The HTTP cache is now stable enough that we can ask more people to help
us testing it. So let's turn it on by default! It can be turned off with
--disable-http-cache if needed.
...navigable shared attribute processing steps.
After e095bf3a5fhttps://wpt.live/html/rendering/pixel-length-attributes.html ends up in
this function with null navigable, which leads to crash while executing
steps. Adding early return in case iframe's document doesn't have a
navigable is enough to make the test pass again.
This commit doesn't import the WPT test, because it's already imported
and we have it disabled since it takes too much time on CI.
This behaviour should only apply to literal percentages as clarified
here: 4ee8429
We were also doing this wrong by converting the numeric type of the calc
to Length which was causing values to be defaulted to 1 instead (hence
the new passing tests for computed values as well)
This ensures that we clamp values for properties like padding-* to valid
ranges (non-negative in this case) if they are specified with `calc()`.
The length-related changes in this commit combined with the ones from
the previous commit fix the primary layout issue on https://lwn.net
(yes, not the first place I would have expected problems either).
According to ECMA-262 §15.4.5 (MethodDefinitionEvaluation),
getters and setters defined in class bodies
must create property descriptors with
[[Enumerable]]: false. Previously we incorrectly marked them enumerable.
This patch updates `ClassMethod::class_element_evaluation` so that both
getter and setter descriptors use `.enumerable = false`.
Previously, the hosts file was updated at the same time the WPT repo
was cloned, but updating the hosts file is only necessary for commands
that start the WPT test runner.
SIMD loads/stores checked bit 0x20 of the align immediate to detect a
following memory index, unlike scalar mem ops which use 0x40 per the
multi-memory encoding. This caused the memidx byte to be misparsed as
the next immediate (e.g. offset).
Update both SIMD sites (v128 load/store and lane variants) to check and
clear 0x40, then read LEB128<u32> memidx.
Repro:
(module (memory $m0 1) (memory $m1 1)
(func (export "go")
i32.const 0
v128.load (memory 1)
drop))
Before: printed memidx 0 with offset 1.
After: prints memidx 1 with offset 0.
The opcode entry declared i16x8_replace_lane with pushes = -1, but
replace_lane pops 2 (vector, lane value) and pushes 1 result vector.
Set pushes to 1 to match the other replace_lane opcodes.
Negate was incorrectly returning "== 0", a copy/paste from EqualsZero.
This patch corrects it to return "neg", matching the operator's actual
semantics and WebAssembly mnemonics (f32.neg, f64.neg).
Replaces spin until with GC-allocated counting object that invokes
destruction callback once all child navigable documents are destroyed.
The change doesn't have a test but not using spin until is strictly
better than using it. Also improves https://www.rottentomatoes.com/
where previously we would hang or crash after loading.
...selector. Grammar per spec: `::slotted( <compound-selector> )`, so
we should reject selector as invalid if first compound selector is
followed by something else.
This change makes layout more correct on https://www.rottentomatoes.com/
This has always been a bit of a hack. Initially it made sense as a lot
of properties that accept a length also accept `auto`, but while
convenient, that leads to problems: It's easy to forget to check if a
length is `auto`, and places that don't accept it end up with an
invalid state lurking in the type system, which makes things unclear.
Not every user of this requires an `auto` state, but most do.
This has quite a big diff but most of that is mechanical:
LengthPercentageOrAuto has `resolved_or_auto()` instead of `resolved()`,
and `to_px_or_zero()` instead of `to_px()`, to make their output
clearer.
...instead of `auto` Lengths.
This also fixes interpolating between two `auto` `<bg-size>`s, which
fixes a lot of animation tests for both `background-size` and `mask`.
Any type of Size which has no LengthPercentage value now uses an empty
optional instead of making an auto Length as before.
We also now serialize a `fit-content` Size as `fit-content` instead of
`fit-content(auto)`, though this doesn't affect test results and I
didn't identify where it's actually used.
The spec is a bit awkward here: A few algorithms create an "empty"
SourceSet, and then assign its source-size value a few steps later, so
we have a temporary state with no length. In order to avoid complicating
the types with Optional, I've chosen to just assign it to 0px.
Previously we used `auto`, but `auto` is not a valid value here - it is
used inside the "parse a sizes attribute" algorithm, but that always
returns an actual length (or calc).
We only had one user of this which allows `auto`, which is length_box().
So, inline the code there, remove the `auto` branch from
length_percentage(), and then remove length_percentage_or_fallback()
entirely as nobody uses it any more.
