Re: [bug-gawk] Does gawk character classes follow this?

[arnold]

Eli Zaretskii <address@hidden> wrote:

> > From: Wolfgang Laun <address@hidden>
> > Date: Fri, 15 Feb 2019 20:40:38 +0100
> > 
> > With the character data type for any programming language you have these
> > two fundamental options: to C or not to C; less poetically: you can base
> > your implementation on the C library (GNU C library in gawk's case), or
> > roll your own, as Java does. If you go the C way, the language will be the
> > tail, and POSIX will, sooner or later, tell you how to wag.
>
> Yeah, and if you go the other way, you get Emacs.  Which is not bad,
> either (I love Emacs and use it every day), but do we really want Gawk
> to become such a large system?  And who will implement all those
> features from scratch in Gawk, anyway?  It took Emacs 12 years to get
> that right.

Indeed, the decision to go with the C library was committed to well over
10 years ago, and isn't going to change; I am definitely not going to
go down the Emacs road.

I have to point out that what gawk uses are standard C interfaces;
locale and multibyte support work on all OS's supporting them, not
just GLIBC. Examples include Solaris and the other legacy Unix systems,
for instance.

Thanks,

Arnold