> On 24 Mar 2018, at 22:10, Paul Eggert <egg...@cs.ucla.edu> wrote: > > Hans Åberg wrote: >> Might it possible to get an option > > Sounds like a reasonable option to me, if someone wants to write and document > it.
I wrote some C++ code translating it back to UTF-8 for error messages, but it is sort of roundabout, and for the future it might good to prepare for UTF-8. For UTF-8, it would suffice in such explicit token names that they come back verbatim, and thus translated into the yytname array only for characters (octets) that C code "..." strings can't contain: I saw you wrote the code that does the actual translation. :-) In 'prepare_symbols' output.c, there is a section "tname -- token names" with code: set_quoting_style (qo, c_quoting_style); which is perhaps the part doing the translation. Perhaps it would suffice to change that.