[GRASSLIST:8033] Re: r.proj takes a very long time / does not complete

[Glynn Clements]

Morten Hulden wrote:

> > On 8/19/05, Jason Horn <jhorn at bu.edu> wrote:
> 
> >>I'll try your idea of projecting the 4 corners,
> >>and then computing the resolution.  One question, isn't this what
> >>r.proj would do anyway?

By default, r.proj projects the sampled boundary of the current region
to the source projection to determine the portion of the source map to
read into memory. If the actual projection operation requests cells
outside of this region, they will be read as nulls.

This behaviour can be disabled with the -n switch, in which case the
entire source map will be read into memory.

This optimisation should be acceptable for most cases, although there
are some extreme cases where projected interior points can fall
outside of the projected boundary (e.g. cylindrical to polar) where it
will fall down. Such cases tend not to occur in practice.

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