[whatwg] Graceful Degradation and Mime Types [was: trailing slash]

[Mike Schinkel]

Sam Ruby wrote:
>> I think that it is fair to say that if there is any benefit to 
>> my serving my content as application/xhtml+xml, I get it.  
>> I get the "benefit" of draconian error processing.  I get the 
>> benefit of being able to embed both SVG and MathML on 
>> my blog and on my planet.

That brings up another thought.  Consider a blog/wiki/forum/cms/whatever
that allows markup written in either XHTML or HTML5 (let's consider both
cases separately.)  All of those web apps allow users to post comments. And
many of the genre allow users to embed (a subset of) HTML to format their
documents. 

Considering that the average person is not a web professional and is likely
very lucky to just understand the basics of HTML let alone understand all
the gooble-de-ga of HTML 4.01 vs. XHTML 1.0 vs. HTML5 vs. XHTML 2.0 vs.
whatever. And of those who do know the basics, they learned them taking
varying paths, i.e. some self taught, some had the blind leading them, and
some had didactic developers indoctrinating them in ways that conflict with
the lessons of other didactic developers.  

So what we have is people who will be contributing content to an app where
their markup needs to be integrated into the markup of the application.  And
these people will likely have no clue which version of (X)HTML the site they
are contributing to is running, and even if they did there is huge chance
they will have no clue what the difference means in terms of the syntax
required for the markup they need to contribute. Which means they will
contribute to a (potentially) conforming site markup that does not conform.

So what are the various answers to this solution?

-- Have the apps tell the user contributing how to clean up their
contribution?  Nah, because most developers of these web apps won't include
such validation then instruction, and most users wouldn't understand it if
they did.

-- Have the apps convert to the proper format. Possibly, but unlikely.  It
is only likely if good and liberally-licensed open-source implements for
*all* significant platforms are provided day one and they are referenced in
the spec in a manner than makes them impossible to ignore even by the most
clueless web developer (although how to ensure the clueless get a clue, I
have no suggestions.)  So this option is not very likely to produce good
results.

-- Let HTML5 be the lax superset of XHTML.  Everything works great, users
contribute what they know how to contribute, and there is no problem.

Are there other options?  And if so, please consider with both XHTML and
HTML5 conformaning systems.

-Mike Schinkel
http://www.mikeschinkel.com/blogs/
http://www.welldesignedurls.org/