[GRASS-dev] Strange behaviour of some v.* in system() in R

[Roger Bivand]

I do not know where to begin to try to debug the following problem. In 
6.1.0 released, in its beta 1, but not in my last 6.1 cvs (from October 
2005, sorry, I find tracking moving GDAL, GRASS, R difficult), I see the 
following problem.

Some v.* still run from R through system()

> system("v.info bugsites")

for example. But:

> system("v.out.ascii bugsites")

now fails on parsing its arguments, and v.category and crucially v.out.ogr 
fail silently. The failing v.* also cause seg faults after using 
system("sh") or system("bash") from R to create a new shell within R that 
"knows about" GRASS.

Is anyone aware of which changes to the argument parsing for these v.*
might cause this behaviour? Is there any debugging flag that can be turned
in to see why this is happening? R's system() is the same, so the change
appears to be on the GRASS side.

I am also seeing this behaviour in the 6.2 snapshot from yesterday.  
However, it is not in the FOSS4G LiveCD GRASS 6.1 CVS version.

There is also the possibility that this is a platform-related issue, both 
machines tried with a single R release are RHEL 4, fully updated, and the 
crucial C library function called by R is system(). I've re-built R to try 
to take into account any change in the C library, but the problem remains.

Best wishes,

Roger

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no