On Tue, Jan 28, 2025 at 06:45:40PM +0000, Werner LEMBERG wrote:
>
> [texinfo.tex 2025-01-20.21]
>
>
> Consider the following example file `tr.texi`.
>
> ```
> \input texinfo
>
> @documentlanguage tr
>
> @tracingall
> @tracingonline0
> @contents
>
> @node A
> @chapter A
> A
> @page
>
> @bye
> ```
>
> If compiled with pdfTeX or LuaTeX this shows 'Icinderkiler' in the PDF
> outline. However, IIUC, the current version of `texinfo.tex` is
> expected to support the complete Unicode BMP within PDF outlines.
Not exactly. At the moment only characters that are actually defined
in texinfo.tex are passed through to the outline.
> In other words, it should show the correct Turkish word for 'contents',
> namely 'İçindekiler'. What am I missing?
The reason is the way that the translation is given in the translation file,
(txi-tr.tex), which uses the accent commands:
\gdef\putwordTOC{\dotaccent{I}\,cindekiler}
If I change this to use UTF-8 instead, the outline is produced correctly
with the accented characters:
\gdef\putwordTOC{İçindekiler}
It should be possible to fix this by outputting the combining accent
character, but this will take me time to implement.
> Another problem is that compilation with XeTeX fails. Calling
> `PDFTEX=xetex texi2pdf ...` yields the following error.
I can confirm. This bug was introduced in the most recent commit
to texinfo.tex. I've attempted to fix it in the most recent version
[2025-01-29.19].
