new `\textsuperscript` definition breaks bigfoot's color management

[lemzwerg]

Using current TeXLive (commit 2103b15de04c8 from 2026-Jun-27), this document

\documentclass{article}

\usepackage{perpage}
\MakePerPage{footnote}
\usepackage{bigfoot}
\DeclareNewFootnote[para]{default}

\usepackage[colorlinks=true]{hyperref}

\begin{document}
foo\footnote{bar}
\end{document}

aborts with the following error.

! Undefined control sequence.
\set@color ...lt {\current@color }{\FN@colorstack 
                                                  }}\marks \FN@color {\FN@co...

l.11 foo\footnote{bar}

A larger document of mine that uses both hyperref and bigfoot with the mentioned options worked fine with TeXLive 0308fcb0 (from 2026-May-09); this corresponds to hyperref v7.01q.

[u-fischer]

it is not hyperref (and it errors also with older hyperref versions). This here errors already

\documentclass{article}
\usepackage{bigfoot}
\DeclareNewFootnote[para]{default}
\usepackage{color}
\begin{document}
foo\footnote{bar}
\end{document}

bigfoot doesn't like the new \textsuperscript definition. It works again if one uses color groups there, as bigfoots redefinition then can properly reset the color stacks:

\makeatletter 
\protected\def\@textsuperscript#1%
  {%
    \check@mathfonts
    \leavevmode
    \color@begingroup %<-----------
      \sbox\z@{\fontsize\sf@size\sf@size#1}%
      \raise\dimexpr\textsuperscript@offset\relax\box\z@
      \textsuperscript@space
    \color@endgroup %<----------
  }
\makeatletter

@josephwright should there be color groups by default?

[davidcarlisle]

adding color groups looks a bit odd (even if it allows the current bigfoot patch) as \sbox is already adding color groups around #1

[u-fischer]

\sbox uses \color@setgroup and not \color@begingroup, and bigfoot doesn't patch that, which means that it can't reset the color commands at the begin of the box.

[lemzwerg]

Thanks for the analysis. Assuming that it takes some time until this gets fixed and published in LaTeX (and thus in TeXLive), how can I circumvent the issue? Shall I use Ulrike's suggested patch?

[lemzwerg]

Ping! How to proceed? Shall bigfoot be patched (i.e., you apply a 'firstaid' fix), or is this something for the LaTeX kernel? In case it's the former, I would add the issue to the bigfoot tracker.

[davidcarlisle]

@lemzwerg it's a bigfoot issue and should be addressed there, firstaid is only a last resort if a package has no effective maintenance

[lemzwerg]

OK, thanks. Alas, 'bigfoot' currently has no effective maintenance (the last release was 10 years ago) – David Kastrup has no time in the foreseeable future to take care of this package...