Eric Clack

I think exploring the pruning might be beneficial. Here's an example, which renders some circles - about 25 are visible, but 5001 get rendered: ``` #lang s-exp stamps/lang (define-shape row...

Hi Rodrigo, My first strategy is to understand fully the recording paths code (especially recursive ones), so I'm writing some test code in a branch here: https://github.com/ericclack/racket-stamps/blob/performance/private/tests/renderer-recursive-tests.rkt Once I've done...

Ah I see how it works now, with the let-loop. I have a theory that at some point we can discard the functions returned by `(renderer pr)` in function `record-paths`...

More progress here: https://github.com/ericclack/racket-stamps/tree/performance More to do...

Testing seems to show no performance improvement! However it's been interesting for me to see how the code works... we'll see if further work leads anywhere...

Looking further I can see that the pruning is not having the desired effect -- certainly we are drawing less, but the work to record the paths seems to still...

See option #18

OK, some progress here, but no performance improvement that I can detect yet: https://github.com/ericclack/racket-stamps/tree/flonum

Thanks for your feedback. Here's one of the use-cases for better-than-rect collision detection: a maze game where the coder draws the walls and the whole maze is rendered as a...

Well I'd be happy to explore too. If you want to give me some pointers, things you'd explore first, I'll have a go.