Ian Calvert

Duckdb builds without other issues, although I've not tried building the extensions. https://github.com/iancal/pyodide Pyarrow would solve a lot of other issues though, including loading remote parquet files, and play nicely...

A possible option here is to allow people to explicitly define which headers they want to pass through on the CLI. That way it's flexible for anyone's setup, but makes...

@devenjarvis one setup would be to use an authentication layer in front - such as https://github.com/cloudflare/nginx-google-oauth (there is a docker image pre-built), then the streamlit app or anything behind it...

@ryanking13 ah I see the right bit now, and hadn't realised how to do the tests, thanks for the heads up :) will sort that and update thanks!

Oh - if the tests are passing but it's failing on a "no-numpy-dependents" is there an extra flag I should add to say it is dependent on that? At the...

You may choose some default approach (perhaps the lru_cache with a copy), but it would be great if the caching function can be passed in. That allows things like TTLs,...

Interesting - it's not an issue I've really hit since generally I'm trying to get a lot of output (more stuff per prompt = more bang for your buck). It...

[edit - this is all with regards to the building, distributing and running part] Docker would be an obvious candidate for this kind of thing: https://www.docker.io/ A public repository with...

@Carreau There's nothing stopping you from reusing the code within it. You'd publish your code, a script for building it, and a fully contained runnable system. This means you know...

> The contents of the descriptor files would need to be agreed by community consensus, could this be a viable goal for this group? I'd advocate "ask for forgiveness, not...