Draft for Firefox 155 release notes

[jakearchibald]

I've collated release notes based on Bugzilla & WPT pass changes. When it comes to WebDriver & WebRTC, I'm not sure which of these changes are worth going in the release notes, so that needs review.

I'm also not sure how 'complete' we are with experiments. For example, I don't think custom <select> is complete enough for testing, so maybe it shouldn't be included?

[github-actions]

Preview URLs (1 page)

• /en-US/docs/Mozilla/Firefox/Releases/155

Flaws (4)

Found an unexpected or unresolvable flaw? Please report it here.

URL: /en-US/docs/Mozilla/Firefox/Releases/155
Title: Firefox 155 release notes for developers (Beta)
Flaw count: 4

• broken_links:
  • Link /en-US/docs/Web/JavaScript/Reference/Regular_expressions/Buffer_boundary_assertion doesn't resolve
• macros:
  • Macro jsxref produces link /en-US/docs/Web/JavaScript/Reference/Promise.allKeyed() which doesn't resolve
  • Macro jsxref produces link /en-US/docs/Web/JavaScript/Reference/Promise.allSettledKeyed() which doesn't resolve
  • Macro domxref produces link /en-US/docs/Web/API/WebTransport/draining which doesn't resolve

External URLs (52)

URL: /en-US/docs/Mozilla/Firefox/Releases/155
Title: Firefox 155 release notes for developers (Beta)

• https://bugzil.la/1278697 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/1477920 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/1642578 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/1692434 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/1911075 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/1924328 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2007150 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2007160 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2007165 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2007174 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2007200 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2014357 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2018900 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2019361 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2029965 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2037673 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2038940 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2042999 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2043000 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2045230 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2046602 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2047345 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2047706 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2052949 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2053296 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2053297 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2055211 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2055710 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2055774 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2055911 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2056412 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2057270 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2059738 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2059947 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2059988 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2060529 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2060754 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2060873 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2060972 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2061646 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2062344 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2062374 (1 time) (Note! This may be a new URL 👀)
• https://bugzil.la/2062892 (1 time) (Note! This may be a new URL 👀)
• https://datatracker.ietf.org/doc/html/draft-ietf-happy-happyeyeballs-v3 (1 time) (Note! This may be a new URL 👀)
• https://firefox-source-docs.mozilla.org/devtools-user/debugger/index.html (1 time) (Note! This may be a new URL 👀)
• https://firefox-source-docs.mozilla.org/devtools-user/json_viewer/index.html (1 time) (Note! This may be a new URL 👀)
• https://firefox-source-docs.mozilla.org/devtools-user/page_inspector/how_to/examine_and_edit_css/index.html (1 time) (Note! This may be a new URL 👀)
• https://github.com/WebAssembly/compact-import-section (1 time) (Note! This may be a new URL 👀)
• https://github.com/WebAssembly/wide-arithmetic (1 time) (Note! This may be a new URL 👀)
• https://github.com/tc39/proposal-await-dictionary (1 time) (Note! This may be a new URL 👀)
• https://github.com/tc39/proposal-regexp-buffer-boundaries (1 time) (Note! This may be a new URL 👀)
• https://jsonlines.org/ (1 time) (Note! This may be a new URL 👀)

(comment last updated: 2026-08-27 13:10:07)

[hamishwillee]

I'm also not sure how 'complete' we are with experiments. For example, I don't think custom <select> is complete enough for testing, so maybe it shouldn't be included?

@jakearchibald FWIW I don't think we have a hard rule. I chose not to include Web API stuff for this topic in release notes or experimental features pages for FF149 relnote because it is untestable, but another author decided to include the base-select value.

My own opinion is that the point of the release note and experimental features updates is to surface testable features: if they aren't testable then the notes are confusing. That said, if we're semi-automatic notes, it is probably easier to include everything than to work out how to exclude things that aren't ready yet.

[jakearchibald]

That said, if we're semi-automatic notes, it is probably easier to include everything than to work out how to exclude things that aren't ready yet.

The way I did this:

1. I used https://github.com/jakearchibald/gather-new-browser-features to generate comprehensive release notes. That tool uses WPT as its primary source, and cross-references with bugzilla and Firefox source.
2. I human-chose which features should be part of the MDN release notes. This is where my lack of knowledge of on particular topics has an impact.
3. I had Claude generate release notes in the style of previous release notes.

If we wanted to explore a pattern like this going forward, I think I'd rather have AI generate too much, then have a human go through and delete anything that doesn't look useful (and of course correct anything that looks wrong).

But, again, this is just an experiment, so if it's not useful, I'll stop 😄

[hamishwillee]

But, again, this is just an experiment, so if it's not useful, I'll stop 😄

@jakearchibald Definitely useful automation. Please don't stop!! For now I will copy paste what I see as useful.

Part 2 is obviously a problem (not least because it is manual), but consider that your uninformed opinion is likely more accurate than mine most of the time.

Your process looks like what I would do. Two thoughts:

• I'd get claude to do a first pass with a prompt like "mark up the items you think are likely to have no developer impact, such
• When/if this is used we should change the docs issue template to review the release note for each area rather than create it. That should help us create a feedback loop to iterate this.

[jakearchibald]

• I'd get claude to do a first pass with a prompt like "mark up the items you think are likely to have no developer impact, such

I'm already doing that, but from the subjects I know in detail, I know it isn't particularly good at it. That's why I erred on leaving more things in when it came to topics I'm less familiar with.

[jakearchibald]

https://docs.google.com/document/d/1E6TA04Lcr3T8ctiC0Pti6b_UlxLlGAx9WYaZjXTm6zk/edit?usp=sharing here's the full generated release notes that I used to generate these MDN release notes.

Definitely some sus bits in there, but it's been useful for me to gather topics to look into for the next release.

[Rumyra]

This looks good - I'd actually be ok if we wanted to merge this, I just wonder how many conflicts we would get with other open release note prs...

I would want developer tools to be further down - but that's personal preference.

Also we'd still have to update the experimental features page here https://developer.mozilla.org/en-US/docs/Mozilla/Firefox/Experimental_features

[dipikabh]

Reconciling with updates from #45268

[dipikabh]

Reconciling with notes from #45267

[dipikabh]

Reconciling with note from #45244

[dipikabh]

Reconciling with note in #45212

[dipikabh]

Reconciling with the note in #45207

[dipikabh]

Reconciling with #45236

[dipikabh]

Reconciling with #45224

[dipikabh]

Thanks for putting this together!

I've completed a readthrough.
Let's also wait to hear from @Rumyra before we land this.

[Rumyra]

With @dipikabh 's comments accepted I'm happy to get this merged 👍

[hamishwillee]

@Rumyra FWIW I love the automation and we need it.
But my initial experience was two questionable notes in WebTransport - which may be fine if we have some appetite for these release notes being wrong.

Specifically, if we're doing it this way, am I still responsible for my project issues to have correct release notes?

If so, proposal - update the release note issue template from:

- [ ] Add entry to Firefox release notes for enabled/preview features

To

- [ ] Review Firefox release notes for feature

[jakearchibald]

@hamishwillee my take is: the AI is not too be trusted. I reviewed these notes myself, but like I said, there are areas like WebRTC, WebTransport, and WebGL which are blind spots for me, so those areas in particular need a close look from others.

If the effort is fixing these generated notes is greater than the previous process, then this new process is bad 😀

Fwiw I put up a PR for 156 #45357 - I did fact checking for the stuff I know, but it should be considered slop until it's properly reviewed.