Re: agenda+ SVG 2 Features and Approach

Hi,
> 1. Form the intersection of superpath, vector effects and connectors, preserving basic topology of orientation. (Markup-based relations between DOM objects need to be known, whether in SVG or in HTML5 or both)
> 2. declarative drawing (including at least some of)
> 	a. replicate
> 	b. extensions of <path> that include classic turtle graphics, parentheses and iteration), and relative coordinates
> 3. extended gradients (including at least some of)
> 	a. replicate
> 	b. contours
> 	c. diffusion curves
> 4. Extensions of the timing module of svg <animate> to include the ability to draw upon functional input from multivariate sources (rather than just keysplines and keytimes --as with <replicate>), e.g.,  from multiple curvilinear 2D paths.
> 5. enhanced text support 
> 	a. to allow multiple alignment of text along multiple lines: base, top and middle
> 	b. to allow glyphs to be squished into convex polygons
> 6. extensive support for randomization (random attribute values, etc)
> 7. alternative transforms (mesh, warp, spherical, conic, Mercator)
> 8. enhanced browser consistency in basic SVG interface
> 	a. consistency of zoom and pan (give us a widget if you can't agree!)
> 	b. a drawing widget 
> 		Rationale: HTML (hyperTEXTmarkuplanguage) has <textarea>
> 			<textarea> comes armed with word-wrap, backspace/delete, newline/return, cursor insertion, 							selection/replacement, copy/paste (&rich text attributes?)
> 		Donc/therefore: SVG (svGRAPHICS) should have <tablet>
> 			<tablet> could come armed with basic drawing tools (polygon, Bezier, select, delete, copy/paste
> 				Drag, resize, rotate) *
> 	c. layers control (as with Photoshop layers)
> 9. non rectilinear layout models -- (consider how often graffiti flows into rectangles)
> 10. magnetism/gravity that allows labels of geographic entities to draw their labels (is not difficult in today's computational environment since most of 2D physics avoids undecidability)
> 11. 2.5 dimensional effects 
> 	a. local z-index --if global z-index is computationally impractical (think knot theory and planar four regular graphs)
> 	b. <replicate>
> 	c. perspective transforms (I think 2.0 already has these)
> 	d. global z-index (I get the sense jwatt is doing this?)
> 12. modeling translucency as beyond transparency 
> 13. SMIL data feedback (as per STELLA circa 1986 on Macs -- see http://www.iseesystems.com/softwares/Education/StellaSoftware.aspx )
> 14. accessibility: no it's not really an afterthought! What does it mean for geometry to be accessible? Let's think about it!
> 
> Fourteen is a nice number, but of course, I might like some time to think about the subject.

I like many of these features, but we should think about how hard it would be to implement them in viewers. Also, most parts of SVG are animatable either directly with SMIL / CSS, or indirectly with JavaScript. Some of these features would need direct pixel manipulation, because of the lack of functionality in current graphic libraries (CG, Skia, Cairo, Qt for example). This might cause bigger performance problems. Just take the animated masks of Eriks 'time travel' as example [1]. I couldn't check Internet Explorer, but most of the other viewers don't perform very well on this example.

Dirk

[1] http://my.opera.com/MacDev_ed/blog/2008/05/18/timetravel