crash loading .adf files

[timaranz]

gdal 1.4.2 and 1.4.3 seg-fault when loading the attached landsurface data via python bindings. However OpenEV 1.3.7 can load it fine. I have tried using both the swig and pymod python bindings and both have the same problem. I have built from source on Windows XP using python 2.4.

[timaranz]

folder with landsurface adf files

[warmerdam]

Tim,

I have tried reproducing the problem with "gdalinfo" 1.4.3 on windows, and with gdal 1.4 python bindings on linux but everything seems to work fine. It will be a non-trivial effort for me build python bindings on win32 for 1.4.3. Would it be possible for you to check if "gdalinfo -mm" on the coverage has any problems?

It seems odd that this particular dataset would cause the python bindings problems, but not normal GDAL C++ applications.

[timaranz]

Frank,

I have downloaded the 1.4.3 windows binaries and gdal_info works fine - giving the output below. I also tried the gdal_contour command and that worked as well, so it looks like it could be a problem with our build. It seems unlikely that the bug is in the python wrapper as it occurs in both wrappers. I'll look into it further at my end.

Thanks Tim

Driver: AIG/Arc/Info Binary Grid Size is 405, 420 Coordinate System is `' Origin = (2778500.000006000000000,6163500.000000000000000) Pixel Size = (100.000000000000000,-100.000000000000000) Corner Coordinates: Upper Left ( 2778500.000, 6163500.000) Lower Left ( 2778500.000, 6121500.000) Upper Right ( 2819000.000, 6163500.000) Lower Right ( 2819000.000, 6121500.000) Center ( 2798750.000, 6142500.000) Band 1 Block=256x4 Type=Float32, ColorInterp=Undefined

Min=112.819 Max=896.425 NoData Value=-3.40282346638529e+038

[timaranz]

OK it works now. It appears I didn't have GDAL_HOME set properly, (and I was also using the system gdal14.dll instead of the one I'd just built!!).

Thanks

Tim