Re: handling fallback content for still images

On Jul 5, 2007, at 5:33 PM, scott lewis wrote:

>
> On 5 Jul 2007, at 1602, Thomas Broyer wrote:
>
>> 2007/7/5, scott lewis:
>>
>>> HTML5 is a language with two serializations (I'll call them): HTML/
>>> xml and HTML5/html. These are both representations of the same
>>> document. Both serializations of a document must parse identically,
>>> otherwise they aren't serializations of the same language. There  
>>> is a
>>> simple test to ensure that: take a document in one serialization,
>>> parse it, generate the other serialization from it, then parse the
>>> other serialization and require the parsed documents are identical.
>>>
>>
>> ...with the exception of <tbody>'s in <table>'s (are there others?).
>>
>> Converting this XHTML fragment:
>>    <table><tr><td>Cell</td></tr></table>
>> to HTML and then back to XHTML will produce:
>>    <table><tbody><tr><td>Cell</td><tr></tbody></table>
>> except if your converter is able to omit the <tbody> in the XHTML
>> re-serialization because it's the only child of the <table> (it means
>> that you're not just parsing and serializing a DOM tree).
>>
>
> I think you're confusing the serialized bytestream with the HTML5  
> document. You must compare the output of your parser (which may be  
> a DOM tree or some intermediary form -- it's entirely an  
> implementation detail) not the serialized form. There are a number  
> of variations in the serialized form which are normalized by the  
> parser.

I'm not sure if Thomas is confused. There is certainly an issue that  
our recommendations should deal with. In other words when serializing  
as XML, should a translating UA include explicit <tbody> elements  
when serializing to XMl? There may problems with doing so, but there  
will also be problems with not doing so. For example, a user may  
wonder why the CSS stopped working simply from saving to a different  
serialization.

Take care,
Rob

Received on Friday, 6 July 2007 22:45:36 UTC