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

[Eli Zaretskii <eliz@gnu.org>]

> Date: Mon, 30 Dec 2013 11:07:02 -0800
> From: Paul Eggert <eggert@cs.ucla.edu>
> CC: 16292@debbugs.gnu.org, grfz@gmx.de
> 
> Latin-1 markup (which you're advocating) is not legible in all locales.

Yes, but Latin-1 characters are only used in a handful of words (names
of people), so using Latin-1 doesn't hamper users' ability to read the
manual in any way, because the quote characters are not affected.
Only the UTF-8 encoding also affects the quote characters, which are
used a lot in Info.

So Latin-1 is a much lesser evil.

> It's not just the few files that specify "@documentencoding UTF-8".
> It's several other files that contain UTF-8 characters, files like
> dbus.texi.  These files are being mishandled now, because of the revert,
> because makeinfo isn't being told what their encoding is.  And the
> first of these files that I found, namely dbus.texi, is specifically
> talking about multibyte characters in its example, so converting it
> to Latin-1 would destroy the point of that example.

The job is not yet finished, so naturally there are problems we need
to solve.  I've just reverted the situation back to where it was
before, to the point where we should discuss the possible solutions
and choose the best one.

> I sympathize with the goal of producing .info files portable to
> all locales, but this revert is not solving that problem and it is
> making matters significantly worse in other areas.

The problem of quotes _is_ solved, in most of the Info files.  Again,
I'm not claiming we should release the files in their present shape.
There's still work to be done, the question is what to do.

> The other change you're advocating (converting to Latin-1) would be
> worse yet.

I don't see why it would be worse, please elaborate.

> Instead, I suggest undoing the revert, and adding something to
> postprocess the info files, to produce ASCII-only info files.
> These could be used in locales that don't grok UTF-8.  Doing
> this will avoid screwing up the documentation in .html and .pdf
> formats.  I'll volunteer to write that.

Are you suggesting to produce 2 separate sets of Info files?  That
could work, although it will be unusual (I don't think any other
project does that).  But if we go that way, there's no need to
postprocess anything, just run makeinfo again with --disable-encoding
and direct the output to a different place.

If you suggest to have a single set of Info files, then I don't
understand the proposal, or how it would help.