François Daoust

> Probably related to [tabatkins/bikeshed#2379](https://github.com/tabatkins/bikeshed/issues/2379) 😕 That was the cause. Bug has now been fixed and Bikeshed generation no longer chokes on missing headings.

If the logic is to look at `series.nightlyUrl` and pick up the first match, I think it needs to be amended. The nightly URL of a spec is under `nightly.url`....

> Arguably an issue split between here and web-specs, but there are cases where we might want multiple levels of specs. That is already somewhat doable but it is hard...

Actually triggered by invalid IDL in the WebGL1 spec introduced 3 days ago, see https://github.com/KhronosGroup/WebGL/issues/3407. It would be good to catch that invalid IDL error before attempting to generate IDL...

Some notes as I'm looking into adding a check for unicity for CSS property definitions. There are a number of properties that are defined more than once in practice. In...

> shape-inside should rather be defined with a newValues field in css-round-display. Actually, the definition should rather be dropped entirely since the new `display` value. Filed https://github.com/w3c/csswg-drafts/issues/6433

> I suppose that properties duplicated in a higher version level of the same specification, eg. `background-clip` in `css-backgrounds` and `css-backgrounds-4`, are filtered out from your above list, because it...

Looking into de-duplicating CSS properties again with the goal of dropping duplicates during the data curation phase, I think that the following logic, to be discussed, would work: 1. Ignore...

> I think I'd rather an entire delta spec be ignored than only some parts of it. I'm thinking to CSS types extracted in valuespaces that may only be used...

Looking at value spaces, applying the same logic would allow us to get down to a list of 6 duplicates, which I don't think we can easily get rid of:...