Re: [ServiceWorker] Please deploy dfn.js or something similar (#489)

Is the only thing stopping us from moving to bikeshed that nobody has volunteered yet to put in the work needed to do so? Or are there particular reasons why we might want to delay moving to bikeshed. If the consensus is that yes, we should move to bikeshed, I'd be happy to volunteer to do so (I've converted two other specs to bikeshed, so I've some idea what I'm signing up for, although those were of course way smaller than the service worker spec).

Anyway, some thoughts about the consequences of a conversion to bikeshed:

- Styling will be different, hopefully for the better.
- Deep links to sections and definitions should mostly remain the same, but links to IDL definitions will change as bikeshed doesn't have a way to specify your own anchors for those (even if it would I don't think we'd want to hardcode these anyway). Some section anchors might have to change too for this, since they conflict with the bikeshed generated IDL anchors.
- Any outstanding pull requests or other changes people are working on will have to be manually rebased. This also might be a reason to do the conversion while v1 and v2 haven't diverged too much yet, as then we can convert both without extra effort.
- Major changes to the spec while I'm working on the conversion to bikeshed would of course be annoying as I'd have to manually re-apply them on top of the bikeshed version. But the spec is probably stable enough at this point for this to not matter too much for however long this will take; re-applying a week (or a few weeks) worth of changes on top of a bikeshed conversion should be relatively painless.
- Actually reviewing the conversion might be hard, since it'll be a fairly substantial change. Although splitting the conversion in several commits might make this slightly better, in the end still pretty much every line of spec source will change for this:
  - one commit to update the overall structure to be bikeshed compatible while keeping most of the actual content unchanged
  - one commit to update definitions and references to them to be bikeshed compatible; we could choose to keep explicit links to locally defined terms to keep the diff somewhat managable (and optionally clean those up in a separate commit), or just go all the way and clean up internal links as well. Either way this will be a pretty nasty diff to review as it'll change so much all over the place.
  - other cleanups as we see fit.

So is there consensus that moving to bikeshed is a good thing, and if so, when would it make sense to do so?

---
Reply to this email directly or view it on GitHub:
https://github.com/slightlyoff/ServiceWorker/issues/489#issuecomment-181654310