- From: Sebastian Hennebrueder <usenet@laliluna.de>
- Date: Tue, 20 Jul 2010 10:41:47 +0200
- To: Aryeh Gregor <Simetrical+w3c@gmail.com>
- CC: www-style list <www-style@w3.org>
Am 20.07.10 01:15, schrieb Aryeh Gregor: > On Mon, Jul 19, 2010 at 3:47 AM, Sebastian Hennebrueder > <usenet@laliluna.de> wrote: >> This is leeds to a general question. How do we want browsers deal with CSS >> versions? I thought that it is simpler to create CSS rules, if you just need >> to do think like: >> >> I have CSS support for 2.1 so I can solve my look and feel this way but if I >> have support for CSS 3, then I can solve my problem this way. >> >> If every browser implements another combination of features, then in worst >> case you need to think like: >> >> 20 % of my visitors use a browser that supports x, y and z so I can create >> my look and feel with the following CSS rules and another 10 % have the >> combination of y, k and l so I need to use another approach. >> >> Many layout decisions are not based on a single CSS feature but requires >> many to build the complete CSS rules. You need a lot more browser specific >> knowledge to check for a reasonable set of support features. > > The only way this would work is if browsers all implemented all CSS > features in the same order. This might actually simplify things a > lot, and it would waste a lot less implementer effort, but it would > require much tighter coordination and central control over > implementation. It would also mean that problems would be more likely > to be uncovered once several implementations are already shipping > rather than just one, and it would be a hassle when a particular > feature is much easier for one implementer to implement than another. > > This is really up to implementers to decide on, but it's a radical > change and I don't see it happening soon to the extent necessary for > your approach to be even possible. > I know very little about the implementers side or the idea/strategy behind the CSS modules and releases. I am a software developer building web applications and CSS is only one of many building blocks. But I am fully aware that my proposal is not a quick fix but a fundemental change with a transition phase of many years. The intention is to make CSS development more stable and easier on the long run. From a developers point of view the current CSS causes a lot of difficulties. 5 different browsers implements 5 different combinations of CSS features. The consequence is that you cannot just write CSS rules following version 2, 2.1 or 3. It transforms CSS writing into a very complex expert job. There is no way to fix browser specific CSS problems inside of CSS without using browser hacks. There is no guaranty that the browser hacks will work in the future which causes a fair chance to fill the web with garbage. It is the same problem like all the websites optimized for browsers using non standard behaviour. I believe that CSS developers needs a clean way to deal with browser specific bugs in order to make the CSS more stable. I don't think that forcing use of JavaScript to load CSS or to configure the webserver to deliver different CSS depending on request headers is the right way. And I believe as well that developers need some control to validate if a feature or standard is supported. The situation is of course not simple nor fast to repair. We have connected elements here. If CSS rules are implemented completely, it is easier to write CSS rules. If CSS versions should be implemented completely in a single browser versions, the implementers need to take greater care and modify its release cycles. If CSS versions should be implemented completely, the W3 needs to write them in a way and timeline that implementors don't need years to implement or wait for them to be defined. Question to everybody: Is it useful, if CSS versions are implemented completely? Is it useful, if you can write CSS rules and use if conditions to check for supported standards, supported features or browser versions. Question to the W3: Do you want that CSS standards are implemented completely or is the current approach - browser foo implements x this year and y next year - perfect for you? A question to implementers: Could you imagine to deliver only completely implemented CSS versions for the future? -- Best Regards / Viele Grüße Sebastian Hennebrueder ----- Software Developer and Trainer for Hibernate / Java Persistence http://www.laliluna.de
Received on Tuesday, 20 July 2010 08:42:23 UTC