Daniel Ritchie

Here's a thought, to keep the conversation going: wrap all deterministic JS functions in a webppl shim that handles continuations/stores/addresses. It's easy to do this for all deterministic JS provided...

This is surprising to me. I was under the impression that V8 uses string interning (i.e. hash consing), so that every time it goes to concat two strings it's already...

Here's some relevant info: http://stackoverflow.com/questions/5276915/do-common-javascript-implementations-use-string-interning It looks like while V8 does intern string literals, it does not intern the results of string operations (e.g. concat). Under these assumptions, the n^2...

I'm a fan of this idea. It's really simple, and it addresses the biggest of my concerns.

Also: decorated/wrapped functions will break the compiler if they are recursive, since the name of the function at its declaration gets mangled but its name at any recursive callsites stays...

@stuhlmueller pointed out that we might able to extend this to models with discrete choices by alternating optimizing the continuous choices via gradient descent and optimizing the discrete ones via...

Did anything ever happen with this? I'm asking because I'm running into situations with one of my current projects where stack traces would be really useful. Here's an example error:...

I think I was a bit unclear in my request--I'd be interested in the part of the stack that corresponds to JS code (i.e. not webppl code) that occurs between...

Ah, yep, that looks like it'd do the trick (plus some extra cruft).

> couldn't we overload \* to be matrix multiply (aka dot)? In most linear algebra libraries I've seen, the infix operators do element-wise operations, and things like dot products and...