[whatwg] Chipset support is a good argument

[Aryeh Gregor]

On Mon, Jul 6, 2009 at 7:30 PM, Joshua Cranmer<Pidgeot18 at verizon.net> wrote:
> Perhaps what could break the deadlock would be Apple conceding to
> implementing Theora, or Mozilla conceding to implementing H.264. In either
> case, the decision to implement would most likely be a result of market
> pressure, not some arcane specification.

Given that Apple's only stated reason for not implementing Theora
support alongside H.264 is the patent risk, and given that a company
nearly as large as Apple (Google) has taken the risk and not gotten
sued yet, I'm personally hoping that market pressure won't be
necessary to get Apple to change its mind.  Every other good argument
I've heard advanced against Theora argues against supporting it as the
only standard, but not against supporting it as a baseline minimal
standard.

> I think you have a misconceived notion of the world here, too. Most of the
> HTML is not manually written by authors, it is automatically generated from
> programs, be it a Wiki-style generator, or a discrete utility like
> Dreamweaver. For the most part, those who write these programs--the people
> who will truly be writing and using the <video> tags--will be driven by what
> works in practice, not a statement in a specification that everyone ignores.

As someone who does write one of those programs, I have to say it's
remarkable how many people actually do want to pointlessly follow
meaningless requirements of specifications.  How many respectable web
app packages *don't* slavishly include things like type="text/css" so
that the W3C validator will magically bestow conformance upon them?
(Until the validator gets updated and they have to scramble to fix the
new pointless things it covers.)  How many have gone out of their way
to eliminate table-based markup, because they heard it was bad
practice -- and replaced it with divs and spans with presentational
classes?

I once did some quick manual hacking and determined that a MediaWiki
page could have the size of its HTML source cut by 5% or so if you
took out all the pointless stuff that XHTML 1 requires but HTML 5
permits you to omit.  And that's comparing the *gzipped* files.
Seriously.  Keep in mind that all browsers simply ignore this garbage.
 The pages were totally identical after parsing (unless I messed up).

That said, the above admittedly only applies if the specification does
actually work in practice.  But don't underestimate web app authors'
desire to conform to standards, even ones that make no sense or that
they don't understand.