Re: A draft for a multibyte and multi-codepoint C string interface

[tlaronde%polynum.com@localhost]

On Mon, Apr 01, 2013 at 10:30:59AM -0400, Thor Lancelot Simon wrote:
> 
> Think of the opportunities for application misbehavior and programmer
> error that will be missed if the kernel simply treats byte strings as
> byte strings.
> 

ISTR that there is a "not imperative" programming language that
provides instructions such as "PLEASE DO" and "PLEASE DON'T".
Perhaps can we imagine a more human like behavior with an OS that
is totally unpredictable (we could add two main flavours too: male
OS and female OS; because it seems the two "species" don't usually
have the same logical rules...). Code name: april fish.

To return more to the subject, this reminds me of the time handling,
where Unix knows only one time: UTC, the representation to the user
being customizable, while Windows knew only "as is" local time,
stored in CMOS "as is", meaning that whatever depends on time
depends on user level settings: great when one has to synchronize
a network, or even simply to share a computer between OSes treating the
time differently (remember rc.conf and RTC_LOCAL_TIME or whatever?)...

Unix agnosticism should be kept, specially with such a fundamental
feature as file name, since if "everything is a file (accessed by 
hierarchical namespace)" is not universally true, it is almost true and
at the core.

And UTF-8 allows to have more than ASCII with every program able to
still work (at least for cases where it worked before) by using C char
functions.
-- 
        Thierry Laronde <tlaronde +AT+ polynum +dot+ com>
                      http://www.kergis.com/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C