[dojo-contributors] SVG/VML/Canvas API discussion

Bill Keese bill at dojotoolkit.org
Mon Apr 17 21:27:19 EDT 2006


Hi Tom,

Thanks for handling this, and also for doing the bug triage.

There's a guy named Gavin that promised to do this API but I haven't 
heard anything from him for a while; maybe he lost interest.  Here's the 
link:

http://thread.gmane.org/gmane.comp.web.dojo.user/5504/focus=5726

(this is why I made that InProgressProjects page on the wiki)

As for my comments: I'm not a graphics expert, but I wonder whether we 
should support the canvas API model (like you suggested) or the SVG 
model.  The difference being that in SVG there's a handle to every 
object you draw, so you can attach event handlers / erase previously 
drawn objects, etc.  Presumably canvas is faster but SVG/VML are more 
powerful.

Bill

Tom Trenka wrote:
> Hi all,
> 
> Since the Iterator discussion went *so* well (<sarcasm/> but not 
> really), I figure it's time to start asking questions about an 
> SVG/VML/Canvas library, since it would seem that some think this would 
> be a very useful thing.  So let me kick off the discussion (since I'll 
> probably write it)...
> 
> The nominal purpose of an SVG+VML+Canvas library would be to add a 
> single layer of abstraction over the two vector (and one bitmap) drawing 
> languages so that a developer can write one set of code, and have it 
> render properly on the major browser platforms.  However, I have some 
> issues with this, so let me lay this out first before diving into a 
> request for purpose+proposal.
> 
> 1.  My personal feeling is that widgets (one of the core uses for 
> something like this) should not be limited to an encompassing API for 
> drawing, and that there's no real difference (aside from platform) 
> between writing markup in VML and SVG and writing markup in HTML.  All 
> three are valid, existing markup languages; to force an API over two of 
> these languages would be a detriment to the system.
> 
> Note the use of the term "force".  I'm not suggesting the API isn't 
> needed, just that we shouldn't force things to rely on it that we 
> release ourselves.
> 
> 2.  In the same vein, we shouldn't ignore the need for core files in the 
> same vein as style.js and html.js for the other markup languages.  
> svg.js is already a part of the library (even though a good portion of 
> the functionality in that file is stub functionality that isn't working 
> yet).  I might expect there to be a vml.js at some point as well; I will 
> certainly add it when I get a better sense of what kind of helper 
> functions are needed (particularly when we finally address DnD with 
> vector markup).
> ---
> All of which to say is that I'd like it to be clear that any vector API 
> could (and probably will) rely on some core files but it should not be 
> the other way around.
> 
> Now that that disclaimer is out of the way...what do we think we want 
> out of such an API?  What kind of opinions are there as to the shape of 
> such an API?  What does Dojo really need in these terms?
> 
> I'll put forth a suggestion, and then let's discuss:  I would say that 
> we would start with a script-based API, modeled on the Canvas API, upon 
> which we could write general purpose widgets to represent drawing 
> primitives (and eventually more complex ones).  Before anyone objects, 
> let me try to be clear:  when I say "modeled" on the Canvas API, I'm 
> referring to methods and the general results of those methods.  The 
> Canvas API is based on OpenGL, which makes sense from a scripting 
> standpoint; I don't see any reason why we shouldn't use the same approach.
> 
> I know one of the things Alex has asked me for is that any primitive 
> drawn be able to fire and respond to events.  Not sure how we'll pull 
> that off with actual Canvas drawings, but I'll cross that bridge when we 
> get to it (probably it will involve looking for a mouse click on the 
> canvas element itself--if that's supported--and then analyzing the 
> current rendering to see what the top most element is that one clicked 
> on, what fun).  What other things would you want out of such an API?
> 
> trt
> 
> (for the record, I will certainly be looking at the Google 
> ExplorerCanvas code but I'm pretty sure we won't incorporate it because 
> of licensing...unless, Alex, you want to take a close look at it and 
> see?  My inclination is to learn from it but not use it, just to be 
> safe, plus I'd hate for Dojo to be reliant on other projects--since it 
> seems like one of the goals here is to be the other way around).
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> dojo-contributors mailing list
> dojo-contributors at dojotoolkit.org
> http://dojotoolkit.org/mailman/listinfo/dojo-contributors



More information about the dojo-contributors mailing list