Re: The argument for |bugmode|

Chris Wilson wrote:
> Matthew Raymond
>>   Since a version number isn't sufficient, and Microsoft won't accept a
>> standardized attribute, what other option is there?
> 
> A version number (in the doctype, e.g.) will make the IE-specific opt-ins go away over time.

   If I write a valid HTML5 page, and it doesn't render correctly in IE
in spite of the fact that it should render correctly according to
standards, how does the version number help? In theory IE could use
Version 6 as a switch to turn on a mode with better standards support,
but if HTML truly needs versioning, this means that IE would NEVER
correctly support HTML5 because the specified nature of the markup may
change in HTML6. So my HTML5 page never renders correctly, and I end up
starting all over again with HTML6 hoping that Microsoft will correctly
implement THAT standard.

   If HTML, however, doesn't need versioning, then the version number is
 nothing more than a new quirks mode switch and I might as well just
have   |version="1000000000000"| in my document so that it always
triggers standards mode. (That's assuming that you don't implement the
switch to only work on HTML version 6, in which case HTML5-compliant
documents are really, really screwed.)

   Furthermore, all this assumes that significant IE releases are
synchronous with HTML spec versions, which may not be the case. Or are
we going to release HTML 5.999 just to accommodate all the versions of IE?

   The bottom line is that HTML versioning, if needed at all, has
nothing at all to do with Microsoft's comformance with W3C
specifications. If your implementation has a bug in its compliance with
HTML, then that by definition has nothing to do with HTML. Using a
system that makes it easier to write documents that take advantage of
non-conformant browser rendering and behavior is fundamentally counter
to the standards process.

   Furthermore, there are many people opposed to versioning, and some
like Håkon Wium Lie even question if versioning will ever even be used
correctly and consistently by authors. If they are, in fact, correct,
then adding a version is absolutely nothing more that putting in a
confusing, spec-condoned switch for Microsoft that has some serious
drawbacks. Why should the HTML WG involve itself with such a thing?

Received on Friday, 20 April 2007 02:46:20 UTC