alamb commented on code in PR #9546:
URL: https://github.com/apache/arrow-rs/pull/9546#discussion_r2930170348
##########
arrow-json/benches/json_reader.rs:
##########
@@ -323,13 +325,83 @@ fn bench_serialize_list(c: &mut Criterion) {
});
}
+fn bench_schema_inference(c: &mut Criterion) {
+ const ROWS: usize = 1000;
+
+ #[derive(Serialize, Arbitrary, Debug)]
+ struct Row {
+ a: Option<i16>,
+ b: Option<String>,
+ c: Option<[i16; 8]>,
+ d: Option<[bool; 8]>,
+ e: Option<Inner>,
+ f: f64,
+ }
+
+ #[derive(Serialize, Arbitrary, Debug)]
+ struct Inner {
+ a: Option<i16>,
+ b: Option<String>,
+ c: Option<bool>,
+ }
+
+ let mut data = vec![];
+ for row in pseudorandom_sequence::<Row>(ROWS) {
Review Comment:
I think other benchmarks we have use seedable_rng to get repeatable psuedo
random numbers. Is there a reason we shouldn't follow the same pattern here?
##########
arrow-json/Cargo.toml:
##########
@@ -61,6 +61,7 @@ tokio = { version = "1.27", default-features = false,
features = ["io-util"] }
bytes = "1.4"
criterion = { workspace = true, default-features = false }
rand = { version = "0.9", default-features = false, features = ["std",
"std_rng", "thread_rng"] }
+arbitrary = { version = "1.4.2", features = ["derive"] }
Review Comment:
I would prefer not to add a new dependency (even a dev one) unless really
necessary as that is then one more thing to chase down / maintain. I think you
could get the same effect using a random number generator directly
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
