Re: Character Sets

[starner@okstate.edu (David Starner)]

vgodunko@vipmail.ru (Vadim Godunko) wrote in message news:<665e587a.0212141453.42386f5d@posting.google.com>...
>
> How many memory required for save all data from Unicode Character
> Database? 

After stripping the converters, ICU takes up 3 MB.
<http://oss.software.ibm.com/icu/userguide/icudata.html> But that
includes a lot of locale data, and could probably be compressed more
with work.
There's no reason it would need to be paged into memory;

> What you do if this constant changed? Retest all existing
> applications?

If the constant changed, then your version of the compiler changed,
and it's certainly possible that it broke your program, constant or
not. Given a stable API, a program should not break from a change in
the Unicode data, especially as they try not to make major changes to
the data between versions.
 
> Latest public GNAT version and GCC3/GNAT both support ISO-8859-5
> encoding in identifiers. 

Which may explain why people weren't using it in earlier versions. 

> And don't known any GNAT users who use
> KOI8-R/U/B encodings outside comment, character and string literals.

The problem is, source encoding is tied into the encoding that I/O
uses.

> The best way for implement localization/internationalization support
> in Ada is define special needs annex, 

The non-BMP Unicode is not l10n/i18n - it's basic text handling just
like the rest of Unicode. As for the character data and encodings -
sure, whatever. Just so long as it's supported in some way.