It breaks workflows, not browsers. The conformance rules are so ad hoc that they have to be hand-coded into XML processing streams. Might not be a problem for your hobbyist PHP coder, but for the serious maintainer... well, we've all been there before, right?

Have you even tried to read section 9 of the proposal? It's a nightmare. The problem with the spec as it stands is that it contains 10% original material and 90% post facto justification for existing browser behavior.

My personal gripe is with the haphazard way of handling DOCTYPEs (9.1.1). Granted, no browser handles DOCTYPEs correctly, but that doesn't justify the requirement of the "useless" element, and "about:legacy-compat" is just sad, an afterthought when somebody explained to Hickson that the DOCTYPE actually has meaning outside of the world of Netscape-begotten browsers. If you're not going to use it, lose it. So it'll break ancient browsers. If you are using a significant chunk of the new features that HTML 5 offers to justify its use, then ancient browsers are going to break anyway.

Gah, I just glanced at the preliminary spec again to look up section numbers. Compare the parsing rules for "raw" HTML 5 (9) to that of the XML version (10). How many browser vendors are going to implement that algorithm completely and correctly? Apple, that's who; and only because they are a major driving force of the WHAT-WG. Hell, the error recovery operations (9.2.8) are non-normative, which is a fancy way of saying "we're afraid that if we commit to something vendors won't like us."

Yes, I know: I should hurt before I holler. But let's face it: it's wag the dog, redux. The acknowledgement section alone pretty much admits this. Developers aren't going to reference this document at all; they're going to see what WebKit does with their docs and take that as gospel. And since error checking is "soft," bad behavior will proliferate, become entrenched, and all other browsers will be forced to follow the WebKit interpretation. It's justification for cargo cult coding, but this time WebKit will dominate instead of MSIE. Just because it is open source does not make it right.

Amen, brother.

The problem with the HTML5 "spec" is that it is so much penis-programming: purposely breaking existing tools and workflows so that everyone will be forced to rewrite everything from scratch. The parsing model is so ad-hoc that it is sure to create another code-for-the-browser generation of web developers.

Actually, what Apple offers in retail is an upgrade, and is labeled as such. That it allows you to do a full wipe of the system without checking for a previous installation is a kindness on their part; should this Psystar case go south for them, expect more draconian restrictions. I wouldn't be surprised if Apple required you to bring in your Mac to an authorized repair center for an upgrade, especially since the majority of their (non-technical) customers never bother upgrading and wouldn't be inconvenienced. Any loss from retail would be easily balanced by protected hardware sales.

Try selling white boxes and bundling in copies of a Vista upgrade pack instead of the OEM package, and see how long it is before Microsoft is banging on your door. Actually, Microsoft wouldn't care unless you flaunted it, just like Psystar did; then they would be forced to act.