> I think this is a large part of the difference between Glenn, Erik and
> Keld: Glenn is already using Unicode (I understand); it's less clear
> to some of the other people on this list how they should go from where
> they are now to where hey would like to be in three or five years'
> time, without spending more money tham they have.
My comments origine from the experience with the migration from
7-bit national ISO 646 variants, and national EBCDICs (Danish) to
the 8-bit world of ISO 8859-1 and PC, HP, and MacIntosh character sets.
This has been a long and painfull process, starting 10 years ago, and we
are still in the midst of this process. In some areas it has just begun,
and it seems that the software is slower to convert than the hardware;
maybe the data is the slowest to convert.
The 7 to 8-bit migration should then be not that difficult compared
to an 8-bit to 16 or 32 bit world; the 7 -> 8 bit conversion does not
change the allocation of data needed as text can still be stored
in bytes. But I do not think it is a coincidense that Erik and I have
these considerations: Nordic countries like Norway and Denmark have
had widespread use of national 7-bit character sets, while it is my
impression that the rest of Europe have not had the same emphasis
on national character sets. From my experience in heterogeneous
computer environments and as a national and international internet
electronic mail provider for heterogeneous systems it is clear that
the migration in character set support is very slow - we are talking
multiple decades - and there is a need for intercommunication
between the different character set worlds during these many years,
which schemes like the SGML or the Mnemonic internet specification
But then I agree with Glenn that the best thing is to have the real thing:
full support for all of the characters of the world in all products.
10646 (implies also UNICODE) is the only candidate for this, with
its various forms.