Re: Doctypes and the dialects of HTML 5

On Mar 25, 2007, at 12:38, Jirka Kosek wrote:

> Henri Sivonen wrote:
>
>>> In XSLT you are not able to output just:
>>>
>>> <!DOCTYPE html>
>>>
>>> This is serious limitation of HTML5 as amount of content produced by
>>> XSLT is enormous.
>>
>> It is a limitation of XSLT!
>
> Interesting. I'm quite surprised that many people coming from WHATWG
> camp are somewhat flexible with arguments.
>
> Once support for legacy stuff is treated as something to which HTML5
> must fits completely (e.g. <!DOCTYPE html>, <meta charset="...">).  
> Other
> time, 8 years old technology which is widely deployed (XSLT 1.0) is
> accused of not supporting HTML5.

Server-side technologies are upgradable (subject to cost). Deployed  
browsers are out there.

In general, changes that don't bring new functionality and are  
incompatible with existing *browsers* are not OK in the WHATWG.  
Requiring ocean boiling and rearchitecting server-side technologies  
is not OK. Requiring incremental changes to the server side is, in  
general, OK.

(XSLT is bit of a hot button. XSL-FO more than a bit. ;-)

> The question is what do will less harm (I'm talking only about HTML
> serialization):
>
> 1. Requiring <!DOCTYPE html>
>
> 2. Allowing <!DOCTYPE html>
>
> 3. Requiring either <!DOCTYPE html> or <!DOCTYPE html PUBLIC "...">

I'd be OK with spec text that makes a given PUBLIC id conforming,  
says that it doesn't map to any DTD, says that it exists for  
compatibility with tools that can't produce a doctype without a  
public id and says that those who don't suffer from such tool  
limitations should not bother to use it.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/

Received on Sunday, 25 March 2007 10:11:00 UTC