The outstanding issues include:

How will the HTML specification be published? through the IETF? through SGML Open? the Davenport group? W3O?
The HTML specification was published as an IETF draft at one point. It has since expired. Draft versions for review will probably continue to be published as IETF drafts in the short term. But IETF documents are subject to review by anyone, and this tends to slow things down, since newcomers take a while to ramp up. An HTML task force is shaping up. It will probably take the form of an SGML Open technical committee, and the HTML specification will be published as an SGML Open technical report. (Thanks to Yuri Rubinsky for the idea.)
Will FORMs be part of the spec? How about TABLEs? MATH?
The various features of HTML go through several "states" -- proposed, required, optional, and obsolete. As the specification evolves, proposed features may become required, optional, or obsolete.

The current thinking is that as of the 2.0 specification, forms will be optional (i.e. they will be formally specified, but not requred for conformance), and tables and math will only be proposed (i.e. they may be discussed in specially labeled parts of the spec, and they may be supported by some implementations, but they are not yet formally specified)

Must a client support forms to be conforming?
No, but it must explicitly say that it does not support forms. There are "levels" of conformance. If a client doesn't explicitly give a level (as is the case in current implementations) level 2 is assumed, which means that forms are supported. But servers will be able to detect clients that do not support forms, if the client says:
	Accept: text/html; level=1
And a client can advertise its lack of support for images by specifying level=0.
Is P a separator or a container? Why does it matter?
More later...