- From: Boris Zbarsky <bzbarsky@MIT.EDU>
- Date: Wed, 28 Sep 2011 15:06:29 -0400
- To: public-css-testsuite@w3.org
On 9/28/11 2:45 PM, Arron Eicholz wrote: > On Wednesday, September 28, 2011 7:14 AM Tab Atkins Jr. wrote: >> On Mon, Sep 26, 2011 at 1:52 PM, Arron Eicholz >> <Arron.Eicholz@microsoft.com> wrote: >>> Unfortunately we defined 0 to be both positive and negative. >> >> Wait, we did? Dammit, that means I probably have language that needs >> adjusting in my specs. I assumed that if you wanted to include zero you used >> "non-negative/positive". > > The problem is that 0 can take signs. Since it can then we have to be explicit about it. If you say 'positive' then +0 is valid, similar issue with 'non-negative'. However, -0 would of course not be valid for 'positive' or 'non-negative'. I assume you're talking about -0 in the IEEE double sense here? Why is that a concern for CSS? CSS numbers are real numbers, not IEEE doubles. Implementations that implement them as IEEE doubles need to handle any resulting inconsistencies, of course.... Note that CSS2.1 explicitly says at http://www.w3.org/TR/CSS2/syndata.html#numbers : -0 is equivalent to 0 and is not a negative number. -Boris
Received on Wednesday, 28 September 2011 19:06:57 UTC