- From: Braden N. McDaniel <braden@endoframe.com>
- Date: Wed, 23 Feb 2000 12:56:06 -0500 (EST)
- To: Jonny Axelsson <jonny@metastasis.net>
- cc: www-html@w3.org
On Wed, 23 Feb 2000, Jonny Axelsson wrote: > COST OF CHANGE > 1. ID selectors are unreliable today. Removing style is only an option when > today's bunch of browsers are gone. ID selectors are more widely available than XHTML browsers are, so I don't see the relevance of this. > 2. They would have to learn a new way of doing it. It shouldn't be too > hard, since they know .class {style}, #id {style} isn't too different. But > it is a switch from "Do *not* use #id (unreliable)" to "*Do* use #id". I don't buy this one. Web authors edge into using techniques that had previously been unreliable quite frequently. The very idea of using CSS at all is an example of this. > 3. The cost of change applies to UAs everywhere too, but this is not an > immediate concern for a designer (though it can be for a developer) It's already quite lucid that a switch to XHTML browsers will involve substantial investment from browser developers--this is not news. Are you suggesting that the absence of a STYLE attribute will significantly affect the size of this investment? > CONCEPTUAL MODEL > 4. Essentially the inline style divides styles into two categories, rules > in the head or preferably in a separate style sheet, exceptions inline. This is not a robust model, since it insists that all exceptions be persistent. Many, if not most, exceptions should be context-sensitive in how they behave. > Some designers (and particularily some visual editors) are not good at > making rules, for them everything will be an exception. Apart from > excluding them from the good parties, there is nothing you can do about > that. That they use style="" instead of selector{} is a symptom, not a cause. > > Yes, there will always be exceptions. I think I am very good at making > rules, but apart from a small class of documents like aircraft manuals, > documents will have parts with unique formatting you cannot generalize into > a rule (unless you count "a rule of one", which is essentially what an ID > selector is). For particularily unruly documents, the ID selectors will > crowd out the element and attribute selectors. STYLE affords no means of naming the styles it introduces, thus all such styles are always "persistent". This is weak, as even "rules of one" as you describe often need to be dependent on aspects of the context in which they occur. So few persistent styles are immune to this that I am not sure they shouldn't be done away with altogether. -- Braden N. McDaniel braden@endoframe.com <URL:http://www.endoframe.com>
Received on Wednesday, 23 February 2000 12:54:53 UTC