[GRASS-dev] GRASS 6.1.0 release preparation

[Glynn Clements]

Markus Neteler wrote:

> > > > Are the any (detailed) instructions for how to build and package
> > > > a native QGIS + GRASS distribution from the regular GRASS CVS
> > > > repository?
> > > > 
> > > > I would love to add my archaeology modules and produce a version
> > > > of GRASS for my colleagues working on Win32 systems.
> > > > I am sure this could hit big in a science with low budget and
> > > > very specific GIS needs.
> > > 
> > > lemmi jump in here and give you the link to Radims detailed description
> > > how to build GRASS/QGIS on windows[1].
> > 
> > It would be more useful to have a guide to building a native Windows
> > version on Windows, rather than having to cross-compile.
> 
> Agreed. Do you have a suggestion for a free (freedom) compiler?
> Maybe obvious, but I am not informed.

gcc. Either from MinGW or Cygwin (Cygwin's gcc has the -mno-cygwin
switch to allow native (non-Cygwin) executables to be built).

> > Also, it would be useful to offer pre-compiled Windows binaries of
> > essential libraries (proj, GDAL etc).
> 
> These could be published as by-product of GRASS, for now as
> cross-compiled version. There is however an outstanding issue with
> proj (nad2nad is not called within MinGW, so the NAD grid import fails
> when cross-compiling).

FWIW, I've managed to compile most of GRASS with MinGW/MSys, but the
Windows stdio implementation appears to be broken with regard to
fseek() on files opened for update (read+write).

The biggest problem amongst the mandatory dependencies was the sunrpc
library (used for the XDR functions) due to the use of the BSD socket
functions.

-- 
Glynn Clements <glynn at gclements.plus.com>