Re: /srv/bzr/emacs/trunk r102591: * src/lread.c (read1): Allow newstyle unquote outside of backquote.

Re: /srv/bzr/emacs/trunk r102591: * src/lread.c (read1): Allow newstyle unquote outside of backquote.

[Glenn Morris]

Stefan Monnier wrote:

>   Extend the restriction no-unescaped-commas-and-backquotes-in-symbols
>   to all contexts.

landmark.el is unhappy with this (or vice versa):

  landmark.el:1140:55:Warning: reference to free variable `s'

etc. It uses commas in function names.

Re: /srv/bzr/emacs/trunk r102591: * src/lread.c (read1): Allow newstyle unquote outside of backquote.

[Stefan Monnier]

>> Extend the restriction no-unescaped-commas-and-backquotes-in-symbols
>> to all contexts.

> landmark.el is unhappy with this (or vice versa):
>   landmark.el:1140:55:Warning: reference to free variable `s'
> etc. It uses commas in function names.

Great, then let's fix landmark.el.
If we ever want to fully support new-style backquotes, we'll have to
disallow commas from symbols.


        Stefan

Re: /srv/bzr/emacs/trunk r102591: * src/lread.c (read1): Allow newstyle unquote outside of backquote.

[Glenn Morris]

Stefan Monnier wrote:

> Great, then let's fix landmark.el.

OK; done. (I wasn't sure if this was an intended consequence.)

Re: /srv/bzr/emacs/trunk r102591: * src/lread.c (read1): Allow newstyle unquote outside of backquote.

[Stefan Monnier]

>> Great, then let's fix landmark.el.
> OK; done.

Thanks a lot Glenn,


        Stefan

Re: /srv/bzr/emacs/trunk r102591: * src/lread.c (read1): Allow newstyle unquote outside of backquote.

[Stephen J. Turnbull]

Stefan Monnier writes:

 > If we ever want to fully support new-style backquotes, we'll have to
 > disallow commas from symbols.

You also have the alternative of distinguishing the first character
(ie, allow a different set of characters there).

On a related topic, has anybody ever run into an issue with lookalike
characters (eg, Japanese full-width latin and numerals, Latin A
vs. Cyrillic A vs. Greek A, etc)?  Unicode alleges this to be a
security risk.  (The reason this comes up here is that JIS mirrors the
entire ASCII set into its multibyte encoding -- including the comma.
So if we don't unify ASCII ?, with JIS ?,, you could have symbols that
have commas in them but those commas are syntactically non-commas. ;-)