Upgrade to DataFusion 17, fix a couple of bugs, add some tests (#18)

Cargo.toml

@@ -12,9 +12,9 @@ rust-version = "1.62"
 build = "build.rs"
 
 [dependencies]
-datafusion = { git = "https://github.com/apache/arrow-datafusion", rev = "5238e8c97f998b4d2cb9fab85fb182f325a1a7fb", features = ["pyarrow", "avro"] }
-datafusion-proto = { git = "https://github.com/apache/arrow-datafusion", rev = "5238e8c97f998b4d2cb9fab85fb182f325a1a7fb" }
-datafusion-python = { git = "https://github.com/apache/arrow-datafusion-python", rev = "0a6743c2a7f27015890c5546efda2d4e85d45a97" }
+datafusion = { version = "17.0.0", features = ["pyarrow", "avro"] }
+datafusion-proto = "17.0.0"
+datafusion-python = { git = "https://github.com/andygrove/arrow-datafusion-python", rev = "7164900a32b25f0644a3294389f02e28e94da8a3" }
 futures = "0.3"
 glob = "0.3"
 log = "0.4"

README.md

@@ -26,7 +26,7 @@ ctx.sql('select sex, smoker, avg(tip/total_bill) as tip_pct from tips group by s
 
 ## Status
 
-- Partially working. Can run about half of the TPC-H benchmark queries.
+- RaySQL can run 21 of the 22 TPC-H benchmark queries (query 15 needs DDL and that is not yet supported).
 
 ## Features
 
@@ -41,8 +41,7 @@ ctx.sql('select sex, smoker, avg(tip/total_bill) as tip_pct from tips group by s
 
 This chart shows the relative performance of RaySQL compared to other open-source distributed SQL frameworks.
 
-Performance does not look very promising so far, but this project was created in one weekend, so this is probably not 
-too bad a start.
+Performance is looking pretty respectable!
 
 ![SQLBench-H Performance Chart](./docs/sqlbench-h-workstation-10-distributed-perquery.png)
 
@@ -72,10 +71,7 @@ python -m pytest
 Create a release build when running benchmarks, then use pip to install the wheel.
 
 ```bash
-cargo clean
-maturin develop
-maturin build --release
-pip install ./target/wheels/raysql-0.1.0-cp37-abi3-manylinux_2_31_x86_64.whl --force-reinstall
+maturin develop --release
 ```
 
 ## How to update dependencies