- From: Brendan Kenny <bckenny@gmail.com>
- Date: Mon, 30 Aug 2010 17:48:18 -0500
- To: fantasai <fantasai.lists@inkedblade.net>
- Cc: Doug Schepers <schepers@w3.org>, "Tab Atkins Jr." <jackalmage@gmail.com>, Håkon Wium Lie <howcome@opera.com>, Chris Lilley <chris@w3.org>, www-style@w3.org, www-svg@w3.org
On Sun, Aug 29, 2010 at 12:07 PM, fantasai <fantasai.lists@inkedblade.net> wrote: >> What does everyone else think? > > See above. > > I understand the arguments for harmonizing number literal notation > across languages, but I also agree with Håkon's argument that e-notation, > especially with negative exponents and in a context where unit identifiers > and subtraction are natural, is confusing. > > ~fantasai I definitely understand this argument, though I agree with Tab that it is a bit of a red herring. Few authors will ever need to manually use "e notation" in a stylesheet, let alone want to use it. The issue I encounter daily is using JavaScript to set CSS properties that accept <number>s. Every time real numbers are allowed and zero is a possible value, the number must be checked since floating-point issues ensure that if the exact solution to an equation is zero, the numerical solution will often only be close to zero and so be serialized using e-notation. While CSSOM will fix some of this serialization madness (for properties like opacity), more complicated style properties (like transform functions) will continue to be discarded if they contain a number expressed using scientific notation anywhere within the value string. If there isn't a fundamental grammar issue (I'll defer to others on that one), the possibility of someone writing terrible stylesheets (which would have to be pretty much willfully done, at which point there are bigger problems) seems completely outweighed by the help this would give JavaScript authors.
Received on Monday, 30 August 2010 22:48:50 UTC