- From: Andrew Fedoniouk <news@terrainformatica.com>
- Date: Fri, 09 May 2008 10:09:27 -0700
- To: Bert Bos <bert@w3.org>
- CC: www-style <www-style@w3.org>
Bert Bos wrote: > > Andrew Fedoniouk wrote: > >> Speaking about functions. set-focus() for example can be useful for >> implementing <label for> functionality but for any type of element. > > Setting focus to some element in a Web page is a bad idea. It means that > the keyboard works differently depending on whether the page has a > focusable element or not. Bert, I am not sure I understand you well here. In my sample I've reproduced standard functionality of <label for>. It is not anyhow different from what UA already do. > > In a typical browser, keys like the arrows, the tab and the space bar > navigate through the page. Some browsers offer even more handy keys, > e.g. to jump to the next <Hn> element. It would be very confusing for a > user if the keys worked in some pages and not in others. Bert, have you seen such thing as @autofocus? http://www.whatwg.org/specs/web-forms/current-work/#autofocus0 All that nice wording under the link above is just one declaration: [autofocus] { when-assigned: self.set-focus(); } I mean that such things shall really be a part of default style sheet of HTML5 but not to pollute HTML5 specification. That behavioral styling can be applied to many input languages - not to only HTML. > > It's different in an application (such as a "widget"). Whatever UIDL is > used to design the (G)UI of that application, one would hope that it > provides control over the initial focus. > > CSS is meant for layout of documents, not for (G)UIs. The assumption > behind style sheet languages is that the document viewer (browser) > already has a UI and we neither need nor should interfere with that UI. Beg my pardon but that is a bit idealistic. This simple behavioral (sic!) declaration: div { overflow:scroll; } will make all DIV elements a) focusable and b) will change (G)UI. > > Turning CSS into a language that is both a style sheet language and a > UIDL can only lead to an ugly language that is neither very well. > CSSS! is not a procedural language in the common sense. It is rather declarative - it defines set of triggers. And it is not a replacement of general procedural/imperative languages like JS and the like. One of goals of proposed CSSS! is to solve some fundamental limitations of CSS selectors. Some desired CSS selectors like (:has-child(), :root-has-child()) may lead to combinatorial explosion when they are used as pure static declarations. > > > Bert -- Andrew Fedoniouk. http://terrainformatica.com
Received on Friday, 9 May 2008 17:10:09 UTC