Re: [Bug-gnulib] what encodings should be used in gnulib source files?

[Bruno Haible]

Paul Eggert wrote:
> I noticed that a couple of gnulib files (diacrit.c, strverscmp.c) use
> ISO 8859-1 encodings.  For source files, some gnulib-using packages
> seem to prefer UTF-8, others iso-2022-7bit, others Ascii.

I'd suggest to convert them to UTF-8 when there is need. UTF-8 is the
encoding of choice when non-ASCII characters are needed.

> Any objections if I turn these files into Ascii?  The only usages here
> are contributors' names.

I object: maintainers (especially one who has done so much for i18n as
François Pinard) would likely be hurt if their name is incorrectly spelled.

> Also, I notice that lib/ChangeLog uses utf-8 and m4/ChangeLog uses
> ISO-8859-1 (I think -- it confuses Emacs, whatever it is).  OK to
> standardize on utf-8 here?

Yes. m4/ChangeLog was actually alrady in UTF-8, with a few uses of U+FFFD
(the "REPLACEMENT CHARACTER"). I fixed it now.

> I would put a "coding:" entry at the bottom, for Emacs's sake.

Fine with me.

> The other possibilities would be ASCII and/or iso-2022-7bit.

ASCII cannot encode François Pinard's name, and iso-2022-7bit is not
a standardized encoding, with the consequence that glibc has no converter
for it.

> If some people use pre-21.3 versions of Emacs then utf-8 wouldn't be
> that suitable: ascii's probably the way to go.

You can just assume that people use emacs 21.3 or xemacs 21.4 or vim 6.0.

Bruno