W3C home > Mailing lists > Public > public-css-testsuite@w3.org > September 2011

RE: [CSS3-mediaqueries]: Invalid test cases in test suite

From: Arron Eicholz <Arron.Eicholz@microsoft.com>
Date: Thu, 29 Sep 2011 21:53:12 +0000
To: Chris Lilley <chris@w3.org>
CC: Boris Zbarsky <bzbarsky@MIT.EDU>, "public-css-testsuite@w3.org" <public-css-testsuite@w3.org>
Message-ID: <07349ECFC3608F48BC3B10459913E70B148E74E0@TK5EX14MBXC293.redmond.corp.microsoft.com>
On Thursday, September 29, 2011 1:37 PM Chris Lilley wrote:
> On Thursday, September 29, 2011, 6:22:23 PM, Arron wrote:
> AE> On Wednesday, September 28, 2011 1:24 PM Boris Zbarsky wrote:
> >> On 9/28/11 4:13 PM, Arron Eicholz wrote:
> >> > CSS 2.1 still doesn't cover 0 exactly. What is 0?
> >> The CSS 2.1 spec says its numbers are real numbers.
> AE> But what is a real number to CSS? There is no reference from the CSS
> AE> spec to go off of for real numbers.
> Hopefully CSS WG will not redefine real numbers.

CSS is unfortunately redefining 'real numbers' right now because we don't have a reference to a definition of 'real numbers'. We just use the term we don't reference what the term 'real number' is actually is defined as in all our minds. Even the link to the Wikipedia page would be better than nothing. Nothing means CSS defines what a 'real number' is and we seem to agree this is bad. We just need to be explicit in the spec for such a low level concept.

> AE> Again I am asking for
> AE> reference or us being explicit about 0 and +0. Maybe we need to do
> both.
> I think that is mixing levels (syntactic and semantic levels).

That's fine with me. Now we just need to decide if we want to reference something or provide actual text to change the spec.

> AE> You are correct that we say real numbers but the spec doesn't say if
> AE> 0 is included in positive or negative numbers (for example when we
> AE> discuss property values that take positive numbers).
> In terms of semantics, zero is neither positive nor negative.
> http://en.wikipedia.org/wiki/Zero#As_a_number

Yep I read that too but the CSS spec doesn't reference it and as far as CSS goes it just uses terms and assumes that there are normative references but there aren't any references.

> In terms of syntax, CSS allows "0" or "-0" or "+0" and these forms all result in
> the number zero.

Actually +0 doesn't per CSS 2.1. We don't define what +0 is in CSS 2.1 normatively. We all have assumed over the years and implementations that it results in 0 and we seem to agree on it that it should. We just don't say it.

> So if the intent is to allow zero, one etc then 'non-negative numbers' is the
> phrase to use.

The intent was to use the term 'positive numbers' which many had thought excluded 0 and +0. Unfortunately we don't define in CSS, or reference anything in the spec, that says if 0 and or +0 are positive numbers or not. So we must assume that they are until proven otherwise by text or a normative reference. The reason I say that is because we are very explicit about -0 so you have to assume that +0 is positive. And 0 is just completely ambiguous.

> >> -0 is equivalent to 0 and is not a negative number. +0 is equivalent
> >> to 0 and is not a positive number. 0 is neither positive nor negative
> number.
> Yes.

If you are agreeing with me to add the text to the spec then that is great. We just need to get it added.

I think what everyone is missing in this thread is, I am well aware of what 0 and +0 are per the 'real numbers' and 'zero' definitions people have referenced in these emails. Those references are great unfortunately these emails are not the spec and the spec needs those references to be clear on what it is explaining/defining.

Arron Eicholz
Received on Thursday, 29 September 2011 21:53:53 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:13:23 UTC