[dojo-contributors] 2D Graphics Recap

Tom Trenka ttrenka at gmail.com
Sat Jun 10 18:41:59 EDT 2006

Going to do just a *tiny* bit of clarifying here...

On 6/9/06, Gavin Doughtie <gavin at dfmm.org> wrote:
> I think there's several things being discussed and thought we might do a
> little refactoring so I don't stump for the wrong functionality at the
> wrong time:
> 1) Graphics for declarative widget authoring. This is Tom's proposal to
> declare widget graphics in dojo-decorated SVG and have an XSLT-driven
> translation layer that will create equivalent VML+JS in IE.

Clarifications: I am *not* planning on trying to use JS to make up for
VML's shortcomings; any script I'd be using would be within the
translation process itself.  The end result will be pure VML; there
won't be any JS library that you'll be forced to include as a part of
the transformation to make up for limitations.

I'm going to be trying to write this so that it actually isn't
Dojo-specific at all; I don't have to be explicit about preserving
Dojo-specific attributes, I'll just be copying them over en masse.
Hopefully it means that anyone trying to use SVG can simply include
the XSL with a couple of lines of JS code for VML translation.

(Also, just for the record, it's a joint proposal and the original
suggestion came from Jon Ferriaolo of IBM, who is one of the major
movers on the SVG specification)

> 2) Retained-mode Javascript objects, suitable for procedurally
> generating complicated charts or for use in a drawing editor.
> 3) An immediate-mode API
> These systems would cross over at various points:
> a) The immediate-mode API would produce retained-mode Javascript objects.
> b) The Javascript objects could be instantiated/retrieved via a "get"
> API on widgets declared in markup
> (probably some I'm forgetting)
> The applications we're thinking of fall into layers (lightest-weight to
> heaviest):
> 1) Widgets/UI
> 2) Charting Library
> 3) Drawing Editor

I'd suggest that 2 and 3 should probably be switched, but that's a
really minor point ;)

> That roughly correspond to the application areas Tom, Eugene and I are
> personally interested in. (Please forgive me if I've misread what you're
> trying to make sure the API can accomplish).
> I'm all jet-lagged and fuzzy at the moment, but I'll be on #dojo and IM
> most of the day tomorrow and possibly even later on tonight. I hope that
> we, and anybody following quietly, can solidify the layers of this
> system we want to build.

The only other thing I'd mention is that if we end up trying to go
down a specific, complex path (such as creating a meta-vector
declarative language), I will probably speak out pretty hard against
it; we need to keep usage KISS in mind for this, as it has the
potential to get out of hand fairly quickly.  And with that warning
I'll shut my trap :)


More information about the dojo-contributors mailing list