Re: [css3-syntax] UNICODE-RANGE production rule

On Tue, Jul 17, 2012 at 12:37 AM, Slaven Rezic <srezic@iconmobile.com> wrote:
> Hello,
>
> I think that the production rule for UNICODE-RANGE in
> http://www.w3.org/TR/css3-syntax/#tokenization should have the first
> question mark stripped. That is, instead of
>
>     UNICODE-RANGE     ::=     "U+" [0-9A-F?]{1,6} ('-' [0-9A-F]{1,6})?
>
> it should be
>
>     UNICODE-RANGE     ::=     "U+" [0-9A-F]{1,6} ('-' [0-9A-F]{1,6})?

As Simon says, the WD of Syntax that you're looking at is nearly a
decade old and long-abandoned, and I'm in the process of reviving it
in a better form.  The issue you raise is irrelevant in my new draft.

Ignoring that for a moment, though, your issue is correct, though your
correction isn't.  The grammar is *trying* to capture the
possibilities expressed in
http://dev.w3.org/csswg/css3-fonts/#unicode-range-desc , where a
unicode-range value with a single term can have trailing ?s as a
shorthand for a larger range.  For example, U+12?? is the same as
U+1200-12FF.  Obviously, though, the grammar as written doesn't
capture this correctly, as it allows ?s in non-trailing positions, and
allows them to be mixed with a second term.

So, thanks for paying attention and noticing this error that we've all
overlooked, even though it happens to be obsoleted by other work we're
doing!  ^_^

~TJ