Dave Hirschfeld

Yeah, I figured that wouldn't work 🤦‍♂️ Will have to think how I can yield as results arrive 🤔

> *I guess the function results feed into a trio memory channel, and the main loop just yields incoming items?* That sounds like it could work - will give it...

> I agree that a general abstract expression system for arrays would be valuable. That sounds a bit like what [`metadsl`](https://github.com/Quansight-Labs/metadsl) was intended to support? Unfortunately development has stalled there...

There might possibly be some useful typing examples in the `array-api` project: * https://github.com/data-apis/array-api/tree/main/spec/API_specification/signatures

There is also the DataFrame API which would seem a better fit for `polars`: * https://github.com/data-apis/dataframe-api > *It would be pretty neat if consumers like scikit-learn could work with the...

> *I'd think we'd want to conform with both, no?* I think they're two separate things. You're either trying to provide a 2D DataFrame api or an nD Tensor api....

Is @yuvipanda's [`simplest-notebook` ](https://github.com/yuvipanda/simplest-notebook) an example or is that not `jupyterlab` based?

> *You can test Katex or MathJax 3 by installing an extension* I'm very interested if the extensions improve the situation at all! 👀

Also requested in https://github.com/jupyterlab/jupyterlab/issues/3011#issuecomment-360021201 ...but it's good for it to have its own issue.

It would be great if this could work on Windows too! Possibly via a configurable.