Re: [SVGMobile12] Comments: Basic Data Types and Color Keywords

On Tuesday, November 29, 2005, 6:47:54 PM, Boris wrote:

BZ> Chris Lilley wrote:
>> The intent is that CSS units MUST be supported for width and height on
>> <svg> and MUST NOT be supported elsewhere.

BZ> This seems to me to place an undue burden on implementations that DO
BZ> have support for CSS -- instead of using the same code for both sets
BZ> of attributes, they are now forced to write separate code to handle
BZ> the two.

How many overlapping sets of attributes are we talking about here?

BZ> Also, since CSS-less implementations still have to handle units for
BZ> width/height on <svg>, I don't see how their life becomes any
BZ> easier... In fact, they have to have two separate codepaths instead
BZ> of just one, which seems unfortunate to me if they're trying to
BZ> minimize codesize.

BZ> In short, from my point of view as an implementor I see no benefits
BZ> to this approach and lots of drawbacks.

BZ> What am I missing?

Well, I could invite you to try implementing SVG where fixed, real-world
units like mm and in can be mixed with lengths in a coordinate system,
and then try to implement a nested transformation coordinate system. But
people already tried that early in the 1.0 timeframe.

Feedback from those implementors was that it made the code extremely
complex, slow, produced non-intuitive results, and was very rarely
exercised (since the entire point is to make scalable graphics; graphics
that are semi-scalable are more complex) so was more likely to be buggy.

In the real world, the number of times that someone wants to make a
rounded rectangle that is scalable except that the y-radius of the
rounded corder is 2em is very small.


-- 
 Chris Lilley                    mailto:chris@w3.org
 Chair, W3C SVG Working Group
 W3C Graphics Activity Lead
 Co-Chair, W3C Hypertext CG

Received on Tuesday, 29 November 2005 18:00:17 UTC