W3C home > Mailing lists > Public > www-style@w3.org > January 2008

Re: Re[8]: css with attribues (was: Re[2]: [CSS] Would be nice if...)

From: David Dorward <david@dorward.me.uk>
Date: Tue, 15 Jan 2008 13:57:01 +0000
Message-Id: <6447C6D9-6542-4FDA-ABE2-FC8617EFB80E@dorward.me.uk>
To: CSS Style <www-style@w3.org>

On 14 Jan 2008, at 08:38, Dmitry Turin wrote:
>>> DD> It also needs to implement media types
>>> Strongly disagreed !
>>> Now we don't assign or not assign attributes depending media types.

> I propose, that attbitutes are considered as assigned by css
> _for all media_. Clarify, how is this proposal make barrier
> for debug.

I misunderstood what you are trying to say, and, while that last  
paragraph clearly shows that what I thought you meant was wrong, I'm  
still not sure what you actually mean.

Do you mean that they should be treated just like any other property?  
i.e. that an attribute can be assigned based on the media type of the  
stylesheet? If so, then my original point stands, the client would  
have to be able to handle HTML and CSS media type syntax.

or do you mean that an attribute specified in a screen media  
stylesheet should apply to all media? (which would be horribly  

>>> Re-rendering during downloading.
> DD> I don't understand what you are trying to say here.
>   Imagine, that you adjust browser to not download pictures.
> <snip>

So we go back to my earlier point. Some user agents are already  
geared up to handle a document that is adjusted on the fly based on  
other files (such as images, and stylesheets) - but others are not  
and adding this feature would require that they greatly complicate  
themselves in order to cope.

> DD> I don't understand what you are trying to say here.
>   There exist at least two types of attributes:
> 1) values of which have visual effect (e.g. @cellspacing)
> 2) values of which change functionality (e.g. INPUT/@type)
>   Attributes of first type are not differ from properties.
> Attributes of second type are small part of document objects,
> and they are not burden for browser.

If the user agent (please stop using the term 'browser', there are  
plenty of non-browser user-agents out there, and if semantics start  
showing up in CSS then they can't be ignored as far as the  
development of CSS is concerned) isn't parsing the document to an in- 
memory representation, and then modifying it based on factors such as  
the stylesheet, then this WOULD add a burden to it.

>>> DD> It is significantly more complicated then simply parsing HTML.
>>> This depend of nature of browser engine - to what is it more  
>>> similar:
>>> to end-to-end linear transformation of html-text into screen image;
> DD> Screen image? There are plenty of user agents which don't  
> output to
> DD> the screen. GoogleBot springs to mind.
> I expect, that we are saying about browser.
> In general case, phrase sounds
> "end-to-end linear transformation of html-text into database of UA".

That is one use case, and a very real one.

David Dorward
Received on Tuesday, 15 January 2008 13:57:32 UTC

This archive was generated by hypermail 2.3.1 : Monday, 2 May 2016 14:27:33 UTC