[GRASS5] Re: [GRASS-CVS] hamish: grass/src/tcltkgrass/module s.in.ascii,1.1.1.1,1.2

[H Bowman]

> > Update of /grassrepository/grass/src/tcltkgrass/module
> > Modified Files:
> > 	s.in.ascii 
> > Log Message:
> > I didn't add the missing date= option as I don't know the correct
> > usage.
>
> it works the same as for r.timestamp. In fact a related note should
> be added to
> s.in.ascii.html and
> s.in.dbf.html

Right, that makes more sense then. But before I fix them up, some questions:


The [r|v].timestamp html pages seem to need updating too, as the example
for date range:
v.timestamp map=soils date='15 sep 1987,20 feb 1988'
  doesn't work. What does seem to work is:
v.timestamp map=soils date='15 sep 1987/20 feb 1988'

Is that correct?


What's more, [r|v].timestamp will write a bad timestamp string to the file 
without question; you get the error when you try to display the timestamp:
"WARNING: Invalid timestamp file for raster map ..."

I assume this is wrong; the following patch makes it right..
[This is against r.timestamp, but v.timestamp is almost the same, and 
  s.in.ascii & s.in.dbf's G_scan_timestamp() calls should probably get 
  the same test as well.]

Should this be applied?
Is G_fatal_error() more appropriate than exit(1)?


--- src/raster/r.timestamp/main.c.CVS	2003-03-02 15:48:12.000000000 +1300
+++ src/raster/r.timestamp/main.c	2003-03-02 16:05:30.000000000 +1300
@@ -62,8 +62,16 @@
 	G_remove_raster_timestamp(name);
 	exit(0);
     }
-
-    G_scan_timestamp (&ts, date->answer);
-    G_write_raster_timestamp(name, &ts);
-    exit(0);
+    
+    if(1 == G_scan_timestamp (&ts, date->answer))
+    {
+	G_write_raster_timestamp(name, &ts);
+	exit(0);
+    }
+    else
+    {
+	fprintf (stderr, "ERROR: Invalid timestamp\n");
+	exit(1);
+    }
+    return(1);
 }