Re: [css3-values][css3-background][css3-transforms] <position> vs. transform-origin

On Mar 23, 2013, at 2:35 PM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:

> On Sat, Mar 23, 2013 at 9:52 AM, Simon Sapin <simon.sapin@exyr.org> wrote:
>> Le 20/03/2013 23:11, Tab Atkins Jr. a écrit :
>>> On Wed, Mar 20, 2013 at 7:23 AM, Øyvind Stenhaug <oyvinds@opera.com>
>>> wrote:
>>>> The transform-origin value type is different from that of
>>>> background-position in that a third component value represents a position
>>>> on
>>>> the Z-axis rather than an offset to a preceding percentage/keyword.
>>>> (<http://lists.w3.org/Archives/Public/www-style/2012Mar/0195.html> has a
>>>> resolution concerning this.)
>>> 
>>> I wish we could solve this. :/  The grammatical ambiguity is annoying.
>>>  It would probably be worthwhile to define a <3d-position> type,
>>> though, that just matches transform-origin's current grammar.
>> 
>> Using '<position> <length>?' would be ambiguous, right? Even with tricks
>> like "parsing <position> should consume as many tokens as possible"?
> 
> With a trick like that, it's not ambiguous.  But we try to avoid those
> kinds of things when possible.

We thought about that and realized that it might not be possible to identify the current third value easily since if will be interpreted as offset.

> 
> More importantly, specifying something like that wouldn't be backwards
> compatible.

Correct. A huge amount of content definitely will be broken after that change. Not good for implementations that already unprefixed 'transform-origin' :/

> 
>>>> That difference doesn't apply to perspective-origin, though, so that does
>>>> seem to be an issue. Has been reported earlier here:
>>>> http://lists.w3.org/Archives/Public/www-style/2012Jan/0964.html
>>> 
>>> Looks like it got forgotten in the craziness of that thread.  It would
>>> indeed be nice (and backwards-compatible) to upgrade
>>> perspective-origin to a <position>.
>> 
>> I read half of that thread but I’ll probably stop here :)
>> 
>> Can we ask in the next conf call to have perspective-origin reference Values
>> 3 for <position>, and leave the discussion of what <position> should be for
>> another time?
> 
> It's an editorial change, but it would probably be nice to inform the
> WG anyway. ^_^

To my knowledge, no implementation supports three of four values for 'perspective-origin'. I do not see a problem with backward compatibility doing that. However, this is not just an editorial change, but needs a resolution of the WG. Again, we already have unprefixed implementations and that requires to discuss changes like this on the WG call.

Greetings,
Dirk

> 
> ~TJ
>

Received on Sunday, 24 March 2013 16:04:38 UTC