Xuanda Yang

Thanks for the elaboration, this is definitely gold! > The way I see it, we want to be able to describe a primitive op as "call C function X with...

> we currently have operations that will be inlined, and we can't rely on inline C functions or macros in a non-C back end. I try to name some examples...

Also, I start drafting the proposal, it will be updated continuously(so far it contains some template content), check it at https://docs.google.com/document/d/1mDh_Dm7YKoQlBDkG7_W9dO0ZwuS02pu94f1cZHONapI/edit?usp=sharing Please take a look and feel free to comment...

Hi @JukkaL and @msullivan, I've submitted python/mypy#8570 to demonstrate some of the ideas we've discussed here, feel free to take a look. Thanks! And I've finished my first draft of...

Some updates: - I've posted my newest understanding of the design in the PR, please be sure to take a look so we can push this forward. - Since GSOC's...

I believe for OS it should cover Windows/MacOS/Linux with ease, ~~architecture wise it should be targeting x86_64 for the most part at this moment~~ see better option below

I believe so, I recently did something similar to the issue's description and triggered the proper check and error messages

Not sure if I fully understand your question here, but as for my repros, the quick is to put them as a new file in `test/correctness` and you should be...

Because Taichi is a JIT compiler, which means it will compile the Taichi kernels during running the code. In your loop, kernel `assemble_k` is compiled during the first iteration, all...