Jarkko Hietaniemi <jhi(_at_)iki(_dot_)fi> writes:
What I wish is that the whole current locale system would curl up and
die.
As you'd agree, it's only 'encoding' part that has to die.
Oh no, there are plenty of parts in it that I wish would die :-)
(though the coupling of encoding is a major booboo). Other parts
include (some of which Nick already listed):
- that you can't really know what you get across different environments
- if there are outright errors in the locale data (a) or
you don't agree with something in the locale data (b)
or if you just want to tweak some little detail (c), there isn't
much you can't do short of submitting bug reports/fixing locale
data (which you normally cannot do in legacy Unixes, short of binary
patching)
- the whole structure of locale data bothers me. Currently, it's a
bundle
of "flat" statements. "This is the date format."
That one is my pet-hate I would like dates in YYYY/mm/dd order so
they sort right - I certainly wanted a YYYY over the century change.
"This the currency format".
How many of us deal with just one currency?
"This is the set of characters." Now we for example have
What I would like to see is something more like a tree of data,
and the possibility in runtime to inspect and override the individual
subtrees and leaves. Also, I think there should be more "shades" to
the
formats, like "long, medium, and short date formats". One should be
able
to serialize the tree of all the current locale settings (original and
overriden) and deserialize that back again.
- it's a process-wide setting
Everybody should switch to UTF-8 on Unix
Yes. UTF-8 and NFD, I would say.