> > Personally, I would prefer that quotes render sensibly in > > utf8, html and dvi, and that correctly written man pages do > > not need any modification. Also, the use of `foo' is very > > convenient (even if it no longer appears balanced in the > > source).
I agree. In case of doubt, stick to existing practice. It's rather unfortunate that a de-facto font encoding standard used by most coders (and also supported by a big font vendor (Adobe) -- remember that Adobe's 75 and 100 dpi bitmap fonts used in the X window system were encoded using Adobe's modified version of the Latin-1 encoding, which had open-quote and close-quote at 0x60 and 0x27) was subverted by reverting to a standard set down a long time ago (and which had from the beginning been designed to be a compromise). I think the Unicode designers made a grave mistake in ignoring this existing practice. Also, there are still fonts in which ASCII 0x27 and 0x60 are symmetric (you can also create your own (or remap an existing one) if necessary) which are still useful and convenient for programmers. > Let me try to summarize: > > . What people like to input in man pages is this: > > `foo' should be equal to \[oq]foo\[cq] > > \`foo\' should be equal to <ASCII 0x60>foo<ASCII 0x27> > > I say `in man pages' because \` and \' act as accents too, but > this is, AFAIK, not used (a quick search in the man pages on my > SuSE GNU/Linux box showed not a single man file which uses \' or > \` as an accent). > > . Depending on the output device, ` and ' should be mapped to > different characters. However, \` and \' should always map to > ASCII 0x60 and ASCII 0x27 for proper cut and paste support. > > Do you all agree with this? Not sure. The manual clearly states that \' represents an acute accent, which is a glyph very distinct from the ascii quote. The only way to get a consistent behavior across all devices would then be to give up on compatibility with AT&T troff, discard the equivalence of \` and \' with grave and acute accent, define two new glyphs "ascii backquote" and "ascii quote" (okay, the latter already exists) and consider these to be equivalent to \` and \'. The accents will still be accessible as \(ga and \(aa.
