- From: Craig Northway <craign@cisra.canon.com.au>
- Date: Thu, 25 Aug 2005 18:03:58 +0200
- To: Ian Hickson <ian@hixie.ch>
- Cc: www-style@w3.org
Hi, Ian Hickson wrote: >On Thu, 25 Aug 2005, Craig Northway wrote: > > >>>No, it's not optional, but that is because section 3.2 requires that >>>user agents implement CSS consistent with the descriptions laid out in >>>the spec, including section 6. >>> >>> >>I do not think this is clear enough. I think that each statement >>throughout the specification that requires conformance from a user agent >>should use terminology that indicates this. >> >> > >This is the case. > >CSS is not defined in terms of functional steps ("do this then do this >then do this"), it is defined in terms of a model ("it is this"), an >implementation for which must then be found. > > The example I quote from section 6.1.2 and 6.1.3 appear to be a functional steps in the model. >Most of the CSS spec describes the model. The only thing a UA implementor >has to do to claim conformance is be consistent with the model. > > > > >>The QA Framework: Specification Guidelines >>(http://www.w3.org/TR/2005/REC-qaframe-spec-20050817) requirement 7 states: >> >> Use a consistent style for conformance requirements and >> explain how to distinguish them >> >>Requirement 8 states: >> >> Indicate which conformance requirements are mandatory, which >> are recommended, and which are optional. >> >>Specifying in section 3.2 that a UA must conform with the rest of the >>specification does not make it easy to distinguish conformance >>requirements. Nor does it make it easy to determine which comments are >>mandatory, recommended or optional. >> >> > >It should be pretty clear; if something describes the model, it's >something the implementation has to be consistent with. > How do I know precisely what is defining the model? Parts of the model that must be performed should be expressed using 'must' or 'shall' or other suitable terms. > Certain things are >explicitly stated as being outside the model ("The behavior of the 'auto' >value is user agent-dependent") and RFC2119 terms are used when >conformance criteria outside the model are required ("but should cause a >scrolling mechanism to be provided for overflowing boxes") > But, everything that may be in the model is not obvious. > > >>>Specifically, in the case of 6.1.3, the two "must" requirements that >>>apply are "For each element in a document tree, it must assign a value >>>for every applicable property according to the property's definition >>>and the rules of cascading and inheritance" and "A user agent that >>>renders a document with associated style sheets must respect points >>>1-5 and render the document according to the media-specific >>>requirements set forth in this specification", which are list item 4 >>>and the third bullet point in section 3.2 respectively. >>> >>>Please let us know if that addresses your concern. >>> >>> >>Where are these musts you indicate that apply to section 6.1.3? they >>are not linked to section 6.1.3. >> >> > >Section 3.2. > >If this does not address your concern please explain how the CSS spec >should be rewritten to use a different conformance model. > Each statement that is required to be peformed by the UA in implementing the model should be specified with must or shall. > It is not clear >to me how we can require conformance to a model without using the >mechanism used by CSS2.1. For example, take margin collapsing: > > http://www.w3.org/TR/CSS21/box.html#collapsing-margins > > I fail to see how this is relevant to my initial comment. Regards, Craig Northway
Received on Thursday, 25 August 2005 16:04:14 UTC