Daniël Heres
Daniël Heres
One small other optimization is s `Box` for `Vec` and `Box` instead of `String`. `Arc` should be fine too. Idealy they would only be there for the expensive to clone...
Haha cool!
I am wondering if the current `2` should just be based on a configation setting for the default number of partitions (just like Spark uses 200 partitions as a default)....
Sounds great! The first indeed is the more low hanging fruit I meant with this story (`TokenizerError`). The second one is really valuable but probably requires much more changes.
@nickolay I saw that in the fork of Materialize column / line indexes are supported. For that, the following is done: * add range to result of tokenizer https://github.com/MaterializeInc/materialize/blob/main/src/sql-parser/src/tokenizer.rs#L328 *...
https://github.com/ballista-compute/sqlparser-rs/pull/217 contains a fix
I would like to work on this, as I am working on a tool that needs to export queries to different. It would be an awesome feature actually to write...
I am willing to take this up. But I think I need access to the repo to add a secret. If this sounds like a good idea, can anybody @andygrove...
1. Yes, we could also use the displayed output instead of serde, that would be much better (and cover that displaying functionality / regressions a bit more as well). 2....
Thanks @Jasperav . In think the `?` is not something that's supported yet.