Re: character syntax fixes needed

[Dave Love]

Richard Stallman <address@hidden> writes:

> What I'm telling you is that the standards are not authorities.

They obviously are authorities when they define the charsets Emacs
wants to implement, at least.  As far as I know, Unicode is the single
authority on the world's character usage, as far as it goes.  Why do
you think it isn't (modulo bugs which I doubt it has now for the Latin
scripts at issue but which you could just get fixed)?  What else
codifies this information?

No one has presented any evidence that Unicode is wrong on what it
says about usage and classification.  It doesn't sound as though
people arguing have even read it.

> We do
> not *have to* follow them.  Arguments that presume that the standard
> is an authority we must obey are simply invalid.  The decision about
> whether to follow any given standard on any given issue is a practical
> decision.

I've no idea why you think I'm not talking practicalities, on the
basis of having implemented support for more locales than anyone else,
I think.  I said that if you follow Unicode and other relevant
standards and usage guides, then you don't have to be, or have, an
expert on every script.  That's eminently practical.

> Telling us that "Unicode says XYZ" is an argument in favor of doing
> XYZ.  It is not an open-and-shut decision.

Unicode says how guillemets are used as a matter of fact, which you
can verify.  It's manifestly wrong for single ones to have word syntax
and double ones to have paren syntax and I wish I'd just changed it
when I had the chance.

> A Unicode-based Emacs means an Emacs whose character codes are mostly
> those of Unicode,

The codes are actually mostly not unicodes (by a factor of 64/17).
However they are a proper superset.  That means you can deal sanely
with the national charsets via their definitions in terms of unicodes
from glibc and elsewhere.  Otherwise you're pretty lost.

> and which works more or less according to Unicode
> specifications.  It does not mean an Emacs that slavishly obeys
> whatever Unicode says.  We don't slavishly obey standards.

I don't slavishly do things, and you're ignoring my implementation
caveats (including backwards compatibility).  I've looked at any
information I could find which seemed relevant, and searched quite
hard for some.  I'm talking after a considerable amount of work which
was meant to help.  It's a pity it obviously doesn't help and someone
else will have to repeat it eventually or that Emacs will be backwards
compared with the character handling in most of the rest of GNU.