> From: Bruno Haible <[EMAIL PROTECTED]>
> Date: Sat, 6 Oct 2007 05:20:03 +0200
>
> When I try to format the gettext.info file, I get warnings like this:
>
> $ cd gettext-tools/doc
> $ LANG= LC_MESSAGES= LC_ALL= LANGUAGE= makeinfo -I ../../gettext-runtime/doc
> --no-split -I . -o gettext.info --enable-encoding gettext.texi
> /packages/GETTEXT/work/gettext-work/gettext-tools/doc//msgfilter.texi:228:
> warning: invalid encoded character `szlig'.
> /packages/GETTEXT/work/gettext-work/gettext-tools/doc//iso-639.texi:228:
> warning: invalid encoded character `aring'.
>
> And the resulting info file contains transliterated forms of ß and similar,
> _although_ I have explicitly specified --enable-encoding and
> @documentencoding UTF-8 .
>
> The code has a "fixxme", so apparently this behaviour of makeinfo is
> undesired also by other poeple.
>
> This fixes it for me.
I'm not the Texinfo maintainer, but thanks anyway.
> ! if (document_encoding_code == UTF_8)
> ! {
> ! size_t i;
>
> ! for (i = 0; i < sizeof (unicode_map) / sizeof (unicode_map[0]); i++)
> ! if (strcmp (html, unicode_map[i].html) == 0)
> ! return unicode_map[i].unicode;
unicode_map[] has over 200 entries. I think linear search is not
really appropriate for such a long list.
