[GRASSLIST:2521] Re: Was: r.in.gdal and libgdal.1.1.dylib

[Paul Kelly]

Hello
On Sun, 8 Feb 2004, Christof Bigler wrote:

> The import works. Apparently, the problem was the missing
> libgdal.1.1.dylib in /usr/local/lib. I had a higher libgdal.1.1.9.dylib
> version which had been installed with the GRASS binary distribution for
> OS X. I installed then the libgdal.1.1.dylib from
> http://gdal.velocet.ca/projects/grass.
>
> Another thing: Is it possible to change the man pages for r.in.gdal? It
> took me literally hours to find out that the input name should be the
> directory and not the file itself. I found this information in a mail
> to the list from, I think, 2001, and also in the Open Source GIS book
> from Neteler & Mitasova.

Are you saying that the bare filename worked with gdalinfo but not r.in.gdal?
If so, that is probably because gdalinfo is using the latest gdal library
which perhaps has that bug fixed, but because you downloaded a really old
version of libgdal to use with r.in.gdal, the bug was not fixed in that
version? (perhaps).

I have a feeling there is much more to this situation than meets the eye.
Where did you get your OSX GRASS binary distribtion from? Perhaps it had
some specially hacked gdalbridge code that was specifically looking for
the filename libgdal.1.1.dylib? Normally it only looks for libgdal.1.1.so
so would fail on OSX, which is a good thing because you need to compile
GRASS with the --with-gdal option anyway (avoiding use of the gdalbridge
code) if you want to use the most up-to-date gdal libraries.

But in your case you were content to use 1.1.4 as indicated on Frank's
download page so it was OK. Did simply renaming the libgdal.1.1.9.dylib to
libgdal.1.1.dylib not work? (I suspect it might have started to work
byt lead to a segmentation fault from r.in.gdal.)

The simple answer to most of the GRASS/GDAL problems is to install GDAL
first and then compile GRASS using the --with-gdal configure option.
Because the problem appears to be so complicated there are lots of other
obscure and quite random solutions that 'sort of' work, as you have
discovered.

I hope these thoughts might shed a little more light on the problem but it
isn't very important.

Paul K