[GRASS-dev] Re: [GRASS-user] barin option in r.flow module

[Glynn Clements]

Markus Neteler wrote:

> >> Hello everybody,
> >> I have got some troubles using r.flow with this option..it stays forever
> >> at the:
> >> Reading input files: elevation...
> >> , barrier
> >> If I remove the barin option, all is fine.
> >> I am using the DEM from spearfish with GRASS 6.2.3 and latest 6.4svn.
> >>
> > This is how replicate using Spearfish dataset:
> > g.region rast=elevation.10m
> > v.to.rast input=streams at PERMANENT output=streamsTest use=val type=line layer=1 value=1 rows=4096 --overwrite
> > r.null map=streamsTest null=0
> > r.report units=me map=streamsTest | grep "no data"
> > r.flow elevin=elevation.10m barin=streamsTest skip=37 bound=9432 flout=testvector lgout=testraster
> 
> I can reproduce the crash (one typo fixed in above example):
> 
> 
> (gdb) r elevin=elevation.10m barin=streamTest skip=37 bound=9432 flout=testvector lgout=testraster

> Reading input files: barrier
> 
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 47222862720544 (LWP 24952)]
> 0x00000000004050b8 in read_input_files () at io.c:121
> 121                         put(ds, row, col, -1);

Initialisation of ds is conditional upon parm.dsout, which isn't set.

This should fix the immediate problem:

--- raster/r.flow/io.c	(revision 34960)
+++ raster/r.flow/io.c	(working copy)
@@ -117,7 +117,7 @@
 	    G_get_d_raster_row(fd, barc, row);
 	    for (col = 0; col < region.cols; col++) {
 		BM_set(bitbar, col, row, (barc[col] != 0));
-		if (barc[col] != 0)
+		if (parm.dsout && barc[col] != 0)
 		    put(ds, row, col, -1);
 	    }
 	}

but I don't know if there are similar cases. If the two are intimately
related, then main() will need e.g.:

	if (parm.barin && !parm.dsout)
	    G_fatal_error(_("barin= requires dsout="));

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