[Date Prev][Date Next][Thread Prev][Thread Next]
- Subject: announce full fledged unicode support
- From: Klaus Ripke <paul-lua@...>
- Date: Wed, 23 Feb 2005 13:01:52 +0100
here you go:
docs are in the test script.
There's only very little missing, mostly a thorough testing,
so please beat it!
Little todos: maybe fine tune grapheme.byte, format's %s
and reenable the frontier %f match.
The package now is called unicode with the four modules
ascii, latin1, utf8 and grapheme (actually closures of the same code).
All modules have the same interface as string.
ascii and latin1 are single-byte, but use the unicode character table,
and can be used as locale independent replacements for string.
In a intel linux static link, the lib adds 23K when used additionally,
or 16K when used as a replacement for string
(mostly due to the 13K tables).
find and friends are slower by about a factor of 3, but at least this
is not the order of magnitude some other unicode stuff burns,
and then there is much opportunity for tuning -- right now the
code is naive in most places, with a bias to size over speed.