Been incrementally solving the problem of creating markdown documentation/resources for dataframe. Tweaked the code which this version of hscript was based on. It’s still pretty fragile (well - more fragile than having this be a first class cabal/GHC capability) but it’s been helpful for me.
Dealing with template Haskell made it harder to make this into a preprocessor as @blamario suggested in the thread above - if anyone has ideas please share them.
This reminds me of Chris Done’s hell language. It has excellent ergonomics (the runner is a standalone binary). It’s explicitly designed as a limited Haskell-reminiscent scripting language, rather than running a GHCi script, so doesn’t support importing extra code that it wasn’t built with. But it might be relevant. (And there’s nothing stopping one from forking to add extra modules & instances.)
Thanks! hell is pretty great! I’ve been using it for some years now. This solves a slightly different problem - it’s supposed to run arbitrary Haskell packages and also manage the dependencies. See the example analysis.md when in the server spun up by cabal run sabela-server. So it’s more similar to a pre-processing step for Jupyter-like notebooks.