Martin Robinson

> I didn't quite understand, those tests that fell, although it was expected to pass, is this a problem or are the tests like this? Auto-rendering tests write that 429...

I have pushed a new commit to this branch that: 1. Reworks a bit of the algorithm that you have written and added specification language to make it more readable....

It looks like one of the pre-existing unit tests failed because it expected an invalid `srcset` string to be parsed. I've updated the test and pushed to the branch again.

> ``` > > ▶ OK /html/semantics/embedded-content/the-img-element/srcset/select-an-image-source.html > > │ ▶ FAIL [expected PASS] "data:,a 1w 1h, data:,b 1w" > > │ │ → assert_equals: expected "data:,a" but got "data:,b"...

@jdm Do you have a link to the original log? The top of the stack trace is not included above.

Here is the test case: ```html .container { left: 0; top: 0; position: absolute; width: 100%; height: 100%; overflow: hidden; perspective: 2000px; perspective-origin: 20% 50%; background: yellow; } .content {...

I still see this issue even with all the recent improvements to ClipScrollNode footprint calculation.

Closing this for now as we don't support MagicLeap at the moment.

I'm some ways it is easier if there is only one way to install dependencies. It's quite common that something breaks without anyone realizing and the instructions quickly get out...

> @mrobinson I saw that you previously [mentioned that choco is not used in CI](https://github.com/servo/servo/issues/32342#issuecomment-2124885906). Is that not the case anymore, or am I missing some context? Oh yeah, great...