W3C home > Mailing lists > Public > public-html@w3.org > July 2007

Re: 3.1 Introduction (Draft), review of

From: Robert Burns <rob@robburns.com>
Date: Mon, 16 Jul 2007 22:00:48 -0500
Message-Id: <27B9825C-F535-49E3-AB91-48AEEE2E14CA@robburns.com>
Cc: public-html@w3.org
To: Philip Taylor <philip@zaynar.demon.co.uk>

On Jul 16, 2007, at 9:44 PM, Philip Taylor wrote:

> Robert Burns wrote:
>> First, I think there's a danger of going into too much detail   
>> regarding optional tags. The only things I think might need to be  
>> in  an introductory section (maybe) are:
>> 1) that empty elements must have their closing tag omitted unless  
>> an  author uses the xml-style self-closing tag (e.g., <link />).
>> 2) that empty elements must be closed when using the xml   
>> serialization: i..e., either (<link></link> or <link />)
>> So to avoid this confusion and simplify things, it may make sense  
>> to  always recommend (or as far as this introduction goes, just  
>> gloss- over the difference so that authors use) the self-closing  
>> tag for  empty elements.
> Teaching authors about XML-style self-closing tags is also a cause  
> of confusion.

Just to clarify, when I wrote "empty elements", I meant canonically  
empty elements (i.e., elements required to be empty). Yes, I agree  
that encouraging the shortcut everywhere be a bad thing for the text/ 
html serialization. I don't think any of your following examples  
relate to that.

> It's fairly easy to find examples, many of which are likely to  
> result in unexpected DOMs after parsing:
> http://encarta.msn.com/encyclopedia_761579147/William_I_ 
> (of_England).html
>     <div style="clear:left" /><div style="clear:left" />
> http://www.rollingstone.com/reviews/movie/5948073/forrest_gump
>     <div style="clear:both;" />
> http://money.aol.com/savings
>     <div class="module colorFive" /><div class="header" /><h3>What  
> Will My Savings Be Worth?</h3></div>
> http://www.nmrestaurants.org/
>     <span id="a1"><span id="a1" /></span>
> http://www.princeton.edu/Siteware/Admissions.shtml
>     <a id="startContent"><span /></a>
> http://www.bible.org/series.php?series_id=72
>     <option value="Gen" />Genesis
> http://www.challenge.nl/
>     <p />
> http://www.alphanet.ch/
>     <p />
> http://www.paramotoraustralia.com/shop/
>     <place /><placename /><span lang="EN-AU">Byron</span></ 
> placename />
> All of those are served as text/html, and I don't think any were  
> anywhere near being well-formed XML. (I'm not sure what fraction of  
> pages have errors like that - the statistics get strongly distorted  
> by Microsoft Office 'HTML' documents.)
> Cases like <script src="..." /> are particularly nasty - see the  
> discussion around http://krijnhoetmer.nl/irc-logs/html-wg/ 
> 20070524#l-24
> Almost nobody (relative to the total population of authors) does/ 
> will use the XML serialisation of HTML4/5, and I would expect  
> anybody who uses it is already able to understand XML self-closing  
> tags without needing an HTML 5 introduction to tell them about it,  
> particularly since their XML tools will notify them whenever they  
> make a mistake.
> For HTML-serialisation authors, in either case (with-slash vs  
> without-slash) you would have to remember which elements have an  
> empty content model if you wanted to write correct code; but the  
> with-slash suggestion causes some confusion with XML and encourages  
> people to erroneously use slashes for elements that have empty  
> content but not an empty content model, which leads in some cases  
> to their code not working like they expected it to.
> The failure case when teaching someone that they must not use an  
> end tag for certain elements is that they will use one anyway, and  
> write "<link ...></link>" or "<embed ...></embed>", which is harmless.
> Teaching that the slash is optional, and that it is permitted on  
> certain elements so that it's kind of like XML even though you  
> actually can't use it everywhere like you can in XML, seems like  
> needless complexity, compared to simply saying that some elements  
> do not have end tags.
> (I think a table that lists for each element whether the start and  
> end tags are required/optional/forbidden, like the Elements  
> appendix in HTML4, would be the most effective way to tell authors  
> which tags are optional - the wording in the current HTML 5 spec is  
> too spread out and precise and hard to follow, and unsuitable for  
> authors who just want a quick reference guide. I'm unsure where  
> such a table should be put, though.)

Take care,
Received on Tuesday, 17 July 2007 03:04:25 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:24 UTC