Bruno Haible via Gnulib discussion list <[email protected]> writes:

> But not for the second part:
>   $ ./gnulib-tool --find doc/parse-datetime.texi
>   parse-datetime
> This file gets copied into other packages by 'gnulib-tool', and we don't
> know the version of Texinfo (or merely texi2html) nor the @documentencoding
> used in that other package.

Isn't use of non-UTF8 in any such files mostly an oversight these days?
The texinfo manual says:

  In the default case, the input and output document encoding are
  assumed to be UTF-8, the vast global character encoding, expressed in
  8-bit bytes.  UTF-8 is compatible with 7-bit ASCII. It is recommended
  to use UTF-8 encoding for Texinfo manuals.

We could add a 'make syntax-check' rule to look for Texinfo manuals
(hueristic: *.texi and *.texinfo?) containing a @documentencoding that
isn't 'UTF-8'.

On my laptop, I only found (besides test vectors inside texinfo and old
copies of nettle/lsh) one examples in the GMP manual which uses
ISO-8859-1.  And my libidn/libidn2 which uses lower-case 'utf-8' that no
other package seems to use, but that never caused any trouble.

/Simon

Attachment: signature.asc
Description: PGP signature

Reply via email to