[OT] Re: OT: Gawk match() and numbers in scientific notation

On 9 Mai, 09:44, Hermann Peifer <pei...@[EMAIL PROTECTED]
> wrote:
> Janis Papanagnou wrote:
> > pk wrote:
>
> >> And, even if it is so, that helps only for POSIX character classes;
> >> how does
> >> one answer the question "what characters match, eg, '[A-Z]' under a
giv=
en
> >> locale?".
>
> > Yes. Why does /[A-C]/ (in locale en_GB and de_DE) match the five
charact=
ers
> > 65 A
> > 66 B
> > 67 C
> > 98 b
> > 99 c
>
> I think the *.src files here have some nice human-readable definitions
> about how character ranges like [A-C] are expanded under a given locale.
>
> http://unicode.org/cldr/data/common/posix/
>
> According to the LC_COLLATE section in de_DE.UTF-8.src, [A-C] is
> expanded to the below characters, in the given order (sorry for the long
> list).

Thanks. I understand the technical reason why the results
are created the way we see them; but I think the implicit(!)
and well hidden semantics of an expression like /[A-Z]/ are
not very useful, to say the least.

I suppose it's also disputable whether character classes like
[[:alpha:]] shall match non-English letters in a en_GB locale
or e.g. accent diacritical characters in a de_DE locale. What
is the intended rationale with locales in such cases; shall
they cover multilingual data? Shall [[:lower:]] in a de_DE
locale match a (French) =E1 but not a (Greek) small omega (if
the latter is not in the codepage)?

Janis

>
> Hermann
>
> ...[snip list]