- From: Tab Atkins Jr. <jackalmage@gmail.com>
- Date: Fri, 22 Apr 2011 12:34:07 -0700
- To: Alexander Shpack <shadowkin@gmail.com>
- Cc: www-style <www-style@w3.org>
On Fri, Apr 22, 2011 at 11:29 AM, Alexander Shpack <shadowkin@gmail.com> wrote: >> >> This has been discussed before. It suffers from the fact that, when >> browsers differ from the standard in their experimental >> implementations, they quite commonly differ in slightly incompatible >> ways. Folding them into a single property so that you can't target >> each browser with different code means that the prefix is somewhat >> unusable. >> > > My proposition didn't cancel current behavior. > > Possible 3 statements: > * Spec is incomplete, UA support it in testing mode. > * Spec is complete, UA support feature in standard mode. > * UA feature, no spec available. > > Developers will use something: foo; in second case. > Otherwise -ua-something: foo; in third case. > And -something: foo; for first one; (use - instead *), because every > UA should be as closer as possible to spec, is not right? I just gave the reason why a unified prefix isn't acceptable for #1. For a relatively recent example, the background-clip property is specified to take the value 'padding-box'. Most browsers accepted that in their experimental implementation, but FF 3.6 instead accepted 'padding' (I don't know if this was an accident, or if the spec changed underneath them). > I do not want to change my CSS every 3 month, because UA behavior or > Spec was changed. > CSS hacks legalization is very bad idea, IMO. If you don't want to change your CSS, don't use experimental properties that are guaranteed to change. Stick to tested, stable, unprefixed properties instead. Vendor prefixes are meant to expose the property for experimentation, not stable production use. ~TJ
Received on Friday, 22 April 2011 19:34:48 UTC