A little program I wrote because I'm tired of writing the same Django admin code, maintaining complex SQL queries and JavaScript transformers in Retool, and working around the lack of writability in BI tools. It's an early alpha, so I wouldn't use it in production, but I would appreciate any feedback on it and whether you find it useful. |
My main takeaways have been:
- It mostly just works and simplifies things, but:
- Working with page state (in my case, which sheet is loaded) in individual requests (like updating a cell) is a bit awkward -- you can either supply it via `hx-vals` or via setting a global hidden input which is specified via `hx-include`. I ended up going with the second approach due to https://github.com/bigskysoftware/htmx/issues/1119 but it's not perfect -- in particular you need to duplicate specifying the input if you want to include additional inputs for some node.
- Error handling is also a bit awkward, but it's not too bad once you get used to the idea that you should return HTML elements in 400s. I found https://htmx.org/extensions/response-targets/ helpful as well. - Often HTML structure doesn't allow you to limit your request scope as much as you'd like. For me the big thing is I can't just replace one column since tables are row-major -- I have to replace the whole table if a column is modified.
Plato is a visual database, similar to Airtable, with read-write (configurable) access to your production database. It's used for operations.