W3C home > Mailing lists > Public > www-style@w3.org > November 2011

Re: vendor prefixes: co-cascading

From: Brad Kemper <brad.kemper@gmail.com>
Date: Thu, 17 Nov 2011 20:26:44 -0800
Message-Id: <24B6C113-600B-43D1-98C8-1A4E97F4B177@gmail.com>
Cc: Brian Manthos <brianman@microsoft.com>, Alex Mogilevsky <alexmog@microsoft.com>, "www-style@w3.org list" <www-style@w3.org>
To: David Singer <singer@apple.com>


On Nov 17, 2011, at 4:05 PM, David Singer <singer@apple.com> wrote:

> 
> On Nov 17, 2011, at 15:55 , Brad Kemper wrote:
> 
>> On Nov 17, 2011, at 3:22 PM, David Singer <singer@apple.com> wrote:
>> 
>>> 
>>> On Nov 17, 2011, at 15:04 , Brian Manthos wrote:
>>> 
>>>> And doing that with CSS specs, and then multiplying *that* by the vendor-product-version-matrix-size is better?
>>> 
>>> I am suggesting that rather than seeing
>>> -mozilla-this
>>> -webkit-this
>>> -etc.-this
>>> 
>>> 
>>> all over our specs, it might be better to see
>>> -draft1-this
>>> -draft2-this
>>> 
>>> where the 'this' feature has changed.
>> 
>> I think changing the number would only be beneficial when syntax stays the same but the meaning changes. If the syntax changes, then authors could just include the old and the new with the same prefix, and wouldn't need to include -draft1-this through -draft50-this. 
> 
> well, if the name stays the same but the syntax changes, I'd need to write it differently, wouldn't I (and hence possibly twice)?
> 
> -draft1-this: 1cubit
> -draft2-this: proportion-of( furlong, 20%)

You'd need to write it twice to pick up old and new versions of the UA, but as long as the syntax was different enough, you could use the same prefix both times. Thus: 

-draft-this: 1cubit
-draft-this: proportion-of( furlong, 20%)

Or, if we had this already:

-draft-linear-gradient(bottom left, white, black)
-draft-linear-gradient(to top right, white, black)

The old UAs would only understand the first one, and the newer UAs would only understand the second one (or in Webkit's case, if the first version had to stick around forever, then they might have to parse both, and rely on the cascade to get the second one). It beats multiplying these by the number of UAs to support, and it beats having to add a new one to the style sheet every time the draft changes a little bit.


> and if the semantics changes, likewise; if the length changes from being how far you want the grenade thrown, to how far you need to back off to be safe, then that's a big change, and the author would express the length differently and might need -draft4-this and draft5-this.

Yes, that is when you'd want numbers. As with this:

-draft-linear-gradient(90deg, white, black)
-draft2-linear-gradient(0deg, white, black)

Still better that numbering them into the double digits. 

Of course, the linear-gradient boat has already sailed, and it is too late to do this with that. 

And also, these examples are shown in isolation; for the sake of the examples being shown on the same page, you have to ignore that the degree meaning change occurred before the 'to <corner-or-side>' change. So they aren't ideal examples in that way, but hopefully you get the idea anyway.
Received on Friday, 18 November 2011 04:27:23 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 17:20:46 GMT