W3C home > Mailing lists > Public > www-style@w3.org > September 2015

Re: [css-transforms] 'transform-box' defaults in svg

From: Erik Dahlström <erik@xn--dahlstrm-t4a.net>
Date: Wed, 30 Sep 2015 16:36:14 +0200
To: Erik Dahlström <erik@xn--dahlstrm-t4a.net>, "Dirk Schulze" <dschulze@adobe.com>
Cc: "Amelia Bellamy-Royds" <amelia.bellamy.royds@gmail.com>, "www-style@w3.org" <www-style@w3.org>, "public-fx@w3.org" <public-fx@w3.org>
Message-ID: <op.x5r5iot1fwt8au@localhost.localdomain>
On Wed, 30 Sep 2015 14:43:25 +0200, Dirk Schulze <dschulze@adobe.com>  

>> On Sep 29, 2015, at 12:59 AM, Erik Dahlström  
>> <erik@xn--dahlstrm-t4a.net> wrote:
>> On Mon, 28 Sep 2015 21:44:16 +0200, Amelia Bellamy-Royds  
>> <amelia.bellamy.royds@gmail.com> wrote:
>>> The default `transform-box` for SVG needs to be `view-box` in order to  
>>> be
>>> consistent with current (SVG 1.1) behavior.  SVG transformations are by
>>> default relative to 0,0 in the userSpace coordinate system.  So the SVG
>>> default transform-origin is 0,0 and the default transform-box must  
>>> somehow
>>> work out as equivalent to view-box.
>> Which is fine up to the point where 'transform-origin' is explicitly  
>> set to something else, since IMHO that implies that the author wanted a  
>> different behavior. I'd think that what most authors would like in this  
>> case, is for rotate and scale transforms to operate on the element in  
>> an intuitive way, and not to have to explicitly override two different  
>> properties (transform-origin AND transform-box) in order to get that.
> Ah because the default is still 0 0 on transform-origin even though you  
> use transform-box: fill? You want that transform-origin switches to the  
> behavior of HTML automatically once you set transform-box: fill?

Not quite what I meant no. Let me try to clarify with an example:

If one sets 'transform-origin' to '50% 50%' on a <rect>, I would expect  
that to behave the same way an element in html would, which is to use the  
center point of that element as the origin for transforms, and that  
'transform-box' in that case doesn't default to something nonsensical like  

Would it be possible for 'transform-box' to use an initial value of 'auto'  
to defer the choice of what the computed/used keyword is, based on  
Received on Wednesday, 30 September 2015 14:36:44 UTC

This archive was generated by hypermail 2.4.0 : Monday, 23 January 2023 02:14:53 UTC