- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Fri, 21 Feb 2014 12:15:46 -0800
- To: Bjoern Hoehrmann <derhoermi@gmx.net>
- Cc: "bugzilla@jessica.w3.org" <bugzilla@jessica.w3.org>, www-svg <www-svg@w3.org>
On Fri, Feb 21, 2014 at 12:03 PM, Bjoern Hoehrmann <derhoermi@gmx.net> wrote: > * bugzilla@jessica.w3.org wrote: >>https://www.w3.org/Bugs/Public/show_bug.cgi?id=24769 > >>http://www.w3.org/TR/SVG2/single-page.html#paths-PathDataBNF >> >>There is no description in the text on how this BNF is to be used. For example, >>are the rules for the BNF alternatives to be parsed as 'first match wins'? >> >>If so, then the following rules won't produce good results, as the may match >>part of a floating point number as an integer before matching the >>floating-point-constant rule: >> >>nonnegative-number: >> integer-constant >> | floating-point-constant >>number: >> sign? integer-constant >> | sign? floating-point-constant > > An `integer-constant` is a `digit-sequence` while a `floating-point- > constant` is either a `fractional-constant` followed by an optional > exponent, or a `digit-sequence` followed by a mandatory exponent. A > `fractional-constant` must include `.`, so either a `.` or an expo- > nent's `e` resolve the choice, and it does not seem like either is > a prefix to a following rule. Could you give an example string that > has more than one parse tree? > > Certainly `integer-constant` is prefix of `floating-point-constant` > but there isn't an obvious advantage to putting it last, automated > tools that prefer some order can re-arrange them as they see fit, > and to a human reader less to more complex ordering might actually > be the better. The issue is that the order isn't defined *at all* currently; the spec doesn't even have an *internal* ordering that it defines and follows. ~TJ
Received on Friday, 21 February 2014 20:16:33 UTC