Much has been written and said about the differences of HTML5 canvas and SVG, so I will not repeat all the categorization, but after working with these two new APIs I would like to offer some of my early findings. These findings are in the context to find a good HTML5 charting framework/solution, for one of my new financial websites.
SVG just seem more out of control, than canvas. I left like I was getting more consistency with canvas.
With SVG doing onclicks seemed much easier, where with canvas, x/y tracking was needed to determine where the user was clicking.
To my surprise, canvas did not deliver the same rendering across browsers when using fonts, so I was getting some inconsistent overlapping... oh I just had a glimpse of faith for a while that cross-browser tweaks would be gone in the HTML5 decade, but looks like UI developers will still have a job.
Scripting canvas is still very raw experience. There is no "object-like" approach you find in flash, which makes doing anything just a lot of work. Since SVG uses tags, I get a feeling that the "object oriented" nature might be there, however I have not seen anything comprehensive yet.