Re: [Gnu-arch-users] Re: How does arch/tla handle encodings?

[Aaron Bentley]

Tom Lord wrote:
> A mixed-encoding scheme library is the right way to do it.
>
>
>     > Anything else is half-measures or worse.
>
> False dichotomy, bro.

Fair enough.  I was thinking only about single encodings.

>     > as U+0061 followed by U+0300, then there's no such thing as random 
>     > access to characters in Unicode.  You can't change U+0061 and U+0300 
>     > into U+0062 without altering the number of codepoints in the string.
>
> Yes, C arrays definately aren't (directly) Unicode strings.   Gosh,
> there's a surprise.

I implemented UTF-32 strings at work to get random-access.  But 
nowadays, I'm less certain that random access is useful.  The mantra 
seems to be "perform operations on the string, not the codepoint".  Even 
case conversions seem to work best on strings, not codepoints.

Aaron