Reduce size of spatial node index

[djg]

Fixes #3431.

---

This change is 

[djg]

@gw3583 would you care to review this one too?

[zptan]

@kvark @gw r?

@djg Is it better to have a try run?

[emilio]

How's this needed? It should be Send by default anyway, right?

[kvark]

This really does much more than just "Reduce size of spatial node index", introducing the IndexVec abstraction, which I have mixed feelings about. I wish we could separate these two and land/discuss independenty.

Reviewed 15 of 15 files at r1.
Reviewable status: all files reviewed, 5 unresolved discussions (waiting on @emilio and @djg)

---

webrender/src/clip_scroll_tree.rs, line 52 at r1 (raw file):

impl Idx for SpatialNodeIndex {
    fn new(value: usize) -> Self {
        assert!(value < u32::max_value() as usize);

nit: debug_assert!?

---

webrender/src/clip_scroll_tree.rs, line 61 at r1 (raw file):

Previously, emilio (Emilio Cobos Álvarez) wrote…

How's this needed? It should be Send by default anyway, right?

agreed, this shouldn't need any unsafe impls

---

webrender/src/gpu_types.rs, line 473 at r1 (raw file):

            from_index
        } else if from_index == to_index {
            SpatialNodeIndex::new(0)

perhaps, use ROOT_SPATIAL_NODE_INDEX here instead?

---

webrender/src/gpu_types.rs, line 648 at r1 (raw file):

        let index = metadatas.push(metadata);
        let _index2 = transforms.push(data);
        assert!(index == _index2);

nit: assert_eq!. Also, we shouldn't mark it as unused with _ since it's used in the assertion

---

webrender/src/index_vec.rs, line 31 at r1 (raw file):

#[cfg_attr(feature = "capture", derive(Serialize))]
#[cfg_attr(feature = "replay", derive(Deserialize))]
pub struct IndexVec<I: Idx, T> {

I'm not convinced this is a helpful abstraction. It carries almost zero logic in itself. WR has used u32 indices (newtyped for safety) for a lot of stuff, and that hasn't been a problem (i.e. not a burden to write, and not a source of mistakes).

[bors-servo]

☔ The latest upstream changes (presumably #3441) made this pull request unmergeable. Please resolve the merge conflicts.

[djg]

I've re-written the patch to not use type safe index nor the Vec wrapper.

[djg]

Try: https://treeherder.mozilla.org/#/jobs?repo=try&revision=5ea9b6e7fabb059fda90156aa39e334e382abff9&selectedJob=219536839

[gw3583]

@bors-servo r+

[bors-servo]

📌 Commit 2f79d63 has been approved by gw3583

[bors-servo]

⌛ Testing commit 2f79d63 with merge 075fd68...

[bors-servo]

☀️ Test successful - status-appveyor, status-taskcluster
Approved by: gw3583
Pushing 075fd68 to master...