fix: add limit clause to tpc-h queries (#3437)

Fixes: #3433

Note that the row count for sf=10 query 13 was corrected to 45 rows,
unrelated to introducing the `limit` clause.

---------

Signed-off-by: Alexander Droste <[email protected]>

Author: 0ax1

.github/workflows/labels.yml

@@ -2,27 +2,29 @@ name: PR Labels
 
 on:
   pull_request:
-    types: [ opened, reopened, synchronize, labeled, unlabeled ] # Trigger on these PR activities
+    types: [ opened, reopened, synchronize, labeled, unlabeled ]  # Trigger on these PR activities
 
 jobs:
   check_changelog_label:
     name: Validate Changelog Label
     runs-on: ubuntu-latest
     permissions:
-      pull-requests: read # Grant permission to read PR information
+      pull-requests: read  # Grant permission to read PR information
     steps:
       - name: Get PR Labels from API
         id: get_labels_api
-        uses: octokit/[email protected] # Use an action to make API requests
+        uses: octokit/[email protected]  # Use an action to make API requests
         with:
           route: GET /repos/{owner}/{repo}/issues/{pull_number}/labels
           owner: ${{ github.repository_owner }}
           repo: ${{ github.event.repository.name }}
           pull_number: ${{ github.event.pull_request.number }}
         env:
-          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} # Automatically provided token
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}  # Automatically provided token
 
       - name: Extract and Check Labels
+        env:
+          API_RESPONSE: ${{ steps.get_labels_api.outputs.data }}
         run: |
           REQUIRED_LABELS=(
             "chore"
@@ -38,7 +40,6 @@ jobs:
 
           # Parse the response from the API call
           # The API returns an array of label objects, we need just the 'name' property
-          API_RESPONSE='${{ steps.get_labels_api.outputs.data }}'
           echo "API Response: $API_RESPONSE"
 
           # Extract only the label names into a JSON array

bench-vortex/src/bin/tpch.rs

@@ -142,6 +142,7 @@ fn main() -> anyhow::Result<()> {
                     format
                 };
                 let opts = DuckdbTpcOptions::new("tpch".to_data_path(), TpcDataset::TpcH, format)
+                    .with_scale_factor(args.scale_factor)
                     .with_duckdb_path(duckdb_resolved_path.clone());
                 generate_tpc(opts)?;
             }

bench-vortex/src/tpch/mod.rs

@@ -28,14 +28,14 @@ pub use execute::*;
 pub const TPC_H_ROW_COUNT_ARRAY_LENGTH: usize = 23;
 pub const EXPECTED_ROW_COUNTS_SF1: [usize; TPC_H_ROW_COUNT_ARRAY_LENGTH] = [
     // The 0th entry is a dummy so that Query 1's row count is at index 1.
-    0, 4, 460, 11620, 5, 5, 1, 4, 2, 175, 37967, 1048, 2, 42, 1, 1, 18314, 1, 57, 1, 186, 411, 7,
+    0, 4, 100, 10, 5, 5, 1, 4, 2, 175, 20, 1048, 2, 42, 1, 1, 18314, 1, 57, 1, 186, 100, 7,
 ];
 pub const EXPECTED_ROW_COUNTS_SF10: [usize; TPC_H_ROW_COUNT_ARRAY_LENGTH] = [
     // The 0th entry is a dummy so that Query 1's row count is at index 1.
     //
     // Generated by executing the SQL in each query file using duckdb with the table names replaced
     // by "$NAME.parquet".
-    0, 4, 4667, 114003, 5, 5, 1, 4, 2, 175, 381105, 0, 2, 46, 1, 1, 27840, 1, 624, 1, 1804, 4009, 7,
+    0, 4, 100, 10, 5, 5, 1, 4, 2, 175, 20, 0, 2, 45, 1, 1, 27840, 1, 100, 1, 1804, 100, 7,
 ];
 
 // Generate table dataset.

bench-vortex/tpch/q10.sql

@@ -28,4 +28,5 @@ group by
     c_address,
     c_comment
 order by
-    revenue desc;
+    revenue desc
+limit 20;

bench-vortex/tpch/q18.sql

@@ -29,4 +29,5 @@ group by
     o_totalprice
 order by
     o_totalprice desc,
-    o_orderdate;
+    o_orderdate
+limit 100;

bench-vortex/tpch/q2.sql

@@ -40,4 +40,5 @@ order by
     s_acctbal desc,
     n_name,
     s_name,
-    p_partkey;
+    p_partkey
+limit 100;

bench-vortex/tpch/q21.sql

@@ -36,4 +36,5 @@ group by
     s_name
 order by
     numwait desc,
-    s_name;
+    s_name
+limit 100;

bench-vortex/tpch/q3.sql

@@ -19,4 +19,5 @@ group by
     o_shippriority
 order by
     revenue desc,
-    o_orderdate;
+    o_orderdate
+limit 10;