Assign all ClipIds during display list construction

[mrobinson]

This change makes it so that all ClipIds are assigned during display
list construction. This will allow us to count the number of ClipIds in
each display list during construction, making it possible to more easily
map them onto simple array indices. The final goal of this work is to
eliminate the performance cost of hashing when flattening and rendering
display lists.

---

This change is 

[mrobinson]

Gecko try job is here: https://treeherder.mozilla.org/#/jobs?repo=try&revision=9e17e800819fb5db5e92a17d2ab9dae467b8b3f0

[kvark]

---

Reviewed 6 of 6 files at r1.
Review status: all files reviewed at latest revision, 3 unresolved discussions.

---

webrender_api/src/display_item.rs, line 763 at r1 (raw file):

impl ClipId {
    pub fn root_scroll_node(pipeline_id: PipelineId) -> ClipId {
        ClipId::Clip(1, pipeline_id)

can we define it as a constant?

---

webrender_api/src/display_list.rs, line 812 at r1 (raw file):

        let start_time = precise_time_ns();

        // We start at w here, because the root reference is always 0 and the root scroll

start at w?

---

webrender_api/src/display_list.rs, line 1316 at r1 (raw file):

        filters: Vec<FilterOp>,
    ) {
        let reference_frame_id = if transform.is_some() || perspective.is_some() {

to clarify, we aren't checking for axis alignment here because an animated transform always spawns a reference frame, right?

---

Comments from Reviewable

[mrobinson]

Review status: 4 of 6 files reviewed at latest revision, 3 unresolved discussions.

---

webrender_api/src/display_item.rs, line 763 at r1 (raw file):

Previously, kvark (Dzmitry Malyshau) wrote…

can we define it as a constant?

Done.

---

webrender_api/src/display_list.rs, line 812 at r1 (raw file):

Previously, kvark (Dzmitry Malyshau) wrote…

start at w?

Heh. This was supposed to be a '2'. Fixed. :)

---

webrender_api/src/display_list.rs, line 1316 at r1 (raw file):

Previously, kvark (Dzmitry Malyshau) wrote…

to clarify, we aren't checking for axis alignment here because an animated transform always spawns a reference frame, right?

That's correct. The spec dictates that even an identity transform creates a stacking context.

---

Comments from Reviewable

[kvark]

Reviewed 2 of 2 files at r2.
Review status: all files reviewed at latest revision, 1 unresolved discussion.

---

webrender_api/src/display_list.rs, line 814 at r2 (raw file):

        // We start at 2 here, because the root reference is always 0 and the root scroll
        // node is always 1.
        const FIRST_CLIP_ID: u64 = 2;

can we move it up next to the rest of the constants?

---

Comments from Reviewable

[mrobinson]

Review status: 5 of 6 files reviewed at latest revision, 1 unresolved discussion.

---

webrender_api/src/display_list.rs, line 814 at r2 (raw file):

Previously, kvark (Dzmitry Malyshau) wrote…

can we move it up next to the rest of the constants?

Done.

---

Comments from Reviewable

[kvark]

Reviewed 1 of 1 files at r3.
Review status: all files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable

[kvark]

@bors-servo r+

[bors-servo]

📌 Commit 1e030a2 has been approved by kvark

[bors-servo]

⌛ Testing commit 1e030a2 with merge 7184d6b...

[bors-servo]

☀️ Test successful - status-appveyor, status-taskcluster, status-travis
Approved by: kvark
Pushing 7184d6b to master...