Clutter Canvas and memory limits

[Donn <donn ingle gmail com>]

Hi,
I'm in a planning stage for an wee project - I'd like to know general 
info about clutter's canvas and how it uses memory (video and normal).

In a nutshell:
A drawing app. Imagine circle, path, image, etc. Like Inskcape.

A)
 I could have each object/element drawn as a Canvas-Actor (via Cairo) — 
Imagine this goes into the hundreds (and more..) of such objects.
(I get the benefits of Clutter's stack and events, etc. Bonus.)

vs.

B)
 I could draw all of the shapes (via Cairo) onto one Canvas-Actor. I'd 
have to implement a picking system and a custom retained-mode thing to 
manipulate the "objects" within that Canvas. (PITA).


Is option A a good idea in Clutter? How would I code (Python/C/Vala) in 
order to catch conditions where I am making too many surfaces or 
exceeding the V/RAM?

Are there any web resources I could visit that would help?

Thanks,
\d