bug#16292: 24.3.50; info docs now contain single straight quotes instead of `'

[Drew Adams <drew.adams@oracle.com>]

> >> Perhaps your info+.el user could investigate why makeinfo is
> >> misbehaving for them,
> > Can you suggest how?
> 
> Run 'makeinfo' on the attached file t.texi and look at the quoting in
> the output file t.info. It should use ASCII accent-grave and apostrophe
> with old texinfo, and Unicode curly single quotes with new texinfo.
> Similarly, the double quotes should use ASCII undirected double-quote
> (both times) with old texinfo, and Unicode curly double quotes with new
> texinfo.

OK, I just passed that information along to him.

FWIW, I have received this info from him:

  I built from source with makinfo 5.2. However, makeinfo 5.2 works
  fine for Emacs pretest (emacs 24.3.90), but not for Emacs trunk
  (24.4.50): bzr branch bzr://bzr.savannah.gnu.org/emacs/trunk

  Here are the steps to build:
  - ./configure --with-x-toolkit=gtk
  - make
  - sudo make install

What does old vs new texinfo refer to?  Did you perhaps mean old vs
new makeinfo?  If not, how does he tell whether he has "old" or "new"
Texinfo?

> > I would still want Info+ to be able to easily test, using Lisp, which
> > is being used in the current Info buffer: `...' or '...'.
> 
> Info files should never use the latter in Emacs 24.4, so this shouldn't
> be a problem.
>
> Down the road there may be a problem, as we really need to accommodate
> new Texinfo. It has been suggested that we change the Emacs build
> procedure to transliterate the output of new Texinfo so that by default
> it quotes 'like this' instead of ‘like this’. I think this would be a
> mistake, and your recent bug report gives another argument against it.

Is the problem just for `...' vs '...' or ‘...’?  What about Lisp (and
other) code, where we use ", ', and `, and none of those should be
changed to a curly version of the same?  A Lisp string, for instance,
must be enclosed in straight double-quote chars, the same char at each
end.

It would be madness if such chars were being automatically "improved"
everywhere willy nilly to become curly versions.

> For your planning purposes, here is a list of non-ASCII characters that
> are generated by new Texinfo when applied to the Emacs documentation as
> of January or so. You might want to check that info+.el handles these
> characters, as some users will run into them either because they built
> the manuals themselves with new Texinfo, or because they're reading
> manuals generated by other GNU projects. I think Emacs is the only
> holdout that still insists on old Texinfo.
> 
> ' ' (i.e., NO-BREAK SPACE, U+00A0)
> '真' (i.e., Han character 'real, actual, true, genuine', U+771F)
> 
> ¤ © ¬ » À Å ß à á ä å ç è é ê ì í ï ò ó ö ø ü ć č ł ń ő Š š – — ‘ ’ “ ”
> • … → ↦ ⇒ − ≡ ⊣ ★

Not sure why you let me know this (but I'm glad to have it).  A priori,
I have no problem with any such chars in an Info buffer.  Any problems
that I would have would I think come from substituting chars that
currently have particular meaning to Info (or to Lisp etc. - see above).