[tex-live] clash between babel (french), hyperref and \cite on keys with a colon character

[Zdenek Wagner]

2014-09-22 20:36 GMT+02:00 Robin Fairbairns <Robin.Fairbairns at cl.cam.ac.uk>:
> Vincent Lefevre <vincent at vinc17.net> wrote:
>
>> On 2014-09-22 16:34:21 +0100, Robin Fairbairns wrote:
>> > Vincent Lefevre <vincent at vinc17.net> wrote:
>> > > On 2014-09-22 15:55:24 +0100, Robin Fairbairns wrote:
>> > >
>> > > > it is a feature because it's insoluble when using tex or latex.
>> > >
>> > > Why insoluble? What's the problem with redefining \cite?
>> >
>> > it's trivial, of course.  except that there are quite a few definitions
>> > of \cite, and you need to know which properties are needed by each use
>> > of each version.
>>
>> OK, I didn't think about that. There's still the question whether
>> it could be an hyperref bug (see jfbu's answer).
>
> indeed.  they do still pop up.  it's a rather old package (i think its
> first version -- pre-heiko -- appeared in the 1990s).
>
>> > and indeed, what's wrong with switching off active status before use of
>> > a punctuation mark in a macro?
>> >
>> > imo, the only _real_ solution involves switching to a "cleverer"
>> > tex-like processor.
>>
>> I've just tried lualatex (to compare), but the PDF bookmarks are
>> displayed as if ISO-8859-1 were used, while the source is in UTF-8.
>
> i've never done anything non-trivial with luatex.  getting the encodings
> wrong, like that, sounds like something missing somewhere.  i would poke
> around with heiko's documentation (or possibly his huge collection of
> small packages in macros/latex/contrib/oberdiek on ctan).
>
The problem lies in Adobe. The PDF documentation says that the
bookmarks have AdobeStandardEncoding. In newer versions (nowaday's
versions fall into this category) you can override this default by
starting a unicode string with a BOM. The only thing which is needed
is to supply "unicode" as a hyperref option and it will create unicode
bookmarks with BOM.

>> And xelatex seems fine.
>
> simpler platform, of course.  (and older, bedded in for some time.)
>
Here the mystery lies in xdvipdfmx. It uses unicode only and adds BOM
automatically without any help from the hyperref package. I hope I
remember it well.

>> Now, perhaps the use of pdflatex should be discouraged in a visible
>> way.
>
> i wouldn't do that on the basis of one positive test.
>
> the requirement for utf-8 is going to frighten a lot of people, of
> course.  (i don't use utf-8 in ordinary work, since english is _soo_
> bare-bones simple.  i could work -- with some difficulty -- in french,
> but no other language unless you count american.)
>
Even Indic languages and many others are possible with pdf(la)tex but
nowadays nobody designs fonts for old-style TeX. On the contrary,
OpenType fonts are available, both free and commercial, because they
can be used by many programs. It is thus a natural step to support
these fonts in new TeX engines.

> robin



-- 
Zdeněk Wagner
http://hroch486.icpf.cas.cz/wagner/
http://icebearsoft.euweb.cz