So: it turns out that the regina-normal build failure was not because of the libstdc++6 transition, but because source-highlight was built against an old version of libboost-regex. It seems that between August and now, somebody has binNMUed source-highlight to use boost 1.58, and as a result the build failure for regina-normal has gone away (see the notes for #797292, which I have just closed).
That leaves the question: what to do with this bug? I’m not sure whether libsource-highlight4 needs a rebuild with gcc5 - I do see lots of references to std::__cxx11::basic_string<…> in the output from “nm -gC”, but someone else would be better placed than me to give an opinion on whether this is enough for a rebuild. - Ben. -- A/Prof Benjamin Burton Computational Geometry & Topology Group School of Mathematics and Physics The University of Queensland, Australia