with-wfs
> make
> 
> then I went into the mapscript/python directory and did a
> python setup.py build
> 
> I copied my test script into the mapscript directory, renamed
> mapscript.py (the module) to mapscript2.py just to avoid importing the
> wrong module and did my test again (with import mapscript2 as mapscript)
> - and I get some binaryish characters back on the console, so I can
> report that it works with a self compiled version. I didn't enable all
> the modules (--with-xyz) as in the packaged mapserver, I hope that isn't
> a problem.
> 
> Over on the mapserver list, someone wrote me:
> > Am 14.02.2011 11:58, schrieb Eichner, Andreas - SID-NLKM:
> > Hi,
> > 
> > it looks like the Python MapScript module is compiled against a different
> > regex libary than the Python interpreter uses. The "premature end of
> > script headers" is usually caused by a SIGSEGV where the script gets
> > killed before being able to send any headers. I would suggest to build
> > python-mapscript on your own.
> > 
> > HTH
> 
> Frank
> 
> Am 15.02.2011 17:11, schrieb Alan Boudreault:
> > hi Frank,
> > 
> > Would you be able to try to reproduce that bug using a manual
> > installation of mapserver/python binding. I would be interrested to know
> > if it's a packaging bug or not. It might be a swig binding bug.
> > 
> > Thanks,
> > Alan
> > 
> > On February 15, 2011 10:05:02 am Frank Broniewski wrote:
> >> Hi,
> >> 
> >> I have a problem with python mapscript concerning the OWSRequest.
> >> Whenever I pass an URL to setParameter and call OWSDispatch(request)
> >> mapscript crashes with a segfault. A small snippet for testing:
> >> 
> >> request = mapscript.OWSRequest()
> >> request.setParameter('LAYERS','post')
> >> request.setParameter('SLD', 'http://192.168.0.111/sld/style.sld')
> >> request.setParameter('FORMAT','image/png')
> >> request.setParameter('SERVICE','WMS')
> >> request.setParameter('VERSION','1.1.1')
> >> request.setParameter('REQUEST','GetMap')
> >> request.setParameter('STYLES','')
> >> request.setParameter('EXCEPTIONS','application/vnd.ogc.se_inimage')
> >> request.setParameter('SRS','EPSG:2169')
> >> request.setParameter('BBOX','64757.194165115,66551.046310101,89442.80583
> >> 488 5,83748.953689899') request.setParameter('WIDTH','1399')
> >> request.setParameter('HEIGHT','975')
> >> 
> >> map = mapscript.mapObj('/usr/local/mapconf/share/osm.map')
> >> map.OWSDispatch(request)
> >> 
> >> 
> >> 
> >> Map debug level is 5, console output is:
> >> brfr at E033:/usr/local/www$ python ows_test.py
> >> Content-type: image/png
> >> 
> >> Segmentation fault
> >> 
> >> 
> >> 
> >> I am using the following sources:
> >> deb http://ppa.launchpad.net/ubuntugis/ubuntugis-unstable/ubuntu hardy
> >> main deb-src
> >> http://ppa.launchpad.net/ubuntugis/ubuntugis-unstable/ubuntu hardy main
> >> 
> >> 
> >> Whenever I pass the request to the mapserver cgi all works as expected.
> >> It seems the parameter parser for python mapscript uses another parser
> >> than cgi-mapserver. Using urllib.quote, urllib.quote_plus doesn't help
> >> any further.
> >> 
> >> 
> >> 
> >> Many thanks for help and tipps
> >> 
> >> 
> >> Frank

-- 
Alan Boudreault
Mapgears
http://www.mapgears.com