From: Kenneth Whistler (kenw@sybase.com)
Date: Thu Apr 29 2004 - 14:51:45 EDT
Peter Kirk continued:
> >>... If you want to abolish the PUA,
> >>
> >>
> >
> >I do not, and have no idea why you are imputing that to me, except,
> >I suppose, because I have been speaking against your position.
...
> >
> You did write the following, in response to Dean:
>
> >>Currently, I view the PUA as practically a wasteland, unusable for even
> >>for the most basic research work.
> >>
> >>
> >
> >A wise decision, all in all.
> >
> >
>
> Maybe "abolish" was not the right word as presumable aboition would
> breach stability agreements. But you do seem to want to "deprecate" it,
> or at least to strongly discourage its use,
Not really.
That particular response to Dean was tongue-in-cheek, in case you
missed it. I suppose I should just refrain from attempts at humor
in responses, as they seem to get interpreted by some people as
literal statements.
So here is that comment, restated and explicated in full.
Dean Snyder has come to the conclusion that use of the PUA is
problematical for the kinds of purposes he envisioned putting it
to, because support for custom properties is nonexistent in
easily available software, because support for display is spotty,
because ensuring that other people share the same conventions
as he might want to define is difficult, because he
or the other scholars he might want to work with may lack the
expertise and/or funds to accomplish the custom programming they
might need to do to use a PUA encoding effectively, and because
the time involved in getting the PUA to work for his research
might better be spent doing something else, including working
through an actual standardization proposal for the script(s)
he might be interested in.
I *agree* with those assessments.
Using the PUA for encoding some scholarly text should be a matter
of *last* resort, when no other option is really available. And
then the person who resorts to that should be prepared to use
the PUA with minimal generic support and with plans to export/convert
to other formats for particular kinds of processing and/or
rendering that they may require. And they should be prepared
to get their hands dirty with some programming to accomplish
what they need to do.
The PUA is basically a "wasteland", as Dean indicated. It is a
range of 137,468 code points that are provided for people to
do with what they will. Caveat emptor. It is silly to expect
that what one person might decide to do with them will not run
into problems with what somebody else might decide to do with
them -- after all, their interpretation is *deliberatly* not
standardized -- that's the nature of "private use".
> except apparently for purely
> internal use within one company, which is outside the scope of the standard.
I think you may have some serious misapprehensions about what
"use within one company" means these days in software
development. Software these days is massive, distributed,
and modular. The "private agreement" I have on some PUA
character's use may be shared publicly with some other group
developing some other piece of software. It may involve harmonizing
a decision on private use with private use defined by some *other*
company's software, without that decision ever rising to the level
of end-user visibility. Such issues are *not* outside the
scope of the standard. We depend on a common understanding of
what PUA code points are and how they might be used, as defined
by the standard. The particular intepretation that I might
use in a particular piece of software *is* outside the scope
of the standard, but that isn't correlated with whether that
usage is within one company or not.
As it happens, one of the major uses to which PUA code points are
actually put in major software today is in cross-mapping East
Asian code pages. And in such cases there are implicit "private
agreements" between the company that might define such a
cross-mapping involving an East Asian code page and another
company that may use such a cross-mapping or which may have to
provide a conversion table which emulates the same cross-mapping.
That is a *very* common situation in software development today.
--Ken
This archive was generated by hypermail 2.1.5 : Thu Apr 29 2004 - 16:01:11 EDT