Re: undecipherable (wrong?) text on def* arguments and --

[Karl Berry]

  [...] when an argument name contains @samp{--} [...],
  use @code{@@var} around the argument.  This causes the name to be
  printed in slanted typewriter [...]
  @c except for ?` and !`, but we won't explain that.

This is talking about the texinfo.tex behavior (I suppose I should
change the doc to say so).  I think the makeinfo (and texi2any) behavior
was never in question.  Test file:

-----------------------------------------------------------------------------
\input texinfo
@setfilename defun.info

@defun a--b---c hyphen2--args eslig?`args
No @@var.
@end defun

@defun a--b---c address@hidden address@hidden
With @@var.
@end defun

@bye
-----------------------------------------------------------------------------

In the first case, the -- turns into an en-dash.  This happens at a low
level in TeX and defeating it in general would have all kinds of
undesirable ramifications.

In the second case, putting @var around the -- makes it possible for
texinfo.tex to disable the "ligature" reasonable.  The output is in
slanted typewriter, not just typewriter.  This is desirable in principle
over @code because the arguments in general are typeset in slanted, not
code.  Although I just tried it and I did not see a visible difference
between the slanted typewriter - and the typewriter -, not surprisingly.

The second argument, with the ?` sequence, is just to explicate the
comment line you found.  ?` and !` are the "Spanish ligatures" in TeX
(actually in the Computer Modern fonts, not TeX itself), and so again,
defeating it is hard/undesirable.  I didn't bother to fix it even with
the @var because the reality is that no one ever needs those sequences,
which is why Knuth used them in the first place.

I realize none of this is logical or Texinfo-like but I see no help for
it.  The extra <var> or whatever in the other formats doesn't actually
hurt anything (aside from being illogical), does it?

k