Le 09/11/11 02:39, L. David Baron a écrit :
They have the choice not to use the prefixed properties. Given that they've made that choice to do so *despite* their complaints about it, I suspect they might not like a solution that takes away their ability to make that choice.
No they don't. Prefixed properties offer features that are visible in apps like iTunes or proeminent web sites that adopt them as soon as they are shipped. Releasing a web site is not only releasing a webapp, it's also *almost always* about releasing a web site that is nicer and cooler than the other ones (just as a reminder, I used to be the CTO of a Web Agency). Don't ask people working in Communication to avoid using cutting-edge tech... The only way to solve the problem is to ship the feature but not enabled by default OR to ship the feature only in test builds. Ignorants will say it slows the Web, others will say it prevents the border-radius/gradients mess from happening again. I'm sure we can explain this well to the masses.
(I agree with the feedback that we need to be better about standardizing high-demand features quickly. I think we can do that by keeping them limited in scope and not adding and stabilizing every addition anybody asks for before moving to CR. In both of the cases you mention, the group resolved to advance quickly and the editor then went through all the comments made on the spec and added a bunch of requested features, delaying advancement to CR. Those additions would be better made by developing the next level in parallel.)
That is not enough. Gradients were shipped almost the day they were proposed to the CSS WG. Such a feature cannot not trigger a huge response from the Web Authors' community. Gradients were shipped with a prefix, but as a 'production' thing, and *that* is the root of the problem. </Daniel>
