Michael Kenzel

Well, the intended behavior would be that `pe_assert(expression, "message")` would be ignored if `expression` evaluates to true and, if `expression` evaluates to false, make compilation fail, producing a diagnostic at...

@madmann91 Good question…I guess that should just be a hard error as asserting that a condition be true during PE would obviously require that the condition is actually one that...

I'm afraid the problem does not appear to be resolved completely as I've run into it again. The following example ```rust fn ohnoes(throw: fn(&[u8]) -> ()) { throw(""); } #[export]...

I'm afraid this now seems to break the following code that's part of the runtime (reduced sample): ```rust #[import(cc = "thorin")] fn opencl(_dev: i32, _grid: (i32, i32, i32), _block: (i32,...

> For thread IDs in particular, I think it's good that they are functions, because at the IR level that's what they have to be, you might need to generate...

I revisited these changes. It's now only cleaning up inconsistent uses of the `world()` vs `world_` members.

Sorry, I should have mentioned that up there, this is independent of the plugin and programming model stuff. It just adds a parameter to allocate a given amount of dynamic...

> Do we document the syntax of this thing somewhere (guessing this is based on GCC's but I'm not familiar) ? Well, it's in the example [here](https://anydsl.github.io/Assembly-Statement.html), but I guess...

thorin: fb0f72189c16826f69ad947fbf197ebac3cc520c artic: 0f50091011f2bf82efd460264f2f40781a85f72b runtime: 5c1a78c07cbeaea16fd61c73e92fd7112dd62ab7

Awesome, thanks! It's still about 5× slower (25 s vs 5 s) on my minimal test build here than pre-merge, but that's already *a lot* better than 27× slower 😄👍