[GRASS5] [bug #1075] (grass) bug in r.mapcalc

[Glynn Clements]

Markus Neteler wrote:

> > > Problem #3 (or just a note): in a command mode, r.mapcalc behaves
> > > differently than a few months ago. Now you must have a command
> > > without spaces around "=" while in the past this command worked OK:
> > > 
> > > GRASS:~ > r.mapcalc flow.dx = "flow * cos(aspect)"
> > 
> > I can implement that easily enough.
> 
> Thanks Glynn. This saves us from rewriting lots of scripts and
> docs. I just tested:
> 
> GRASS:~/cvsgrass_exp > r.mapcalc test="34-43"
> test = (sub(34,43))
>    3%...
> GRASS:~/cvsgrass_exp > r.mapcalc "test=34-43"
> test = (sub(34,43))
>    3%...
> 
> nice.

Er, that's just shell behaviour; in both cases, r.mapcalc will get a
single argument.

> > However, given that you have to quote most of the command (e.g. to
> > prevent '*' being expanded by the shell), you may as well just quote
> > the whole thing, i.e.
> > 
> > 	r.mapcalc "flow.dx = flow * cos(aspect)"
> 
> Really? For my bash it works:
> 
> r.mapcalc test="34  * 43"
> test = (mul(34,43))

Same here, this is still a single argument.

The issue was regarding unquoted spaces in the command line, e.g.:

	r.mapcalc test = "34 * 34"

In this case, r.mapcalc would get three arguments, i.e.

	argc: 4
	argv[0]: "r.mapcalc"
	argv[1]: "test"
	argv[2]: "="
	argv[3]: "34 * 34"

The fix duplicates the behaviour of the old r.mapcalc which, when
given multiple arguments, concatenates them into a single string, then
parses that string.

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