These have been published by @emilio:

• https://crates.io/crates/html5ever/0.36.0
• https://crates.io/crates/web_atoms/0.1.4

Hi, according to this, it seems that web_atoms should be bumped to 0.2.0 instead due to phf breaking change from #654. The 0.1.4 release may need to be yanked.

Even if phf had potentially breaking changes, I don't think that web_atoms had any, so I thought that a 0.1.4 was enough?

I’m actually not entirely sure what is the correct answer.

The question is what would happen if old html5ever release pulls the latest web_atom of which a breaking release phf is used underneath. The breaking release phf may cause trait conflicts or inconsistent hash being used. But since html5ever itself does not depend on phf directly. It may be fine? 🤷‍♂️

No problem from my side if you want to bump to 0.2.0

Yes, we'll need to yank and publish 0.2.0 instead.

I can confirm that the Blitz build does actually break if I run cargo update (which causes it to pick up web_atoms 0.1.4), so we definitely need to republish and yank this release. I've put up #691 with new versions.