W3C home > Mailing lists > Public > www-style@w3.org > March 2010

Re: vendor prefixes considered harmful

From: Brad Kemper <brad.kemper@gmail.com>
Date: Wed, 3 Mar 2010 23:36:16 -0800
Cc: "L. David Baron" <dbaron@dbaron.org>, Sylvain Galineau <sylvaing@microsoft.com>, Aryeh Gregor <Simetrical+w3c@gmail.com>, Zack Weinberg <zweinberg@mozilla.com>, "www-style@w3.org" <www-style@w3.org>
Message-Id: <E8C9122C-DE8D-436C-A1C3-2A9229A8390E@gmail.com>
To: Daniel Glazman <daniel.glazman@disruptive-innovations.com>

On Mar 3, 2010, at 10:31 PM, Daniel Glazman wrote:

> Le 03/03/10 19:27, L. David Baron a écrit :
>> On Wednesday 2010-03-03 10:14 +0100, Daniel Glazman wrote:
>>> I said it in the past and I maintain my opinion: the vendor prefixes
>>> as we manage them in CSS don't make sense. We have
>>> border-radius and border-image interoperable enough so web authors copy
>>> precisely styles only changing the prefix. That gives stylesheets that
>> 
>> Well, in the case of border-image, Gecko and WebKit may have
>> interoperable behavior, but it no longer agrees with the spec.  So
>> we have a good bit of work to do before we rename -moz-border-image
>> to border-image.
> 
> But I make a strong difference between border-image and border-radius:
> the latter is already used by zillions of pages, not the former...

Enough are using -webkit-border-image to get iPhone-like buttons for iPhone-specific (or at least mobile-webkit-specific) pages, that Apple cannot just change it. Authors for iPhone pages are all using the old version (from before my changes), and would have pretty awful breakage if that change were just pushed out to the existing users of that property. I think there are quite a lot (not as many as border-radius, but the differences between versions are greater). Apple now has the newer version as un-prefixed in their nightly downloads, where they also still retain the old prefixed version. 

I think if we did a css working group prefix, we would also need to allow for some sort of versioning when something changed more significantly ('-wg-v1-border-image', '-wg-v2-border-image'), or even for some smaller changes (such as what percentages mean to 'border-radius'). This would mean authors would have to opt in to get later versions, but that's not too different from when we move from prefixed to non-prefixed, I suppose. It would also mean the continuance of a bunch of old versions floating around the Web, from authors who liked the old ways better. If we had '-wg-v1-border-radius', '-wg-v2-border-radius',  '-wg-v3-border-radius', and  'border-radius', would we ever be able to get rid of the older versions?

Received on Thursday, 4 March 2010 07:36:51 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:25 GMT