Shipped · Observability

What sitespeedio/sitespeed.io shipped

The public repository of Sitespeed · github.com/sitespeedio/sitespeed.io

Written by FoxPlug from the repository's own releases; not affiliated with Sitespeed. Paste your own repository in the box below and FoxPlug writes the same weekly update for your product. This page is an automatic summary of the releases, pull requests and commits in github.com/sitespeedio/sitespeed.io; Sitespeed did not write it and does not use or endorse FoxPlug. Every line links to the change it describes.

Get a weekly update like this for your product, free

Or Use it as a GitHub Action

See the update FoxPlug writes for sitespeed.io, free

Week of November 2, 2026

What shipped

Why it matters

This week fixes a debug mode that never worked on desktop and removes a warning that showed up on every install in a checkout. It also moves to Browsertime 29 with drivers installed at build time in the Docker images, and narrows a sudo rule that could hand out root in containers. The video docs now explain what you lose when you skip conversion.

Changelog entry

Example posts FoxPlug drafted from these changes. Not written or posted by the project.

Post for X

This week in Sitespeed: --debug works again on desktop, npm install no longer warns about core-js, Browsertime 29 installs drivers at container build time, the throttle sudo rule no longer gives root in Docker, and the video docs explain what you get when you skip conversion.

Post for LinkedIn

This week in Sitespeed: --debug works again on desktop after being blocked by an Android check that always matched and a misspelled timeout key. npm install in a checkout no longer prints the core-js warning from pug-lint. We upgraded to Browsertime 29, which downloads a driver the first time a test needs it, and our Docker images now install drivers when built so container tests do not depend on a run time download. The throttle sudo rule now allows only the tc and ip subcommands, so it no longer gives root in Docker. The video docs explain what you get when you skip conversion.

Week of October 26, 2026

What shipped

Why it matters

Several fixes this week target failures that were easy to miss: a failed script no longer wipes out the whole report, broken result links now match where uploads actually go, and the compare plugin no longer quietly produces an empty tab. The Docker work moves ARM64 builds onto native runners, so ARM64 images are built the same way the ARM64 tests run rather than through emulation.

Changelog entry

Example posts FoxPlug drafted from these changes. Not written or posted by the project.

Post for X

This week: failed scripts no longer drop the whole report, result links follow the S3/GCS upload path, and compare finds python3. Plus ARM64 Docker images now build on native runners.

Post for LinkedIn

This week in the project: a script that fails mid-measurement no longer drops the entire Browsertime result, so the report keeps the steps that measured fine and shows the real error. Result links in Grafana/Graphite annotations, Slack and Matrix now follow the custom S3 or GCS upload path instead of pointing at a page that does not exist. The compare plugin finds python3 on systems that only ship it, instead of finishing with an empty tab. Unknown connectivity engine values are rejected. On the Docker side, release and autobuild images for amd64 and arm64 now build on native runners, the correct WebPageReplay binary is used on ARM64, and build-essential is gone from the image.

Week of October 19, 2026

What shipped

Why it matters

This week fixed several bugs that could stop or confuse a run: a CrUX network error no longer halts sitespeed.io, comma-separated plugins now get their options, and repeated visualElements are kept. Metric lookups with bracket paths work again, so First Contentful Paint appears in reports, and compare no longer crashes with plugin-added metrics. Browser names are case-insensitive and the report opens without a shell.

Changelog entry

Example posts FoxPlug drafted from these changes. Not written or posted by the project.

Post for X

Fixed this week: CrUX errors no longer stop the run, comma-separated plugins get their options, repeated --scriptInput.visualElements are kept, bracket metric paths work again, compare no longer crashes, browser names are case-insensitive, and the report opens without a shell. Re

Post for LinkedIn

This week brings several fixes for sitespeed.io users. A CrUX network error or non-JSON response no longer stops the whole run; it is reported as a plugin error and results are kept. Plugins added comma-separated now register their CLI options, so InfluxDB and similar work with --plugins.add. Repeated --scriptInput.visualElements are all kept. Bracket paths in metric lookups work again, restoring First Contentful Paint in the Pages table, summary box and run page. Compare no longer crashes when plugins add extra metrics. Browser names are accepted in any case. The Third party tab renders when coach is disabled. The HTML report opens without a shell. Release 42.10.0 includes Browsertime 28.5 with new Chrome and Edge drivers and browsers in Docker.

Week of October 12, 2026

What shipped

Why it matters

This week fixes two silent problems: plugin data that vanished when the coach was on, and iOS pre-warming that could pick a different device than the real test. It also brings an updated WebPageReplay whose inject script does less work per replayed response, and both 42.8.0 and 42.9.0 are out.

Changelog entry

Example posts FoxPlug drafted from these changes. Not written or posted by the project.

Post for X

Sitespeed.io 42.9.0 is out with Browsertime 28.4.0. Plugin scripts are kept when the coach is on, iOS pre-warming now uses your configured device, and WebPageReplay prepares injected HTML once per archive load instead of per request.

Post for LinkedIn

Sitespeed.io 42.9.0 is out, with Browsertime 28.4.0 and an updated WebPageReplay. When the coach was enabled, scripts sent by other plugins through browsertime.scripts could be lost, so those plugins never got their data and no error was reported; that is now fixed. Pre-warming for iOS had a typo in the option name, so it ignored --safari.deviceUDID and could run on a different device than the real test when several were available; it now uses the configured device. The new WebPageReplay prepares the injected HTML once when the archive loads rather than on every request, removing CPU work from each replayed response. Result URLs also no longer use the deprecated url.parse.

More from sitespeedio/sitespeed.io on GitHub

Merged pull requests

Releases

Weeks with too little public activity are left out rather than filled in. Last updated 2026-11-10.

Is this your repo? Ask us to remove this page.