Re: Selectors, vendor prefixes (again...) and IE9

On 5/16/10 2:57 PM, François REMY wrote:
> Then, we run into the CR => PR => R problem, where some
> implementations are needed to move on. Well, I see no reason
> why we couldn't implement a feature in nightlies (or Beta, if
> we speak about IE browsers) and don't compile it when we're
> in the RC/RTM stage

Our experience with Gecko is that nightly feedback is wholly 
insufficient to discover web-compatibility problems.  Beta feedback is 
almost good enough.  An actual release is getting there.  Typically we 
get a whole slew of web-compat regression reports for release N when we 
start dropping support for release N-1 and all the people who don't 
update their browser until they have to start upgrading to release N.

Or put another way, I don't think we could provide the implementation 
feedback needed to move out of CR without at the very least shipping the 
feature in an RTM product.  Nightlies definitely won't cut it; betas 
almost certainly won't.

> Vendor prefixes were a solution to a problem. It's not because
> it is a solution that's the best solution. We now see it isn't.

Perhaps; I have yet to see a better solution proposed.

>> Let us try it. But then where does it go ? Are you suggesting taking
>> CSS3 Selectors back to WD ?
>
> It would be a problem if every UA should prefix features of the CSS 3
> Selectors module if they decide it should go back to WD. This solution
> is clearly unacceptable. So, it removes UA's the ability to say "a part of
> this document is not good at all, we need to return to WD until it's
> fixed".

If it came to that, I would think UAs would have no problem requesting 
that the document go back to WD and ignoring any unofficial policies 
(which is all they really are) there may be about prefixes in that 
situation.

-Boris

Received on Monday, 17 May 2010 03:12:10 UTC