Hi Carsten, Carsten Kunze writes: > is there a way (except \N'45') to output an ASCII dash (0x2d) with > nroff in an UTF-8 locale (i.e. without -T or with -Tutf8)? > > I want to copy command lines from nroff output into a xterm, but the > shell complains about the '-' which is not 0x2d. > > E.g. mandoc(1) does output 0x2d for '\-' and even '-' which IMHO is > better for manpages (to allow copiing text into the shell).
Ooh, what a question! Back in the day, nroff only output ASCII, and encoding of troff output was irrelevant. So \- became - in nroff, and a minus/en dash in troff (there was no visual distinction). In Unicode, minus and en dash are distinct (U+2212 vs U+2013). And nroff can output more encodings than just ASCII. And encoding of typeset output matters now because copy/paste is now a thing. Heirloom troff and groff both render \- as en dash, not minus sign, in PDF output. Groff renders \- as minus in HTML output. Mandoc renders \- as ASCII '-' in all outputs. Clearly there is no consistency here. Personally I avoid \- entirely, and use \(mi and \(en instead. I only ever mark up command flags in -mdoc, so I just type Fl for those. I wish Fl became ASCII '-', but we're talking about changing decades of historical practice here. -- Anthony J. Bentley
