This commit itself only moves things around and adds an extra parameter to the
`apply_declarations` function to eventually handle #14079 correctly.
Probably needs a more granular API to query fonts, á la nsFontMetrics, but
that's trivial to do once this is landed.
Then we should make the font provider mandatory, and implement the missing stylo
bits.
This patch introduces infrastructure for the rule tree, and constructs it.
We don't use it yet, nor have good heuristics for GC'ing it, but this should not
happen anymore once we store the rule node reference in the node.
I haven't messed up with memory orders because I want to do a try run with it,
then mess with them.
Take down the ApplicableDeclarationsCache, use the rule tree for doing the cascade.
We also make sure that an element never has an ElementData with ElementDataStyles::Uninitialized,
and eagerly call prepare_for_styling whenever an ElementData is instantiated.
MozReview-Commit-ID: 9YP6eSmdMt0
Instead of maintaining a dummy restyle damage field for text nodes, we can just
perform the necessary parent inheritance and is_changed adjustments on the fly
in ThreadSafeLayoutNode, simplifying the requirements on the style system.
MozReview-Commit-ID: DCqiCRLsLUF
A couple of changes here:
* Remove the option to unset with the dirty bit setters.
* Add an explicit API for setting text node style.
* Hoist has_changed handling into the restyle damage setter and text node style setter.
* Make set_style take a non-Option.
Refactor style logic to avoid direct access to the node data during the cascade
The new restyle architecture doesn't store these things in consistent places, so we need a more abstract API.
<!-- Reviewable:start -->
---
This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/13656)
<!-- Reviewable:end -->
We don't need this for Gecko, and it's hard to implement in that case because
there's nowhere obvious to put it (we don't plan to create TSDs for non-dirty
nodes, and non-dirty nodes can have dirty children which require the
children_to_process atomic). There are various solutions here, but punting is
the easiest.
We'll need to rethink this if/when we need to do a bottom-up traversal for
Gecko.
Start simplifying FFI ownership sugar
This is step one of a series of changes planned to make the ownership sugar easier to use. This was blocked on #13038
*very* unsure about second commit. Don't like the thought of accepting types with destructors over FFI. Probably will revert it. Leaving it in for now in case you have some insight.
Eventually at least for the borrowed stuff I want to use T directly (because the crates are merged now), instead of the fake void types. Perhaps for the others too. I might include those changes in this PR -- was originally planning to but I realized that these steps could be split out.
Tentative plan for `Owned` (assuming it's not removed) is to have `Owned<T> <-> Box<T>` (same `T`, no "FFI type") conversions. We will only use ownership wrapper types for things with destructors, and try to keep the conversion simple. I'm envisioning a couple methods for arc/strong and a few more for box/owned. We may need to introduce new wrapper types for gecko-side managed objects (`RefPtr<T>`, `UniquePtr<T>`) but that should be all the wrapper types we have in the ownership sugar.
This PR relies on the guarantee that `Option<&T>` and `Option<Box<T>>` are pointer-sized via the `NonZero` optimization. I am now less unconvinced that this is a good idea 😉.
r? @emilio
cc @mystor
<!-- Reviewable:start -->
---
This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/13569)
<!-- Reviewable:end -->
