[PR #738] [MERGED] fix(bench): parse low-sample benchstat output in benchmark comments #751

New issue

Closed

opened 2026-02-26 22:32:23 +03:00 by kerem · 0 comments

kerem commented

2026-02-26 22:32:23 +03:00

Owner

📋 Pull Request Information

Original PR: https://github.com/rudrankriyam/App-Store-Connect-CLI/pull/738
Author: @rudrankriyam
Created: 2/23/2026
Status: ✅ Merged
Merged: 2/23/2026
Merged by: @rudrankriyam

Base: main ← Head: fix/benchmark-results-parse

📝 Commits (4)

0ed24c3 fix(bench): parse benchstat output with low-sample formatting
e909b66 fix(bench): render benchmark deltas with direction wording
c080b0d Remove unused change property from bench results
03d81f4 Merge pull request #745 from rudrankriyam/cursor/result-object-change-property-7957

📊 Changes

1 file changed (+52 additions, -17 deletions)

View changed files

📝 scripts/bench-comment.js (+52 -17)

📄 Description

Summary

fix benchmark comment parsing to handle benchstat rows that use ±∞/footnoted confidence output when sample count is low
parse only sec/op sections and skip B/op + allocs/op rows to avoid duplicate benchmark entries
keep fallback raw output rendering unchanged when parsing truly fails

Test plan

mkdir -p .perf && bash scripts/bench-ci.sh .perf/base.txt 1
bash scripts/bench-ci.sh .perf/pr.txt 1
PATH="$(go env GOPATH)/bin:$PATH" node scripts/bench-comment.js .perf/base.txt .perf/pr.txt .perf/comment.md
verified .perf/comment.md contains parsed benchmark table instead of "No benchmark results could be parsed"

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/rudrankriyam/App-Store-Connect-CLI/pull/738 **Author:** [@rudrankriyam](https://github.com/rudrankriyam) **Created:** 2/23/2026 **Status:** ✅ Merged **Merged:** 2/23/2026 **Merged by:** [@rudrankriyam](https://github.com/rudrankriyam) **Base:** `main` ← **Head:** `fix/benchmark-results-parse` --- ### 📝 Commits (4) - [`0ed24c3`](https://github.com/rudrankriyam/App-Store-Connect-CLI/commit/0ed24c357a7d663a89d607ffcc9a7134de0540b3) fix(bench): parse benchstat output with low-sample formatting - [`e909b66`](https://github.com/rudrankriyam/App-Store-Connect-CLI/commit/e909b66d1d41130031f0294e94c630e8d6df5a9a) fix(bench): render benchmark deltas with direction wording - [`c080b0d`](https://github.com/rudrankriyam/App-Store-Connect-CLI/commit/c080b0ded4048c672c62fc0debacaf30f83740ad) Remove unused change property from bench results - [`03d81f4`](https://github.com/rudrankriyam/App-Store-Connect-CLI/commit/03d81f478aa7822e0da7013bc925d55e66233bc4) Merge pull request #745 from rudrankriyam/cursor/result-object-change-property-7957 ### 📊 Changes **1 file changed** (+52 additions, -17 deletions) <details> <summary>View changed files</summary> 📝 `scripts/bench-comment.js` (+52 -17) </details> ### 📄 Description ## Summary - fix benchmark comment parsing to handle benchstat rows that use `±∞`/footnoted confidence output when sample count is low - parse only `sec/op` sections and skip `B/op` + `allocs/op` rows to avoid duplicate benchmark entries - keep fallback raw output rendering unchanged when parsing truly fails ## Test plan - [x] `mkdir -p .perf && bash scripts/bench-ci.sh .perf/base.txt 1` - [x] `bash scripts/bench-ci.sh .perf/pr.txt 1` - [x] `PATH="$(go env GOPATH)/bin:$PATH" node scripts/bench-comment.js .perf/base.txt .perf/pr.txt .perf/comment.md` - [x] verified `.perf/comment.md` contains parsed benchmark table instead of "No benchmark results could be parsed" --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>