Re: RFC 165 (v1) Allow Varibles in tr///

[Bart Lateur]

On Wed, 30 Aug 2000 10:44:55 -0400, Stephen P. Potter wrote:

>| Memory usage is irrelevant compared with speed.
>
>That's interesting.  I could swear I've seen people make a tradeoff before,
>rather than always just implementing the fastest solution.  Nothing is
>irrelevant (except resistance, if you're a Borg or a Dalek).  If we go to
>UTF16 and we start needing gig footprints for tr///, is that going to be OK
>with you?

If you have a gig footprint, speed will suffer too, and I'm not talking
about swapping, but about table initialization, which would take quite a
significant time.

A quick check at Unicode's site reveals that 3x64k characters ought to
be enough, for virtually all current cases (except for the "plane 14"
tags).
        
        <http://www.unicode.org/unicode/alloc/Pipeline.html>

-- 
        Bart.