Re: CVS: herc.mirbsd.org: src
Thorsten Glaser <tg@...
2007-02-01 23:47:31 GMT
Thorsten Glaser dixit:
>Commit ID: 10045C27A5403C6060F
>the standards say these two are undefined, so why shouldn't we
>be so liberal and define them to something that actually makes
>sense? (I could even implement an unget buffer so that all but
>the first octet are eaten, but that'd slow down I guess.) Code
>in the base system is hereby allowed to use this, and all fol-
>lowing, enhancements to the standards, because MirOS is a core
>oriented "comes as a package" OS and we do not need to require
>our own tools to only use normed functions unless it's made to
>a portable package.
This is the first of a series of commits (I think) I plan to be
using the "Plan 9 way" of doing I18N. Existing APIs suck, so we
implement them for third-party software, but use our own better
APIs (specifically, I plan to implement most of Plan 9's APIs &
a few own, e.g. to go back in a MB string to the last character
or to the last column - for e.g. col(1) useful).
I even plan to drop __locale_is_utf8 and make the 'C' locale an
equivalent of the 'UTF-8' locale. This would simplify the code,
speed it up, and the 'C' locale is 7-bit anyway since we do not
support latin1 any more in recent snapshots (and nobody sent me
a complaint yet).
Benny, what do you think? Anyone else?