From c3ed59a683bdc3441d2d68081ed28a621c11b98c Mon Sep 17 00:00:00 2001
From: Claude <noreply@anthropic.com>
Date: Mon, 31 Aug 2026 14:25:13 +0000
Subject: [PATCH] build(workflows): stream PR list when checking for duplicate
 PRs

The `check_duplicate_prs` workflow has failed on four of the last five
nightly runs against `develop`. The "Check for duplicate PRs" step
aborts with `The action 'Check for duplicate PRs' has timed out after
15 minutes.` shortly after reporting `Processed 900 PRs...`.

The script accumulates every open pull request into a single JSON
array and then indexes into that array once per pull request via
`jq -c ".[$i]"`. Each of those invocations re-parses the entire array,
which is roughly 4 MB at the current ~950 open pull requests, making
the loop quadratic. Job logs show a steady 47 seconds per 50 pull
requests, which puts a full pass just past the step's 15 minute
budget.

Stream the array through a single `jq -c '.[]'` invocation and drive
the loop with `read` instead, so that the list is parsed once rather
than once per pull request. Process substitution keeps the loop body
in the current shell, so the accumulated issue-to-PR arrays are
unaffected. Against a mocked API and a 122 pull request fixture
covering null bodies, empty bodies, a null array element, and a null
pull request number, the script's stdout and stderr are byte-identical
to before the change.

Ref: https://github.com/stdlib-js/stdlib/actions/runs/33352880096
---
 .github/workflows/scripts/check_duplicate_prs/run | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/.github/workflows/scripts/check_duplicate_prs/run b/.github/workflows/scripts/check_duplicate_prs/run
index 07c45303a604..ddcd796cdf60 100755
--- a/.github/workflows/scripts/check_duplicate_prs/run
+++ b/.github/workflows/scripts/check_duplicate_prs/run
@@ -255,14 +255,12 @@ main() {
 	echo "Processing PRs for issue references..."
 	debug_log "Starting to process $total_prs PRs for issue references"
 
-	# Process PRs one by one...
+	# Process PRs one by one, streaming the list through a single `jq` invocation in order to avoid re-parsing the entire list once per PR...
 	processed_count=0
 
-	for ((i=0; i<total_prs; i++)); do
-		pr=$(echo "$open_prs" | jq -c ".[$i]")
-
+	while IFS= read -r pr; do
 		if [ -z "$pr" ] || [ "$pr" = "null" ]; then
-			debug_log "Warning: Empty PR data at index $i"
+			debug_log "Warning: Empty PR data"
 			continue
 		fi
 
@@ -300,7 +298,7 @@ main() {
 				issue_prs_values[index]="${issue_prs_values[index]} $pr_number"
 			fi
 		done
-	done
+	done < <(echo "$open_prs" | jq -c '.[]')
 
 	debug_log "Finished processing all PRs for issue references"
 	debug_log "Found ${#issue_prs_keys[@]} unique issues referenced in PRs"