OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help

office message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]

Subject: Re: [office] ISO 14977 EBNF grammar

> My point is that you can use ISO 14977 to define a range operator and 
> the Unicode characters that you want to use with it.


But then you're not using the standard as it is; you're using a nonstandard extension.
Better to use a standard notation that _has_ a range operator.

> See: http://www.open-std.org/jtc1/sc22/wg11/docs/n506.pdf, for example. 
> (It uses ISO 14977 to define a range operator.)

No it doesn't.  I looked at that spec.

Section 5.1 describes the EBNF; it has NO range operator in the EBNF
metalanguage.  Which is expected, because it's using ISO's EBNF.
It also doesn't use ranges where it'd be obvious to do so, e.g.,
note that 7.1 (which uses the EBNF metalanguage) has to list each character,
instead of using a range, because the EBNF has no range operator.

Now it's true that 8.2.1 talks about a "range" operator, but this is
NOT a range operator in the EBNF notation.
That section is using EBNF to define a range operator in the language
the spec is defining. That does NOT give a "range" operator
to the EBNF itself; the ENBF continues to lack a range operator.

Yes, this could be worked around by using prose definitions, set unions
and differences, and defining many more nonterminals.
But I see no need to use those hacks.
ISO has accepted other standards that use the W3C/XML BNF notation, and
I think we have a good technical reason for using W3C/XML BNF: lack of ranges.
The much shorter/simpler resulting spec is in my mind a good justification too.
Clearly W3C/XML is itself defined in a standard, one which is available for
all to use and was developed by a wide consensus. Heck, I suspect ISO
has ratified it (if so, you'd think it'd count as an ISO standard too).

{Heck, ISO is willing to ratify a specification that is explicitly incompatible with
the Gregorian/ISO 8601 calendar, so using ISO standards even when they
clearly DO apply is obviously not THAT important ;-)  ;-)  ;-).  }

--- David A. Wheeler

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]