Re: [O] Bug: When exporting to PDF an Org file where multiple footnotes share the same definition, only the first footnote is clickable [8.3.4 (8.3.4-39-ge0acd8-elpaplus @ /home/jorge/.emacs.d/elpa/org-plus-contrib-20160418/)]

[Rasmus]

Hi,

Nicolas Goaziou <address@hidden> writes:

> Rasmus <address@hidden> writes:
>
>> The transcoded version of the second "[fn:1]", say S2, will be passed to
>> our filter FUN, if FUN is added to
>> org-export-filter-footnote-reference-functions.  But S2 will only contain
>> the estimated number of the footnote¹, e.g. "\footnotemark[1]{}".
>
> Indeed. I don't know any other way to handle repeated footnote markers
> in LaTeX.

Could we inject labels in all footnotes?  If so we could simply use

   \textsuperscript{\ref{FN-LABEL}}

In place of

   \footnotemark[FN-GUESS]{}

It seems \footref of scrextend.sty has some extra robustness built into it
but until someone complains I think the \textsuperscript hack should be
enough and better then what we’ve got.

Example:

     \documentclass[11pt]{article}
     \usepackage[hidelinks]{hyperref}
     \usepackage{parskip}   % to avoid \noindent 
     \usepackage{scrextend} % to get \footref 
     \begin{document}
     \addtocounter{footnote}{1}
     Text\footnote{\label{org925ce52} Footnote.}\\
     Text\footref{org925ce52}\\
     Text\textsuperscript{\ref{org925ce52}}\\
     Text\footnotemark[2]{}\\
     \end{document}


>> We could capture the transcoded version of the link "[[f1]]" in FUN, if
>> FUN is added to org-export-filter-link-functions (I think).  But by then
>> we’d only be served a generic reference, e.g. "\ref{org925ce52}".  I don’t
>> know how we’d then differentiate between a link to a footnote and to some
>> other type of element, e.g. a source block.  Hence, it becomes hard to
>> guess when to use e.g. \footref{.} instead of \ref{.}.
>
> You can retrieve element referenced as "org925ce52" with
>
>   (cdr (assoc "org925ce52" (plist-get info :internal-references)))

Cool!  I hadn’t noticed that these are part of info.

> It's a bit low-level, but it is useful, it could be factored out as
> a function in "ox.el".

It should be enough to allow Jorge to write a simple filter to address
original issue.

Rasmus

-- 
To err is human. To screw up 10⁶ times per second, you need a computer