Re: CSV parsing and other issues (Re: LC_NUMERIC)

[Stefan Monnier]

> First, language is different from locale.  And second, we don't even
> have a buffer-local notion of language yet.  What we can support (but
> seldom if ever do) is to have buffer-local case-conversion table,
> which is a very small part of language- or locale-dependent settings.
>
> So no, buffer-local aspects in general don't affect what you have in
> mind, not yet anyway.

Worse: it's not uncommon to run code which doesn't really care about its
current-buffer, so it's not always correct to presume that the settings
of the current-buffer should be used.  We already suffer from such
problems in some corner cases with code that uses `\<` or `\_<` in
regexps matching on strings (rather than buffer content) where the
result can unexpectedly depend on the buffer which happens to
be current.


        Stefan