W3C home > Mailing lists > Public > www-style@w3.org > May 2014

Re: [geometry] DOMMatrix with DOMString argument

From: Dirk Schulze <dschulze@adobe.com>
Date: Mon, 26 May 2014 07:58:34 +0000
To: Tab Atkins Jr. <jackalmage@gmail.com>
CC: FX <public-fx@w3.org>, www-style <www-style@w3.org>
Message-ID: <96B0B541-FA82-4B06-8585-5F8B2C4D2F69@adobe.com>

On May 26, 2014, at 2:05 AM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:

> On Sun, May 25, 2014 at 12:10 AM, Dirk Schulze <dschulze@adobe.com> wrote:
>> On May 23, 2014, at 11:32 PM, Tab Atkins Jr. <jackalmage@gmail.com> wrote:
>>>> 2) If no Element was passed, what is the fallback for relative of percentage values?
>>> Generally speaking, these types of things are resolved against a 0
>>> length.  I suppose that em/rem/etc should be resolved against the UA
>>> default font and font-size.
>> I checked MSCSSMatrix and WebKitCSSMatrix. Both have a method called setMatrixValue which takes a DOMString as argument. This DOMString takes a <transform-list> in the CSS Transforms notation. Percentage length values and relative length values throw a SyntaxError exception for both engines. It is at least another possibility. I am not sure if UAs can access font data without a document even though it should be possible.
> Yeah, only accepting absolute values is definitely a possibility.
> There are potentially some relative lengths that might work - if
> there's a reasonable way to associate the call with some default
> viewport, it's possible that the viewport-relative units would work,
> for example.  However, that probably wouldn't work in a Worker, for
> instance.
> Thinking about it, it seems likely that any relative units that could
> be resolved to an absolute length without at least a document for
> context are probably actually just absolute lengths, and we'd refer to
> them as such.  So, it's probably fine to declare that only absolute
> lengths are allowed in the string arguments.

I changed the spec to the described behavior of MSCSSMatrix and WebKitCSSMatrix. (Both seem to have the constructor taking a DOMString  as well with the same behavior as 
setMatrixValue.) I close the issue for now. Anyone can feel free to raise it again and we can explore the Element argument further.


> ~TJ
Received on Monday, 26 May 2014 07:59:24 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:08:43 UTC