prql-query
prql-query copied to clipboard
Create snapcraft.yaml
@snth or @max-sixty Can we snap this?
Sorry, I missed this.
@snth — do we want to do this? I'm a big fan of prql-query, but we should ensure you're advocating for more folks to use it if we release more broadly. (Doesn't have to be finished — but hopefully it's working for some subset of use-cases, however small, and the doc are explicit about the breadth of use-cases...)
As I wrote here https://github.com/PRQL/prql-query/issues/22#issuecomment-1468377907, I vote to archive this repository in favor of combinations of prqlc + duckdb cli, datafusion cli, polars cli, etc. While prql-query is a very interesting example of combining PRQL with DataFusion, etc., in practicality it is far more flexible to combine prqlc with other CLI tools and not worry about stale updates...
Thank you for doing this @vanillajonathan .
I have some thoughts which will take longer to write up but are broadly along the lines of @eitsupi . If you can wait a day or two I'll come back to add them.
Alright! I look forward to hearing your thoughts!
@snth Any update?
Hi,
Apologies, I owe an update on this for a long time.
Looking back, what I think killed my momentum on this were the long compile times for DuckDB as well as the version incompatibilities between the different Arrow crates which means that it becomes very difficult to incorporate different projects (like DataFusion, DuckDB, Polars, ...) into one binary while this works quite simply in Python. So while having a small compiled Rust binary is nice, I don't think it's worth the effort and I would rather redo this in Python.
I think we can keep what's there, but I would suggest:
- Remove the DuckDB dependency.
- Just use DataFusion
in order to make it easy to bump the versions to keep up with the
prql-compilerreleases.
I still feel there is a place for a tool like this and when I get some time I want to rewrite this in Python. I also have a similar project as a bash script which I can release once I've fixed a particular bug.
If anyone wants to take over this project and run with it, you're very welcome and I will try to assist where I can.
As a matter of fact, I think that what was set out here has almost already been accomplished in GlareDB CLI.