Re: User agent requirements and document conformance (was: Re: Cleaning House)

On 5/3/07, Anne van Kesteren <> wrote:

> <center> is also unlikely to be allowed. Yet it needs to be part of the
> parsing algorithm and part of the rendering requirements for visual
> desktop browsers to ensure interoperability. ...


> So I'd like to know if your argument is about some of the current elements
> allowed by the specification such as <b> or if it is about user agents
> requirements in the parsing section, rendering section, et cetera. There
> is a line between those two which I think is useful to clearly mark.

My objection is about the parsing/rendering/etc section.

I view these "here is how to handle tag soup the normal way"
requirements roughly the same way I view HTTP requirements.  They are
certainly *related* to HTML, but they are separable, and should be

(Based on as of May
3, 2007), as a first pass, I would separate out the following:

    Section 4 (Web Browsers), except for subsection 4.4 (Links)
    Section 6 (Communication)
    Section 8.2 (Parsing)

You need this information to write an interactive application that
works in the wild, or to write a new fully-functional browser.  You
don't need it to read/write/validate a document.  In fact, you're
probably better off if the people writing the content *don't* think
about these sections in detail, and just leave these concerns isolated
to a few common scripts.

Whether to call this new document "Web Compatibility", or to let it
keep "Web Applications" instead of "Web Documents" -- I'm not sure.

Whether to further simplify (move some elements to the Compatiiblity
document?  rephrase the parsing to distinguish between normal path and
expected recovery from common errors?) ... is a decision to make after
deciding whether to split at all.


Received on Thursday, 3 May 2007 16:26:28 UTC