Re: Upcoming loss of usability of Emacs source files and Emacs.

[Alan Mackenzie]

Hello, Paul.

Sorry it's taken me some time to get around to answering your post.

On Fri, Jun 19, 2015 at 08:36:19AM -0700, Paul Eggert wrote:
> Alan Mackenzie wrote:

> > Any benefits there may be are not _practical_ ones.

> Sure they are.  They [curly quotes] make documentation strings easier
> to read, for people who are used to today's typical displays.

Is that really true?  Even if it is, the difference between curly and
ASCII quotes must be quite small in this respect, surely?

> And there are other practical benefits, e.g., being able to cut and
> paste from help buffers that use modern styles.

I haven't a clue what "help buffers that use modern styles" means.  One
can cut and paste from help buffers equally easily regardless of the
quoting style, if any, used in them.

> > In a true experiment, comment and objections would be actively
> > encouraged at an early stage.  I don't think this has happened wrt
> > these curly quote changes.

> You're quite mistaken.  The doc string fixes were proposed on 20 April
> in Bug#20385.  Comments were solicited and the fixes were improved
> over a period of several weeks, before the patch was installed on May
> 28.

I've just had a look at bug #20385.  I even posted to that discussion a
couple of times myself, yet I wasn't quite aware of the massive changes
you had in mind.  Possibly you weren't either when you started the
thread.

I don't think you were consciously trying to slip things through,
avoiding discussion, but if you had've been, doing precisely what you
did would have been an effective way of going about it.  Even the
subject "Support quoting 'like this' in doc strings", with nice gentle
fuzzy words like "support" instead of harder, more incisive words could
have been calculated not to draw too much attention.

You could have had a subject like "Supersede ASCII quoting convention in
doc strings", and most importantly, put it in emacs-devel where it would
have been much more prominent.  Note how much more vigorous has been the
debate in this thread than the one on the bug list.

> In contrast, you installed commit 52c3946c872c8bd96508f74cdda5cbb90c664306, 
> an 
> equally large change to user behavior, ....

Not really.  I simply made the new behaviour optional, setting the
default to "off".  And I also drew attention to my change.

> .... and without first proposing it so that we could comment on it
> (surely you knew it would be controversial) ....

Yes, I guessed it would be a bit controversial.  But it was never really
intended to be much other than an emergency quick fix.  Having doc
strings mandatorily translated into curly quotes was truly
objectionable, and best fixed quickly.

> .... and catch the obvious errors with it (even assuming we agreed
> with the idea).

The whole idea of curly quotes in doc string is only now being fully
discussed.  As already noted, it could have been discussed much earlier.
You could even have implemented the option yourself, saving me (with
less familiarity in Emacs's C code) from having to do it.

> This was another mistake, one that still needs some work.

[ .... ]

-- 
Alan Mackenzie (Nuremberg, Germany).