From: Hans Aberg (haberg@math.su.se)
Date: Tue May 17 2005 - 07:49:23 CDT
At 13:20 +0100 2005/05/17, Peter Kirk wrote:
>>So if ASCII based software now switches to use say UTF-8 instead,
>>which does not seems to be so difficult to achieve, the 25 year
>>limit on active use may apply to that one, too.
>
>Thank you, Hans. But I would suggest that ASCII is already obsolete,
>in that almost no one still uses it. In practice they have long been
>using supersets like any of the ISO-8859 variants or UTF-8. Of
>course many users in practice restrict themselves largely to the
>original ASCII subset. But I consider a standard which has been
>supersetted in this way as no longer in active use.
When designing a computer language, it is still natural to restrict
to the ASCII subset. But in the theorem prover I am writing on, I
just fed Flex with a UTF-8 .l file, editing it in a UTF-8 editor;
there is a Unicode font (Code2001) that can display the characters I
am interested in, and there is even a keyboard layout editor Ukelele
(for Mac OS X) available. This seems to work just fine; there are no
theoretical reasons for why it should not work. Currently I admit in
the input, in parallel, ASCII and UTF-8 reserved names.
This UTF-8 is still in parallel to the ASCII names; the language is
still expressible via an ASCII only file. But if the right Unicode
aware tools are in place, I tend to think that there is no need for
admitting an ASCII only input. One might compare this with the
situation of the C/C++ trigraphs, supplied in order to admit certain
restricted character encodings, where compilers support that, but are
more likely to have them turned off as a default compile option.
So admitting more than one character set is probably just a bother.
And I may zip out the ASCII keywords at a later stage. Then ASCII is
truly out of active use.
-- Hans Aberg
This archive was generated by hypermail 2.1.5 : Tue May 17 2005 - 07:50:23 CDT