ci: speed up ci by introducing precompiled slt binary

crates/datasources/src/bson/builder.rs

@@ -82,7 +82,6 @@ impl RecordStructBuilder {
                         .field_index
                         .get(key)
                         .ok_or_else(|| BsonError::ColumnNotInInferredSchema(key.to_string()))?;
-                    println!("{}->{}", key, idx);
 
                     if *cols_set.get(idx).unwrap() {
                         continue;

crates/testing/Cargo.toml

@@ -35,3 +35,7 @@ rpcsrv = { path = "../rpcsrv" }
 harness = false
 name = "sqllogictests"
 path = "tests/sqllogictests/main.rs"
+
+[[bin]]
+name = "slt"
+path = "src/main.rs"

crates/testing/src/main.rs

@@ -0,0 +1,20 @@
+use std::sync::Arc;
+
+use testing::slt::{
+    hooks::{AllTestsHook, SshTunnelHook},
+    runner::SltRunner,
+    tests::{PgBinaryEncoding, SshKeysTest},
+};
+
+pub fn main() -> anyhow::Result<()> {
+    SltRunner::new()
+        .test_files_dir("testdata")?
+        // Rust tests
+        .test("sqllogictests/ssh_keys", Box::new(SshKeysTest))?
+        .test("pgproto/binary_encoding", Box::new(PgBinaryEncoding))?
+        // Add hooks
+        .hook("*", Arc::new(AllTestsHook))?
+        // SSH Tunnels hook
+        .hook("*/tunnels/ssh", Arc::new(SshTunnelHook))?
+        .run()
+}

crates/testing/src/slt.rs

@@ -1,5 +1,7 @@
 //! Utility to run SQL Logic Tests.
 
-mod cli;
+pub mod cli;
+pub mod hooks;
 pub mod runner;
 mod test;
+pub mod tests;

crates/testing/src/slt/cli.rs

@@ -145,7 +145,8 @@ impl Cli {
         let mut tests: Vec<_> = if let Some(patterns) = &self.tests_pattern {
             let patterns = patterns
                 .iter()
-                .map(|p| glob::Pattern::new(p))
+                .map(|p| p.trim_end_matches(".slt"))
+                .map(glob::Pattern::new)
                 .collect::<Result<Vec<_>, _>>()?;
 
             tests

crates/testing/tests/sqllogictests/hooks.rs → crates/testing/src/slt/hooks.rs

@@ -2,7 +2,6 @@ use std::{collections::HashMap, time::Duration};
 
 use anyhow::{anyhow, Result};
 use async_trait::async_trait;
-use testing::slt::runner::{Hook, TestClient};
 use tokio::{
     net::TcpListener,
     process::Command,
@@ -11,6 +10,8 @@ use tokio::{
 use tokio_postgres::{Client, Config};
 use tracing::warn;
 
+use super::test::{Hook, TestClient};
+
 /// This [`Hook`] is used to set some local variables that might change for
 /// each test.
 pub struct AllTestsHook;

crates/testing/tests/sqllogictests/tests.rs → crates/testing/src/slt/tests.rs

@@ -2,10 +2,11 @@ use std::collections::HashMap;
 
 use anyhow::{anyhow, Result};
 use async_trait::async_trait;
-use testing::slt::runner::{FnTest, TestClient};
 use tokio_postgres::Config;
 use tracing::warn;
 
+use super::test::{FnTest, TestClient};
+
 macro_rules! test_assert {
     ($e:expr, $err:expr) => {
         if !($e) {

crates/testing/tests/sqllogictests/main.rs

@@ -1,11 +1,10 @@
-mod hooks;
-mod tests;
-
 use anyhow::Result;
-use hooks::{AllTestsHook, SshTunnelHook};
 use std::sync::Arc;
-use testing::slt::runner::SltRunner;
-use tests::{PgBinaryEncoding, SshKeysTest};
+use testing::slt::{
+    hooks::{AllTestsHook, SshTunnelHook},
+    runner::SltRunner,
+    tests::{PgBinaryEncoding, SshKeysTest},
+};
 
 fn main() -> Result<()> {
     SltRunner::new()

justfile

@@ -67,8 +67,8 @@ sql-logic-tests *args: protoc
   just test --test sqllogictests -- {{args}}
 
 # Run SQL Logic Tests over RPC
-rpc-tests: protoc
-  just sql-logic-tests --protocol=rpc \
+rpc-tests:
+  just slt-bin --protocol=rpc \
     'sqllogictests/cast/*' \
     'sqllogictests/cte/*' \
     'sqllogictests/functions/delta_scan' \
@@ -113,6 +113,14 @@ rpc-tests: protoc
     'sqllogictests/describe_rpc' \
     'sqllogictests/allowed_operations'
 
+# Build a pre-compiled slt runner
+build-slt *args:
+  cargo build --bin slt -- {{args}}
+
+# Run SQL Logic Tests with a pre-compiled slt runner
+slt-bin *args:
+  ./target/debug/slt {{args}}
+
 #  Check formatting.
 fmt-check: protoc
   cargo fmt --check

testdata/sqllogictests/catalog/functions.slt

@@ -54,5 +54,5 @@ select
 from glare_catalog.functions
 where function_name = 'read_parquet';
 ----
-read_parquet   table   Utf8/List<Utf8>   t   SELECT * FROM read_parquet('./my_data.parquet')   Returns a table by scanning the given Parquet file(s).
-read_parquet   table   Utf8/List<Utf8>   t   SELECT * FROM read_parquet('./my_data.parquet')   Returns a table by scanning the given Parquet file(s).
+read_parquet table Utf8/List<Utf8> t SELECT * FROM read_parquet('./my_data.parquet') Returns a table by scanning the given Parquet file(s).
+read_parquet table Utf8/List<Utf8> t SELECT * FROM read_parquet('./my_data.parquet') Returns a table by scanning the given Parquet file(s).

testdata/sqllogictests/create_table.slt

@@ -49,7 +49,7 @@ select * from ctas1;
 1
 
 statement ok
-create table ctas2 as select * from '../../testdata/parquet/userdata1.parquet';
+create table ctas2 as select * from '${PWD}/testdata/parquet/userdata1.parquet';
 
 query I
 select id from ctas2 order by id limit 1;
@@ -82,16 +82,16 @@ select count(*) from glare_catalog.tables where builtin = false and table_name =
 
 #2034 case sensitive table names
 statement ok 
-create table case_sensitive as select * from '../../testdata/csv/case_sensitive_columns.csv';
+create table case_sensitive as select * from '${PWD}/testdata/csv/case_sensitive_columns.csv';
 
 statement error Duplicate name: case_sensitive
-create table "case_sensitive" as select * from '../../testdata/csv/case_sensitive_columns.csv';
+create table "case_sensitive" as select * from '${PWD}/testdata/csv/case_sensitive_columns.csv';
 
 statement ok
-create table "Case_Sensitive" as select * from '../../testdata/csv/case_sensitive_columns.csv';
+create table "Case_Sensitive" as select * from '${PWD}/testdata/csv/case_sensitive_columns.csv';
 
 statement ok
-create table "Case Sensitive" as select * from '../../testdata/csv/case_sensitive_columns.csv';
+create table "Case Sensitive" as select * from '${PWD}/testdata/csv/case_sensitive_columns.csv';
 
 query I rowsort
 select name from case_sensitive

testdata/sqllogictests/csv.slt

@@ -23,20 +23,20 @@ select count(*), status from bikeshare_stations group by status order by status;
 # Empty column name (#1750)
 
 query ITT rowsort
-select * from '../../testdata/csv/empty_col.csv'
+select * from '${PWD}/testdata/csv/empty_col.csv'
 ----
 0 a hello
 1 b world
 
 query T rowsort
-select col1 from '../../testdata/csv/empty_col.csv'
+select col1 from '${PWD}/testdata/csv/empty_col.csv'
 ----
 a
 b
 
 # Weird, but it works
 query T rowsort
-select "" from '../../testdata/csv/empty_col.csv'
+select "" from '${PWD}/testdata/csv/empty_col.csv'
 ----
 0
 1

testdata/sqllogictests/external_table.slt

@@ -14,7 +14,7 @@ statement ok
 CREATE OR REPLACE EXTERNAL TABLE T1 FROM DEBUG OPTIONS (TABLE_TYPE = 'never_ending');
 
 statement ok
-CrEaTe ExTeRnAl TaBlE if not exists SuppLIER fRoM lOcAl (LoCaTiOn '../../testdata/parquet/userdata1.parquet');
+CrEaTe ExTeRnAl TaBlE if not exists SuppLIER fRoM lOcAl (LoCaTiOn '${PWD}/testdata/parquet/userdata1.parquet');
 
 statement ok
 drop table supplier;

testdata/sqllogictests/functions/csv_scan.slt

@@ -33,7 +33,7 @@ select count(*) from csv_scan(
 
 # Relative path
 query I
-select count(*) from csv_scan('../../testdata/sqllogictests_datasources_common/data/bikeshare_stations.csv')
+select count(*) from csv_scan('./testdata/sqllogictests_datasources_common/data/bikeshare_stations.csv')
 ----
 102
 

testdata/sqllogictests/functions/delta_scan.slt

@@ -16,7 +16,7 @@ select * from delta_scan('file://${PWD}/testdata/delta/table1') order by a;
 
 # Relative path
 query IT
-select * from delta_scan('../../testdata/delta/table1/') order by a;
+select * from delta_scan('./testdata/delta/table1/') order by a;
 ----
 1   hello
 2   world

testdata/sqllogictests/functions/json_scan.slt

@@ -14,7 +14,7 @@ select count(*) from ndjson_scan('file://${PWD}/testdata/sqllogictests_datasourc
 
 # # Relative path
 query I
-select count(*) from ndjson_scan('../../testdata/sqllogictests_datasources_common/data/bikeshare_stations.ndjson')
+select count(*) from ndjson_scan('./testdata/sqllogictests_datasources_common/data/bikeshare_stations.ndjson')
 ----
 102
 

testdata/sqllogictests/functions/lance_scan.slt

@@ -16,7 +16,7 @@ select * from lance_scan('file://${PWD}/testdata/lance/table1') order by point.l
 
 # Relative path
 query IT
-select * from lance_scan('../../testdata/lance/table1/') order by point.lat;
+select * from lance_scan('.//testdata/lance/table1/') order by point.lat;
 ----
 0.2,1.8 {lat:42.1,long:-74.1}
 1.1,1.2 {lat:45.5,long:-122.7}

testdata/sqllogictests/functions/parquet_scan.slt

@@ -8,7 +8,7 @@ select count(*) from parquet_scan('file://${PWD}/testdata/parquet/userdata1.parq
 
 # Relative path
 query I
-select count(*) from parquet_scan('../../testdata/parquet/userdata1.parquet')
+select count(*) from parquet_scan('./testdata/parquet/userdata1.parquet')
 ----
 1000
 
@@ -45,14 +45,14 @@ select * from parquet_scan('./testdata/parquet/userdata1.paruqet');
 # Ambiguous name.
 # query I
 # select count(*)
-#   from parquet_scan('../../testdata/parquet/userdata1.parquet') p
+#   from parquet_scan('${PWD}/testdata/parquet/userdata1.parquet') p
 #   inner join (values ('Sweden')) as c(country) on p.country = c.country
 # ----
 # 1000
 
 # query I
 # select count(*)
-#   from parquet_scan('../../testdata/parquet/userdata1.parquet') p
+#   from parquet_scan('${PWD}/testdata/parquet/userdata1.parquet') p
 #   inner join (select 'Sweden') as c(country) on p.country = c.country
 # ----
 # 1000

testdata/sqllogictests/functions/read_csv.slt

@@ -33,7 +33,7 @@ select count(*) from read_csv(
 
 # Relative path
 query I
-select count(*) from read_csv('../../testdata/sqllogictests_datasources_common/data/bikeshare_stations.csv')
+select count(*) from read_csv('./testdata/sqllogictests_datasources_common/data/bikeshare_stations.csv')
 ----
 102
 
@@ -66,11 +66,11 @@ select * from read_csv(
 # Alternative delimiters
 
 query ITR
-select * from read_csv('../../testdata/csv/delimiter.csv', delimiter => ';');
+select * from read_csv('./testdata/csv/delimiter.csv', delimiter => ';');
 ----
 1   hello, world   3.9
 2   HELLO, WORLD   4.9
 
 # Invalid delimiter (longer than one byte)
 statement error delimiters for CSV must fit in one byte \(e.g. ','\)
-select * from read_csv('../../testdata/csv/delimiter.csv', delimiter => ';;');
+select * from read_csv('./testdata/csv/delimiter.csv', delimiter => ';;');

testdata/sqllogictests/functions/read_json.slt

@@ -14,7 +14,7 @@ select count(*) from read_ndjson('file://${PWD}/testdata/sqllogictests_datasourc
 
 # # Relative path
 query I
-select count(*) from read_ndjson('../../testdata/sqllogictests_datasources_common/data/bikeshare_stations.ndjson')
+select count(*) from read_ndjson('./testdata/sqllogictests_datasources_common/data/bikeshare_stations.ndjson')
 ----
 102
 

testdata/sqllogictests/functions/read_parquet.slt

@@ -8,7 +8,7 @@ select count(*) from read_parquet('file://${PWD}/testdata/parquet/userdata1.parq
 
 # Relative path
 query I
-select count(*) from read_parquet('../../testdata/parquet/userdata1.parquet')
+select count(*) from read_parquet('./testdata/parquet/userdata1.parquet')
 ----
 1000
 
@@ -45,14 +45,14 @@ select * from read_parquet('./testdata/parquet/userdata1.paruqet');
 # Ambiguous name.
 # query I
 # select count(*)
-#   from read_parquet('../../testdata/parquet/userdata1.parquet') p
+#   from read_parquet('${PWD}/testdata/parquet/userdata1.parquet') p
 #   inner join (values ('Sweden')) as c(country) on p.country = c.country
 # ----
 # 1000
 
 # query I
 # select count(*)
-#   from read_parquet('../../testdata/parquet/userdata1.parquet') p
+#   from read_parquet('${PWD}/testdata/parquet/userdata1.parquet') p
 #   inner join (select 'Sweden') as c(country) on p.country = c.country
 # ----
 # 1000

testdata/sqllogictests/glob.slt

@@ -1,14 +1,14 @@
 query I
-select count(*) from '../../testdata/parquet/userdata1.parquet';
+select count(*) from '${PWD}/testdata/parquet/userdata1.parquet';
 ----
 1000
 
 query I
-select count(*) from '../../testdata/parquet/*.parquet' as pd where pd.id != '';
+select count(*) from '${PWD}/testdata/parquet/*.parquet' as pd where pd.id != '';
 ----
 2000
 
 query I
-select count(*) from '../../testdata/parquet/*.parquet';
+select count(*) from '${PWD}/testdata/parquet/*.parquet';
 ----
 2000