[GRASS-dev] r.proj vs r.proj.seg

[Glynn Clements]

Markus Neteler wrote:

> I wonder if r.proj should be deprecated now
> in favour of r.proj.seg (to be renamed to r.proj then).

Provided that r.proj.seg has received a reasonable amount of testing,
I would be in favour of that.

In the case where the source map can fit into memory (the case that
the current r.proj handles), r.proj.seg doesn't appear to be
noticeably slower than r.proj, so there doesn't seem much point in
maintaining the old r.proj.

The only minor issue is that r.proj.seg is limited to 2GiB unless
--enable-largefile used, while r.proj is limited by available memory
(and that may well be >2GiB nowadays, although you need a 64-bit
architecture to be able to use significantly more than 2GiB for a
single process). OTOH, if you regularly deal with maps that large,
you're probably already relying upon LFS being enabled, so I doubt
that anyone is likely to actually be bitten by this issue.

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