W3C home > Mailing lists > Public > public-fx@w3.org > January to March 2012

Re: [css3-transforms] skew(<angle>[, <angle>]) (again)

From: Dirk Schulze <dschulze@adobe.com>
Date: Thu, 8 Mar 2012 10:20:27 -0800
To: "Dr. Olaf Hoffmann" <Dr.O.Hoffmann@gmx.de>
CC: "public-fx@w3.org" <public-fx@w3.org>
Message-ID: <C4ADC064-7771-483F-AF81-79D6D54671B4@adobe.com>

On Mar 7, 2012, at 5:10 AM, Dr. Olaf Hoffmann wrote:

> Hallo,
> 
> surprisingly in the new draft the transformation skew(<angle>[, <angle>])
> appears again. 
> http://www.w3.org/TR/2012/WD-css3-transforms-20120228/#two-d-transform-functions
> 
> It was already discussed and clarified, that at least the naming is 
> misleading, because the formula, that defines this behaviour now is
> a mixture of scaling and skewing (in general the determinant is not
> 1 as for skewing necessary).
> Therefore this transformation was already removed from a previous draft
> ( http://lists.w3.org/Archives/Public/www-style/2009Mar/0287.html
> and following comments).
> 
> At least now there is a formula at all for this kind of transformation,
> but it is still not related to skewing or shearing - no progress in the past 
> years finally, looks like a regression, this problem was already known as
> the reason, why this transformation was removed.
> 
> Therefore I propose to do one of the following actions:
> a) If there is no indication, that such a transformation is of practical use, 
> remove it again, because is is useless.
skew is different from skewX * skewY (which might make sense), there seems to be no mathematical reason why we have skew. So I agree, it seems not to be very helpful.

> b) If there is an indication, that such a transformation is of practical use,
> rename it, for example to something like 'scaleAndSkew',  'scaw',  'warp',
> 'distort', 'noSkew' or 'weks' to avoid further confusion and obfuscation.
Renaming is no option. It is implemented by a lot of browsers already.

> c) Replace the formula with something more useful for authors, for example
> a skewing in an arbitrary direction.  This would be comparable with the
> option of rotate to provide the rotation center. Here one could rotate
> the intended axis in the x-axis, do a skewX and rotate everything back
> again.
> This was already proposed as a meaningful interpretation of such a
> skew transform with two angles:
> http://lists.w3.org/Archives/Public/www-style/2009Dec/0124.html
Even if that makes more sense, this is against what current implementations are doing at the moment.

I would agree with Olaf and Aryeh that we could remove it from the spec again. Renaming or redefining seems not to be practical at this point of time. Most CSS WG members were pretty clear that the specification should just specify current behavior of implementations. Maybe we can come up with new transformation functions in the next level.

Greetings,
Dirk
Received on Thursday, 8 March 2012 18:21:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 8 March 2012 18:21:14 GMT