Skip to content

Instantly share code, notes, and snippets.

@92kns
92kns / tp6_report_final.md
Created April 16, 2026 19:51
tp6 Pageload Drop Report: Desktop + Mobile (Apr 2026, post-verification)

tp6 Pageload Drop Report: Desktop & Mobile

Context: Bug 2025545 — re-record everything with mitm12. Apr 16, 2026 meeting outcomes: drop login sites, drop noisy sites (bug 1991544), aggressively trim the list.

Metrics used

Metric What it means Threshold for keep
@92kns
92kns / tp6_drop_proposal.md
Created April 16, 2026 19:35
tp6 Pageload Drop Proposal — login + noisy sites (Apr 2026)

tp6 Pageload: Drop Proposal

Context: Meeting outcome Apr 16, 2026. Action items: drop login sites, drop noisy sites (bug 1991544), share query results ahead of re-recording (bug 2025545).

Data: 375 days of Perfherder alert data (autoland + mozilla-central). Unique catches = regressions only this suite detected. Fix rate = (fixed + backedout) / resolved.


Immediate drops: login sites

@92kns
92kns / tp6_site_analysis.md
Created April 16, 2026 16:34
tp6 Pageload Site Analysis: Keep vs Drop (mitm12 re-recording)

tp6 Pageload Site Analysis: Keep vs Drop

Context: We're re-recording everything with mitmproxy 12 (bug 2025545). This is the natural moment to aggressively reduce the site list.

Data: 375 days of Perfherder alert data (autoland + mozilla-central, all platforms). Criteria: unique regression catches + fix ratio + login burden.

A "unique catch" = a regression only this suite detected. Fix ratio = (fixed + backedout) / resolved.