Jarkko:
>>
>> It might be more useful if the default for the non-utf-8 characters
>> were the system-defined default character encoding of the process ...
>
>I can understand the request but the problem is that for this to work
>the legacy eight-bit mappings must first be implemented. ...
I understand. But it might be nice to define the API in terms of converting from the
system character encoding, even if the API only supports ISO Latin-1 in the first
release.
>> Perhaps it would be difficult to use ICU as a utility ...
>... The biggest problem is that
>the ICU will not be everywhere.
>
Since it would be invisible outside of Perl internals, could it be an optional
component? In other words, you don't need it unless you care about character encodings
outside of a easy-to-implement base set.
But in any case, for the near term it might be useful to use their character set name
alias file "convrtrs.txt" and their mapping table plain-text 'source' files (*.ucm).
"convrtrs.txt" is a useful alias list and is in an easy-to-process plain-text format.
The mapping table files are useful because the selection is much larger than the
selection on unicode.org and they are all in the same format, unlike the ones on
unicode.org. Also, I believe them to be of high quality - at least as good as the ones
on unicode.org.
=Ed
--== Sent via Deja.com http://www.deja.com/ ==--
Before you buy.
