But even using a 2 degree wide sliver still didn't stop r.los from
taking hours and hours. So I had to put r.los out of its misery.Apparently if you use small square masks, r.los might even finish sooner
or later, But the man page gives no strategies, and the program gives no
progress indications.
I've been playing around with r.los lately, namely modifying it to return horizontal compass heading, rather than vertical angle to target. This is for for a simple wave exposure index along a coastline. (No refraction or reflection of waves for now, just direction * known wave stats from that direction, iterated over 360 degrees)
As far as I can tell, the r.los in GRASS 5.0.0 totally ignores a null MASK. This is a pain.
Setting the region/window to a smaller area does speed things up of course.
Setting max_dist= just bigger than your area of interest may or may not speed things up as well. I'm not sure.
If you thing r.los is slow, try r.sunmask!