Re: Codespace Anxiety Redux (was: Re: Level of Unicode support required ...)

On 11/2/07, [email protected] <[email protected]> wrote:
>  the point was that setting
> such limits in the first place is not he best computing pratice.

One of the complaints about moving CPU architectures to 64 bits is
that programs almost invariably run slower. This is because in some
cases you double the size of your data, which is very expensive in
terms of cache and memory accesses. (The AMD-64 is an exception,
because it did more than double the number of bits, but PowerPC and
most other processors didn't have the glaring deficiencies to be
fixed.)

Setting limits is required for efficient processing. In this case,
setting limits helped ease transitions for older UCS-2 systems
(backwards compatibility always a good thing). If you need more than
the available number of code points, there's always ways to hack it
in, but I think it's in Unicode's interest to stop people who want to
use more than 130 thousand characters. Almost always, they want to
encode a complete mapping to some other standard that duplicates part
of Unicode, and forcing them through hoops or into their own character
encoding makes it clearer that they aren't real Unicode files.

> Agreed in the same way it took 18 years to fill up most of the BMP, it
> will take about 18 years to fill up plane one

There's little in Plane 1 that anyone, even academics, really care
about. Today, annual non-East Asian proposals are down to about a
couple dozen new Latin characters and an ancient script that nobody
actually uses (even to transcribe stuff originally written in the
script).  Much of the Roadmap for Plane 1 is empty, and what's left
often isn't really wanted by the people who work on the literature
written in the language (they transcribe into Latin). I wouldn't be
surprised if the appropriate committees were disbanded (except
possibly for an 5 or 10 year review) long before most of what's
roadmapped for Plane 1 is encoded.