W3C home > Mailing lists > Public > public-html@w3.org > June 2009

Things in HTML that I disagree with (Was: evidence of harm)

From: Ian Hickson <ian@hixie.ch>
Date: Thu, 25 Jun 2009 20:32:45 +0000 (UTC)
To: Shelley Powers <shelleyp@burningbird.net>
Cc: public-html@w3.org
Message-ID: <Pine.LNX.4.62.0906252015120.16244@hixie.dreamhostps.com>
On Thu, 25 Jun 2009, Shelley Powers wrote:
> Ian has mentioned that parts of the HTML5 specification have been 
> "edited" by others. I have to wonder, though, if there has been a case 
> where Ian has disagreed with the contents of the edit--strongly 
> disagreed, as he does with topics under discussion, such as 
> @summary--but still incorporated it into HTML5?

The text of HTML5 was all written by me, but I do disagree with several 
parts of it. These include, but are not limited to:

 - The allowance of /> syntax on certain tags, e.g. <br/>, in the
   text/html syntax

 - The allowance of the style="" attribute on all elements

 - The inclusion of <div> as a valid element

 - The use of headers="" on <th>

 - The inclusion of the microdata section

 - The semantics of transforms affecting paths in canvas

 - The allowance of "xmlns" attributes all over the place

 - The messy situation regarding parsing <script> in SVG

 - The handling of document.write() in <script> in SVG

The list goes on; this isn't by any means an exhaustive list. In each of 
these cases, the spec says something that differs from what I'd like it to 
say, because arguments were made that showed that my opinion isn't the 
most practical solution.

(For example, I'm not convinced that we need to support the use case of 
annotating pages so that they can be reused, but it has been demonstrated 
that people clearly _do_ want to do this, so that logically we should 
support it, even though I think it's not an interesting problem. 
Similarly, I am not convinced that we should ecourage authors to be 
writing tables so complex that headers need their own headers in ways that 
scope="" can't handle, but data was collected showing that people are 
going to write such tables whether they can make them accessible or not, 
so the spec supports those cases, even though I think that this is not the 
best solution for those tables.)

The complete list would in fact be huge; requiring compatibility with 
legacy documents forces us to make many compromises that I personally am 
unhappy about, but that haven't even been discussed because the backwards- 
compatibility principle is overriding. (e.g. location.search is a 
ridiculous name for that API, but there's no point considering renaming it 
to location.query since that would break millions of pages.)

> I could be wrong in both regards, though. The HTML5 specification could 
> contain sections to which Ian strongly disagrees

This is indeed the case. I tend not to make much noise about those 
sections because once an argument has been presented explaining through 
reason or based on research why I am wrong, I move on.

Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'
Received on Thursday, 25 June 2009 20:33:22 UTC

This archive was generated by hypermail 2.4.0 : Saturday, 9 October 2021 18:44:49 UTC