r.mapcalc compile trouble - was Re: [GRASS5] I18N for GRASS: continued discussion

Scott W Mitchell smitch at eos.geog.utoronto.ca
Tue Nov 19 11:37:50 EST 2002

On Monday, November 18, 2002, at 11:33 PM, Glynn Clements wrote:

> AFAICT, the problem is that the previous rule isn't terminated with a
> semicolon. It appears that my version of bison[1] doesn't care about
> this, but yours does.

Great, that fixed it, thanks.  Now I get the same kind of error for 
r.weight, and to fix it, following your model, I had to add semicolons 
as you did to all the rules in the first ~250 lines - then it compiled 

> [1] Not only my version, but that of everyone else who has compiled
> any version of GRASS since at least -pre4; byacc (Berkeley yacc)
> doesn't care either.

Yeah, my other systems are compiling this fine too.  This one, however, 
bouteloua:/usr/local/src/grass> bison --version
bison (GNU Bison) 1.75
Written by Robert Corbett and Richard Stallman.

It's installed by the "testing" release of Debian, so perhaps it's 
"coming soon" to all sorts of releases, I dunno...  I checked, but 
don't see anything in the man page re: this issue changing, and all I 
see in the info pages is that sure enough they want a semi-colon at the 
end of each rule.

My Mac is "way back" at version 1.35, the solaris machine has 1.28.  
I've got (other) big problems compiling the current release on solaris, 
but will leave that for another day...

Thanks a lot...

Scott Mitchell

> I'm about to commit a fix to this, which essentially just adds a line
> containing an indented semicolon at line 74:
> program		: stmts			{ $$ = result = $1;		}
> 		;
> -- 
> Glynn Clements <glynn.clements at virgin.net>

More information about the grass-dev mailing list