Hi Werner, Werner LEMBERG wrote on Thu, Nov 06, 2014 at 02:57:50PM +0100:
> groff uses different quotes depending on -Tascii or -Tutf8: > > devascii: lq " > devlatin1: lq " > devutf8: lq u201C (via `glyphuni.cpp') This rendering of \[lq] seems a good choice to me and shouldn't be changed. > Using \*[Lq] and \*[Rq] in Anthony's patch is definitely correct. I have registered it in the bug tracker such that it doesn't get lost. > In file `doc-nroff', however, they are mapped to \[lq] and \[rq] for > -Tutf8 only. The other TTY devices get `` and ''. Given that the GNU > standard already moved away from `foo' and ``foo'' to 'foo' and "foo" > for TTY devices that are not capable of representing correct quote > characters, we probably should revise the code in `doc-nroff'. That is a seperate issue, so i changed the subject such that we don't confuse the two. Anthony's patch can be committed independently of this second issue, i think. > At least it is worth a bug tracker issue :-) I have submitted one, including a patch to doc-nroff, which is actually quite simple, works as expected in my tests, and doesn't break the mandoc regression suite (which is also, in a way, a regression suite for groff_mdoc(7), groff_man(7), and parts of groff(1) -Tascii and -Tutf8 because im using groff to produce the *desired* output files i'm checking mandoc against). When you merge either one (or both) of the submitted patches, i'll update mandoc to conform. Carsten, do you think Heirloom can follow that direction, too? Very early in the thread, you said: :: It is generally no problem to differentiate between nroff and :: troff output. I suggest to do this here and to apply the patch :: for typeset output. But i suspect that assessment might have been modified by what was discussed later. Applying both patches, we get nice and consistent output for all cases: - troff - nroff devutf8 - nroff devlatin1 - nroff devascii for all of .%T, .Dq, \*[Lq], and \[lq]. Thanks, Ingo
