Re: [help-texinfo] Umlaut strangeness

[Andreas Falkenhahn]

Hmm, this is really weird. When using

    @documentencoding ISO-8859-1
    ....
    @chapter 0xdc  

the character is rendered correctly on the PDF page but not in the bookmarks!
If I leave out the 

    @documentencoding ISO-8859-1

then the character is rendered correctly in the bookmarks but not on the
page! This is pretty confusing.

Note that the 0xdc above is only a placeholder for the real character. Of
course I'm not passing 0xdc to @chapter but the 8-bit value directly. I'm just
using 0xdc here so that you know what I mean because email programs might
mess things up and make it look like I'm passing UTF-8 or other Unicode
stuff to @chapter which I'm not doing.

I'm not sure whether you're saying that what I want to achieve is actually
impossible or whether it is possible.... from what you've said I'd infer that
it should be possible to use umlauts in the bookmarks *and* on the pages
if I just use ISO-8859-1 because the bookmarks use Latin 1. But it doesn't
work, cf. see above.

So is it possible to use umlauts on pages *and* in the bookmarks or can I
have umlauts only on one side - pages *or* bookmarks but not both?

Thanks!

On 31.05.2016 at 23:32 Karl Berry wrote:

>     When using this in an ISO-8859-1 encoded *.texi:
>     @chapter �Übersicht

> As long as you specify @documentencoding ISO-8859-1 and use the 8-bit
> input character 0xdc (your mail is garbled on my system, sorry), it
> should work.

> Another possibility is to use @"Uber... (all 7-bit ASCII) instead of any
> literal character.

>     overview of all chapters that appears in Acrobat Reader's side panel?

> The PDF overview (aka bookmarks aka outlines) is rendered using a
> different encoding than anything else (PDFDocEncoding, a superset of
> Latin 1) and with system fonts instead of document fonts.  The @"U
> character is at position 0xdc in that encoding (defined in Appendix D of
> the PDF reference manual), like Latin 1.  Therefore a UTF-8-encoded
> character (where U-umlaut is not 0xdc) will not come out right.

> texinfo.tex has never tried to deal with all this perfectly; it would be
> a big job.  Instead, rather than pushing what would amount to binary
> garbage into the PDF, it uses the "sortable ASCII equivalent" -- clearly
> suboptimal in many ways, but that's the status.

> Nowadays, PDF bookmarks can also use Unicode strings (UTF16-BE
> <FFEF...>).  So the right thing would probably be to translate all
> character-generating Texinfo commands into their Unicode equivalents,
> thus assuming that the system font being used is UTF-8.  Only for
> purposes of the PDF bookmarks, nothing else ... not offering to
> implement it, just mentioning :) ... --best, karl.



-- 
Best regards,
 Andreas Falkenhahn                            mailto:address@hidden