[whatwg] rel/rev for <form> ?

[Lachlan Hunt]

Henri Sivonen wrote:
> Of course, the UA does not care about "semantic" class names. In both 
> cases, the UA only sees opaque strings that can be tested for equality 
> with strings present in CSS selectors.

Technically yes, without a profile, they only have author-defined 
semantics which are unknown to anyone else, and I wouldn't expect a 
typical UA to be able to do anything with them.  So, you could write 
<code class="red"> and define that to mean a snippet of HTML markup and 
style it appropriately (not necessarily red), but that wouldn't be very 
clear to anyone else, nor easy to remember.  At least something like 
<code class="html"> is easy for yourself and others to work with.

> The class names in the latter case may be "semantic" in the private 
> universe of the author, but they do not communicate semantics to 
> software developed by someone else without a prior agreement (possibly 
> in the form of a third-party spec) on the meaning of the class names.

With meaningful class names, there doesn't necessarily have to be a spec 
defining them to be useful to anyone else.  There's nothing stopping 
anyone looking at your source code and writing a user style sheet or 
user JS that works with your code.  Without meaningful class names, 
doing that may be harder for the user.

> As far as the UA goes, the "semantic" class names could be translated 
> into Finnish or into Elvish or be replaced with unique random strings.

Indeed they could, and they would have exactly the same formally defined 
semantics as an english class name: none.  The usefulness of a class 
name to others, though, is limited to those that understand or can 
roughly determine its semantics.

> So in the end, home-grown class names are just style hooks when observed
> outside the private universe of the author.

Also useful for scripting hooks.

-- 
Lachlan Hunt
http://lachy.id.au/