::details-content - unfairly partial for Safari?

[jakearchibald]

Related web-platform-dx/web-features#3382

The baseline banner on MDN shows a cross for Safari, but the limited support here feels like an edge-case.

It feels like the missing part should be a sub feature, or at least not mark the whole feature as partial in Safari.

[jakearchibald]

As per https://github.com/mdn/browser-compat-data/tree/main/docs/data-guidelines#partial_implementation-general-usage-guidelines

The browser's support does not implement mandatory specified behavior.

True.

The browser's support is inconsistent with at least one other browser.

True, Chrome & Firefox.

The browser's support causes confusing feature detection results.

False. CSS.supports('selector(::details-content::before)') returns false as expected in Safari.

The browser's support has a demonstrable negative impact on web developers.

I don't think so.

[lukewarlow]

It was originally marked as partial based on the fact it didn't fully implement the features as it's been specced since first shipping in Chromium (and Firefox also deliberately shipped multiple releases later to avoid this partial feature release). The bug I raised for this: #27620

It's subsequently been fixed on WebKit trunk so will at least be fully supported soon ™️

Fwiw my opinion is that something being an edge case doesn't necessarily warrant it not being marked as partial. Introducing an interop difference on a brand new feature is an example where I think partial is warranted even when it doesn't meet the criteria to be a must include.

[nt1m]

Fwiw my opinion is that something being an edge case doesn't necessarily warrant it not being marked as partial. Introducing an interop difference on a brand new feature is an example where I think partial is warranted even when it doesn't meet the criteria to be a must include.

I'm more bothered about the baseline banner than the MDN compat data table to be honest. The baseline banner suggests you can't use ::details-content at all in production, for something that is a very specific edge case (that was intentionally left feature detectable fwiw). I don't think that is helpful for developers at all.

[lukewarlow]

I'll defer to Dev rel and the mdn team on what they think is most useful for developers. If they are looking just to baseline then yeah they might decide not to use even though they can for their use case.
Given this issue will soon be fixed it's less important in the long run, but perhaps the guidance on partial for when an implementation doesn't ship the whole specced feature needs a bit more detail.