[GRASSLIST:3099] Re: segmentation fault with GDAL

Scott Mitchell smitch at mac.com
Mon Apr 5 13:36:47 EDT 2004

Actually, my test file was a gif, but still no spatial ref, so I assume 
that's the same idea.  I did include the 1.2.0 GDAL lib in the binary 
distribution, soft linking to it with the generic library name 
(libgdal.so.1 in this case).

Soo... not clear to me what that means - that the bug that you fixed in 
the release branch IS the culprit here, so that the solution would be 

I agree that there's limited utility in worrying too much about 5.0.3, 
but am happy to try a simple fix if there is one.

On 5-Apr-04, at 13:15, Paul Kelly wrote:

> On Mon, 5 Apr 2004, Scott W Mitchell wrote:
>>> I meant the GRASS binaries: which version and where you downloaded 
>>> them
>>> from? I notice the 5.0.3 Linux binaries on the website are quite 
>>> recent so
>>> if the fault occured with them then it is possible that whoever 
>>> created
>>> them didn't use the --with-gdal configure option and they should be
>>> re-done.
>> Not that this helps much with the current problem (maybe), but I'm 
>> pretty
>> sure that I made those binaries.  I just checked, and (1) I used the
>> --with-gdal switch, (2) r.in.gdal is working on my machine, and (3) I 
>> used
>> GDAL 1.2.0 (CVS) as of March 5.
> Then there is a good chance that the other recently-found bug in 
> r.in.gdal
> will have been causing the problem. 5.0.3 still contains this bug, but 
> I
> copied the fix to the release branch yesterday:
> http://grass.itc.it/pipermail/grass-commit/2004-April/011397.html
> As the bug was something to do with the unit name, it is perhaps more
> likely to occur on a non-georeferenced file, like a JPEG. Maybe this 
> was
> the problem... did you try it with a JPEG image on your machine.
> If that was the problem, we could always do new binaries using the 
> latest
> release branch code, but then it wouldn't be true 5.0.3.
> Really shouldn't bother with 5.0.3 any more though :)
>> Would it cause problems if the binary GDAL build involved is a 
>> different
>> version than the one I built with?
> But did you not include the gdal library in the GRASS binary 
> distribution,
> so it is the same one as you compiled with? I suppose if there is more
> than one GDAL installed on the system the potential for that is there
> though.
> Paul
Scott W. Mitchell		Scott_Mitchell at carleton.ca
Department of Geography and Environmental Studies
Carleton University, B349 Loeb Building (Office A209)
1125 Colonel By Drive, Ottawa, ON Canada K1S 5B6
+1-613-520-2600 ext 2695 Fax: 1-613-520-4301

More information about the grass-user mailing list