Our great sponsors
-
prysm
physical optics: integrated modeling, phase retrieval, segmented systems, polynomials and fitting, sequential raytracing... (by brandondube)
-
InfluxDB
Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.
-
WorkOS
The modern identity platform for B2B SaaS. The APIs are flexible and easy-to-use, supporting authentication, user identity, and complex enterprise features like SSO and SCIM provisioning.
-
praise
reference implementation of contemporary "forward-reverse" or "iterative transform" phase retrieval algorithms
-
cinder
Cinder is Meta's internal performance-oriented production version of CPython. (by facebookincubator)
There's at least one Lisp implementation in Python called Hy
Just in the last week or two, I wrote an algorithm and then made it vectorized over a couple of commits. This process really is not difficult with the tools we have today and zero type annotations.
In my view, we're far more likely see this need met in the tools ecosystem for CPython. For example, the ability to compile Python to C using the current gradual typing system, using tools like mypyc (which is already used by mypy itself for massive performance gains).
I have a feeling the pytorch team at facebook will drop a new compiled or jitted version of a subset of python in the next year or two. They have a really strong compiler/language team that already hacked up large portions of python, including the GIL free interpreter (https://github.com/colesbury/nogil/), using multiple interpreters for inference in python (https://arxiv.org/abs/2104.00254), and a bunch of other efforts (https://dev-discuss.pytorch.org/c/compiler/5)
Already exists: https://github.com/keith-packard/snek
I think for legibility, I expect ides will eventually support toggling type hints on/off. https://github.com/microsoft/pylance-release/issues/2177 is the relevant vscode issue.
I write a lot of library code and don't get tickets about type problems.
I write a lot of library code and don't get tickets about type problems.
I write a lot of library code and don't get tickets about type problems.
I write a lot of library code and don't get tickets about type problems.
Not the PyTorch team, but here's both JIT support (of basically everything) and leveraging type annotations for better performance ("static Python") in Python 3.8: https://github.com/facebookincubator/cinder