Repository navigation
Benchmark regression: insert (main) #678
Description
Bencher flagged these benchmarks in the nightly run (insert, main) on two runs in a row:
insert/batch_100/100k
Branch: main
Run: https://github.com/project-minigraf/minigraf/actions/runs/39559734592
Previous run: https://github.com/project-minigraf/minigraf/actions/runs/39385015408
Please review the Bencher dashboard to confirm the regression.
Activity
adityamukho commentedon Oct 21, 2026
Not a regression: runner noise.
A/B on an idle dedicated machine,
insert/batch_100, dcbc19e (last fast nightly) vs e434148 (v2.0.9), run in parallel and swapped each round, 4 rounds:Criterion found no change in 30 of 32 comparisons; the other two were single 1m rounds on e434148, +8.8% and −7.0%, which cancel.
Bencher history agrees:
insert_file/batch_100runs the same parse + transact path and was flat or faster (0.71–0.95×) on those runs.parse_transact(PRS-108, fix(parser): reject a fact with more than four elements, PRS-108 (#649) [v2.x backport] #656).🤖 Generated with Claude Code
https://claude.ai/code/session_01PtCdzbJMSS1V8qEPaJ4jXz