Infrastructure for joint W3C-WHATWG publication

[foolip]

The Memorandum of Understanding Between W3C and WHATWG lays out the process for WHATWG Review Drafts going through the REC process to be endorsed by the W3C, after which the Review Draft is updated to reflect this.

This has happened for the first time now with DOM, see https://dom.spec.whatwg.org/review-drafts/2019-06/. The details of how it was produced are in https://github.com/whatwg/dom/tree/review-drafts-w3c-additions, and updates were also posted in whatwg/dom#767.

Remaining work to make this smoother for future publications:

• Land Add new feature for injecting MDN panels speced/bikeshed#1564 so that it need not be applied manually
• Extract the scripts in whatwg/dom@be3bc51 somewhere
• Update Wattsi to support making similar changes for HTML
• Track the built output in /review-drafts/ (or more) for each spec in version control
• Graft https://github.com/whatwg/dom/tree/review-drafts-w3c-additions onto that history
• Deploy to whatwg.org from that repository

Related issues:

• Review draft and revamped snapshot work coordination thread is where the review draft was originally developed
• Move ever-growing *.spec.whatwg.org storage off of the VM disk could be solved together with this issue, but need not be
• Deploy changes to policy files to whatwg.org is similar in that it's publishing built output to whatwg.org

@sideshowbarker @whatwg/editors @whatwg/sg

[annevk]

I think we should also extract https://github.com/whatwg/dom/tree/review-drafts-w3c-additions somehow if we want to preserve that. I'd rather not have to keep non-master branches around.

[foolip]

Yes, we should transplant that history onto whatever solution we end up with and then delete the branch. I'll add a step.

[foolip]

Some stats on /var/www on the whatwg.org server (named "marquee") to inform how what we could/should do for version control:

• A total of ~116k files totaling ~40GB uncompressed and without deduplicating
• Putting all of that into a git repo and running git repack -a -d -f --depth=250 --window=250 results in a ~440MB repo (The existing whatwg.org repo is already ~340M.)

By comparison, the review-drafts directories have only ~500 files. Putting only those under version control would be no problem at all.

[TimothyGu]

Has hosting static files on GitHub Pages been considered? It doesn't sound like it would be too much effort, and we get (essentially) free storage.