bug#20385: [PATCH] Support curved quotes in doc strings

[Paul Eggert <eggert@cs.ucla.edu>]

Dmitry Gutov wrote:

> What if the locale mandates the use of some other kind of quotes? Then copying
> and pasting won't work anyway.

Yes it will.  It'll work fine.  If a documentation string is in (say) Hungarian, 
a user will be able to paste quoted Hungarian text into it, and the text will 
carry its Hungarian quotation marks along.  It'll be just like documentation in 
English and with English quotes.

> I've never had a need to copy a diagnostic message with the
> intention to paste it into the source code.

Shrug.  I've done it.  And vice versa.  And there are other examples.  The point 
is that it's simpler and nicer if cutting and pasting Just Works.

> Actually, it's a decent argument against using *any particular locale's*
> quoting, in order not to confuse casual contributors, who'd have to know that
> *these* fancy quotes will be localized, but *those* - won't.

No, there's normally no need to localize those curved quotes, not if the text is 
English and uses English quoting.  The only time one would need to localize is 
for obsolete displays lacking curved quotes, which casual contributors are 
unlikely to be using nowadays anyway.

> Maybe you could paste quotes, but not any other kind of formatting.

First, that's not true: many other kinds of formatting (though obviously not 
all) can also be cut and pasted from info into .texi files.  Second, even if it 
were true it'd be OK.  One improvement at a time.  Quotes are such a common kind 
of formatting that simplifying their cutting and pasting is worth doing even if 
we can't simplify all forms of formatting.

> And you can notice that if we delimited docstrings with curly quotes instead of
> the straight quotation marks, tildify-space wouldn't have been able to use them
> exactly this way inside the docstring.

That's not a significant problem, as it applies to any convention that uses only 
paired quotes.  For example, the current documentation for 
skeleton-pair-insert-maybe says "These are (), [], {}, <> and `' ...", and the 
reader needs to infer that those instances of ` and ' are not quotes, but are 
characters.  This sort of thing happens every now and then in documentation, and 
it's no big deal.

> I know of no serious markup language that utilizes unicode, even now.

Texinfo does.  Unicode curved quotes are part of the markup.  They affect 
spacing among other things.

> I don't see it as clunky

That's the main point of disagreement here.  Although Emacs old-timers are used 
to quoting `like this', to newbies it's a weird and offputting relic from 
ancient history.