Gentle ping @sandandsnow. We’d like to receive your review for the general direction of this PR by our TPAC meeting webmachinelearning/meetings#35

An alternative to this proposal is a minimal change as outlined in #886 (comment)

The group is eager to hear the Privacy WG preference. If you’re indifferent wrt this I’d be inclined to propose the minimal change instead. Thank you for your advice.

@anssiko, thank you for the reminder. I had missed the ask to review the PR.

I have this on our TPAC agenda now but if we're able to resolve this sooner the better and we can free time for other privacy topics.

Thanks for your review and suggestions, @reillyeon.

@sandandsnow, if you feel further discussion is required, we have allocated a slot for privacy topics at TPAC webmachinelearning/meetings#35 within the 10 November 2025 15:30-16:45 JST timebox for a higher bandwidth discussion.

We had a great discussion on privacy considerations in Kobe (minutes). While we did not do a line-by-line review of this particular PR there, I feel comfortable merging this PR now given the directional guidance.

(@deniak, the auto-publish job failure looks like it could be related to the recent specberus 12.0.0 release? I'm not familiar with that codebase, so just a guesstimate.)

@anssiko the publication failed because the spec didn't pass the validator. The role directory was deprecated and the support removed from the validator recently.

@deniak thanks for pointing me to the right direction. I opened an issue for bikeshed: speced/bikeshed#3204

That was fast. The fix landed in bikeshed 5.4.2. When auto-publish updates to the latest bikeshed the build is good again.

FTR, the latest bikeshed with the fix was propagated to auto-publish. I triggered a new build and it passed. The normal publication workflow is operational again. ✅