[GRASS5] A naive opinion on how grass *should* work

[Glynn Clements]

Frank Warmerdam wrote:

> But, I would like to stress that even then I am not sure that GDAL should
> be required. I hate it when systems require me to installed a huge number
> of other support libraries before I can build them.  I think this can
> really put off new developers.

IMHO, the biggest problem with requiring GDAL is the use of C++. On
many commercial Unices, a C++ compiler may cost extra, or it may not
be particularly good (e.g. CFront). System header files (or other
system components) might not even be compatible with C++.

> Perhaps what we need is to streamline
> mechanisms whereby a developer can build selected GRASS commands from
> source and overlay them on an existing built GRASS tree.  This may be
> possible now, but I am not convinced it is easy (or I would do it!)

It's not that hard if you have a grass source tree. However, the
binary distributions only include a minimal set of libraries and
headers (libgis, libdatetime, and headers for those), and don't
include any of the components required for processing a Gmakefile.

For the future, we should support the construction of a "grass-devel"
package, which includes all of the main libraries and header files.

-- 
Glynn Clements <glynn.clements at virgin.net>