Re: HTML has no definition / automated test suite

2014-11-02 0:48, Patrick H. Lauke wrote:
> On 01/11/2014 14:05, Stefan Reich wrote:
>> A proper definition of HTML would include collections of sample HTML
>> source plus IMAGES of how they look rendered.
>
> No, as HTML does not specify look/presentation/default browser rendering.

It’s not quite that simple.

All HTML specifications have said *something* about (default) rendering. 
Not normatively, as a rule (except that good old presentational markup 
is defined in terms of rendering), but still.

HTML5 takes this much farther. It has an entire major section devoted to 
the issue,
http://www.w3.org/TR/html5/rendering.html
Formally, it is not normative, but it describes the “expected” 
rendering, and this is what browsers nowadays try to implement. And it 
is “conditionally normative”: “/For the purposes of conformance for user 
agents designated as supporting the suggested default rendering 
<http://www.w3.org/TR/html5/infrastructure.html#renderingUA>, the term 
"expected" in this section has the same conformance implications as the 
RFC2119-defined term "must".”/

Automated testing for HTML5 conformance for user agents claiming to 
support the suggested default rendering is thus a meaningful idea. 
Whether it is feasible is a different idea. Images are neither necessary 
nor adequate here, since the expected default rendering needs to be 
described in exact terms, usually indicating CSS counterparts, rather 
than images. Images might help as non-normative additional guidance, but 
most of the rules are simple enough so that images would be superfluous.

-- 
Yucca, http://www.cs.tut.fi/~jkorpela/

Received on Sunday, 2 November 2014 11:03:52 UTC