editor: Smooth scrolling by marcocondrache · Pull Request #44827 · zed-industries/zed

marcocondrache · 2025-12-14T20:50:32Z

Closes #4355

Before:

zed-no-smooth.mp4

After:

zed.-.smooth.scroll.mp4

Ref: https://pavelfatin.com/scrolling-with-pleasure

Release Notes:

Editor now supports smooth scrolling

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

niekdomi · 2025-12-15T15:06:06Z

Thanks for this PR, feels smooth. Are there any plans to integrate this into vim-mode as well?

(e.g., ctr-y, ctrl-e, ctrl-b, ctrl-f, ctrl-d, ctrl-u, zz, zt, zb)

or is this not part of this PR?

marcocondrache · 2025-12-15T15:09:17Z

@niekdomi yep, I still need to make the changes. I'm currently perfecting the scroll animation

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

the set_scroll_position of the editor saves to the db the position and creates event which is wasteful for an animation Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

ttytm

Amazing effort!

Hope giving input already doesn't bother you.

Taking it through manual testing, I found that it doesn’t work with the optional preceding [count].

For example, my personal default for C+e / Ctrl+y is to scroll 5 lines. Currently it is reverted to 1 line when smooth scrolling is enabled.
No matter which scrolling motion is pressed [count] is ignored.

I wouldn’t want this to be a blocker, though. So a fix could also be part of a separate PR.

marcocondrache · 2025-12-16T10:03:27Z

@ttytm, thank you for the feedback. I'm primarily focusing on improving the animation, especially for trackpads. I'm waiting for some input from the team before I make significant changes and dive deep into the vim crate

niekdomi · 2026-03-01T10:50:37Z

there seems to be a bug/edge case: If I enter a negative value for the duration, zed crashes. I think there should be some kind of clamp

MrSubidubi

Sorry for the delay, as you know, was out sick and still a bit slow to pick up on some things.

Next round here, some nits although I'd argue we better are safe than sorry here, given that this reaches really deep. There is also the following issue, which we obviously need to figure out, where trackpad scrolling on my Mac is not acceptable with the native acceleration from the OS:

Bildschirmaufnahme.2026-03-09.um.22.24.54.mov

Which I think might very well be related to one of my comments.

Nonetheless, we are getting closer though and I'd say after this round we are getting to the topic of tests, so feel free to start thinking about them (not necessarily add them just yet, but you are obviously free to already start working on these). Arguably, I think we want to test stuff amongst the lines of

expected vs. calculated scroll positions on a per-frame basis.
change in direction during animation
instant scroll request during scroll animation
animation scroll request during animation scroll request
animated scroll after animated scroll to ensure state is in sync (can perhaps be part of the first test
settings update test during animation - perhaps we might not need this, but what I'd like to test here is that we do not end up in a state we cannot escape

and perhaps some other nice cases you can think about. As said, food for thought, I think it might be easier to tacke these after this batch here.

MrSubidubi · 2026-03-09T20:12:51Z

+pub enum ScrollAnimationPhase {
+    Intermediate,
+    Final,
+}
+
+#[derive(Clone, Copy, Debug)]
+pub struct ScrollAnimation {


I think the animation phase should be pub(crate) at best, the state itself should perhaps even only be visible within this file even?

The animation must be at least pub(crate) since it's used in the EditorSnapshot. Should be fixed

MrSubidubi · 2026-03-09T20:14:11Z

-                }
-            }
-        };
+        let (anchor, top_row) = self.calculate_scroll_anchor(scroll_position, map, cx);


Let's revert the change here since calculate_scroll_anchor is only called once now.

Super right, should be done

MrSubidubi · 2026-03-09T20:18:28Z

+    // Whether to animate scrolling with a smooth easing effect.
+    "enabled": false,
+    // Duration of the animation
+    "duration": 0.125,


I am not entirely sure we should make the duration configurable at all - I am sold on having the behavior change based on the input type used, and feel like making the duration configurable might be better to do at a later step if at all. This feels like a case where good defaults might just be more worth than to make this configurable in the first place. The fact that VSCode does not have a setting for this makes me feel like we might not need it as well or at least not in the first place.

That makes sense. The main reason I added it initially was that scrolling felt much slower when smooth scrolling was enabled. Later I discovered that increasing the sensitivity fixes that, so it definitely makes sense to remove it.

Should be done 😊

MrSubidubi · 2026-03-09T20:37:48Z

        self.scroll_anchor.scroll_position(&self.display_snapshot)
    }

+    pub fn scroll_target_or_position(&self) -> gpui::Point<ScrollOffset> {


Can we not just have scroll_position, do we really need to distinguish these here?

We don't need to distinguish them anymore, should be done

MrSubidubi · 2026-03-09T20:38:16Z

    pub placeholder_display_snapshot: Option<DisplaySnapshot>,
    is_focused: bool,
    scroll_anchor: SharedScrollAnchor,
+    pub scroll_animation: Option<ScrollAnimation>,


Is there a good reason for this to be pub?

Should be done

MrSubidubi · 2026-03-09T20:57:32Z

+        if self.is_finished() {
+            self.phase = ScrollAnimationPhase::Final;
+            self.position = self.target_position;
+        } else {
+            let progress = self.progress();


We calculate the progress twice here in rapid succession - Perhaps we can have progress return a newtype ScrollAnimationProgress(f32) that then has the method is_finished on that so we avoid doing this twice?

Nice idea, I have created ScrollAnimationProgress

MrSubidubi · 2026-03-09T21:02:47Z

+impl ScrollAnimation {
+    pub fn instant(target: gpui::Point<ScrollOffset>) -> Self {
+        Self {
+            phase: ScrollAnimationPhase::Final,
+            duration: Duration::ZERO,
+            position: target,
+            start_position: target,
+            target_position: target,
+            start_time: Instant::now(),
+        }
+    }


Looking at this and the surrounding methods, I can't help but wonder whether we should manage the state here using an enum as opposed to the struct - that way, we can get rid of plenty of the duplication here and can avoid some invalid states a bit better.
This might be more verbose in parts, but might also help more with what the current state is? Since ScrollAnimationPhase::Final basically implies position == target_position for example. Would like your view on this obviously

That’s a really nice idea, not sure why it didn’t come to my mind earlier. Looks a lot better now

MrSubidubi · 2026-03-09T21:07:07Z

+    let snapshot = editor.snapshot(window, cx);
+    let scroll_position = snapshot.scroll_target_or_position();
+    let old_top_row = DisplayRow(scroll_position.y as u32);
+    let old_top_column = scroll_position.x as u32;


This is much better already, thank you.

We should make scroll_manager private as part of this then though - if these methods can no longer be used freely due to these changes, we should ensure that no one can actually use them (and again (sorry), ensure that visibility of methods matches what we enforce here). I know our policies on comments, but for this case, might even be worth to have a comment within editor to ensure peole do not easily violate that.

That's right. I checked where scroll_manager is accessed, and it looks like the agent_ui crate is still accessing it directly. Should we address that in this PR?

I think to save you on conflicts, I'd be best if we could move this out into another PR. I'll get that one in then ASAP

MrSubidubi · 2026-03-09T21:29:35Z

+        } else {
+            ScrollAnimationPhase::Intermediate
+        }
+    }


I think one issue we might face here is when we invoke this method, we reset the current acceleration, which I am not sure is what we want - you lose the current scroll momentum as we start the easing method from the very beginning again, which means we might be in the mid of easing - and then are at the very start again. Instead, I think we should consider the current acceleration/speed at which we are animating and try to find a suitable progress at which we can continue the animation/perhaps change the duration slightly. So this should not be a hard restart I believe but instead more calculate a new animation based on the current state.

That’s also why increasing the sensitivity makes such a big difference. I’ve added an initial implementation that computes the velocity of the current animation using the derivative of the easing function, and then adjusts the duration of the next animation accordingly. The approach is inspired by Chromium’s smooth scrolling implementation (ref 1, ref 2).

The results are already much better than the previous approach, but I think with some additional tuning and research we can improve it even further.

MrSubidubi · 2026-03-09T21:30:15Z

Sign in to view

+
+                        let base_scroll_position = if is_precise {
+                            editor.scroll_manager.cancel_animation();
+                            current_scroll_position
+                        } else {
+                            position_map
+                                .snapshot
+                                .scroll_animation
+                                .map(|animation| animation.target_position)
+                                .unwrap_or(current_scroll_position)
+                        };


Could you briefly elaborate on the logic here?

@MrSubidubi I've removed this logic it's not necessary anymore. It was an attempt to understand the scrolling slowness difference between mouse and trackpad.

seanstrom · 2026-04-20T23:25:32Z

@marcocondrache do you have some time to rebase this branch and resolve the conflicts?

MrSubidubi

I agree this is at a good point to resolve conflicts - the PR is in very good shape now and I'd love to see this progress further. Currently, at a high level, see only three things left:

the comment below
tests as mentioned in the prior round
how we want to handle trackpad scrolls - if we can, I am thinking about whether it would make sense to distinguish between already smoothed scrolls and non-smoothed scrolls on the platform level, or do you have a better idea for that? Would love your opinion on this

MrSubidubi · 2026-04-20T23:46:47Z

+    let snapshot = editor.snapshot(window, cx);
+    let scroll_position = snapshot.scroll_target_or_position();
+    let old_top_row = DisplayRow(scroll_position.y as u32);
+    let old_top_column = scroll_position.x as u32;


I think to save you on conflicts, I'd be best if we could move this out into another PR. I'll get that one in then ASAP

marcocondrache · 2026-04-21T07:00:38Z

@MrSubidubi @seanstrom I've resolved the conflicts and merged main.

For everyone following the status of this PR: I’ve recently been pulled into leading a project at my current job, so I have very little, if any, time to dedicate to open source right now. Since this PR is nearly ready and it’s something the community really values, I’m aiming to get this closed out soon.

Apologies for this becoming a bit stale and for the lack of progress recently. I appreciate everyone’s patience

marcocondrache added 4 commits December 14, 2025 20:39

add setting

576c72c

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

wip

408989c

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

move easing function

0afdc8e

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

use easing fn

a647739

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

cla-bot Bot added the cla-signed The user has signed the Contributor License Agreement label Dec 14, 2025

github-actions Bot added the community champion Issues filed by our amazing community champions! 🫶 label Dec 14, 2025

github-project-automation Bot added this to Quality Week – December 2025 Dec 14, 2025

github-project-automation Bot moved this to Community Champion PRs in Quality Week – December 2025 Dec 14, 2025

marcocondrache added 4 commits December 14, 2025 22:00

make duration configurable

2322376

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

forgot this

8efc7f5

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

move duration to settings

a905674

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

140ms is the optimal default

c7c14f8

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

MrSubidubi self-assigned this Dec 15, 2025

marcocondrache added 2 commits December 15, 2025 15:49

delegate to manager

f6abc30

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

revert fmt

5b8c5b1

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

marcocondrache added 4 commits December 15, 2025 16:24

vim: this was bypassing animation

005b280

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

vim: make smooth scrolling work

22e01ee

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

comment

ce5c212

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

avoid sync to db and events

579f8ef

the set_scroll_position of the editor saves to the db the position and creates event which is wasteful for an animation Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

marcocondrache commented Dec 15, 2025

View reviewed changes

Comment thread crates/editor/src/element.rs Outdated

marcocondrache added 2 commits December 15, 2025 20:55

update default

1e88aa1

Signed-off-by: Marco Mihai Condrache <52580954+marcocondrache@users.noreply.github.com>

Merge branch 'main' into feat/smooth-scrolling

bcc7db0

ttytm reviewed Dec 16, 2025

View reviewed changes

marcocondrache marked this pull request as ready for review December 16, 2025 13:43

slavpetroff added a commit to slavpetroff/zed that referenced this pull request Dec 18, 2025

Merge PR zed-industries#44827: smooth scrolling improvements

7746fbc

marcocondrache force-pushed the feat/smooth-scrolling branch from 62fbef4 to c929736 Compare December 18, 2025 12:23

marcocondrache mentioned this pull request Dec 20, 2025

Implement kinetic / inertial scrolling for Linux (Wayland) #14473

Open

1 task

marcocondrache added 3 commits February 19, 2026 21:11

Merge branch 'main' into feat/smooth-scrolling

037ef66

fix: new tests

2c06a7d

fix: scroll screen test

642d169

marcocondrache added 4 commits March 1, 2026 12:15

Merge branch 'main' into feat/smooth-scrolling

3c15aa7

fix negative duration

efd3b7a

Merge branch 'main' into feat/smooth-scrolling

3e8fd59

Merge remote-tracking branch 'upstream/main' into feat/smooth-scrolling

bfe9e4f

MrSubidubi requested changes Mar 9, 2026

View reviewed changes

Merge branch 'main' into feat/smooth-scrolling

7355d20

marcocondrache force-pushed the feat/smooth-scrolling branch 7 times, most recently from 00be4d6 to 49f54fc Compare March 10, 2026 11:00

second iteration

0d8fd32

marcocondrache force-pushed the feat/smooth-scrolling branch from 49f54fc to 0d8fd32 Compare March 10, 2026 12:19

Merge branch 'main' into feat/smooth-scrolling

97119a7

marcocondrache force-pushed the feat/smooth-scrolling branch from 1a8fa0f to 97119a7 Compare March 12, 2026 09:06

This comment was marked as spam.

Sign in to view

preserve scroll velocity

b2d2f6a

marcocondrache force-pushed the feat/smooth-scrolling branch from b9928b1 to b2d2f6a Compare March 14, 2026 17:09

MrSubidubi requested changes Apr 20, 2026

View reviewed changes

marcocondrache force-pushed the feat/smooth-scrolling branch from 3126841 to e9bd4e5 Compare April 21, 2026 06:50

Merge branch 'main' into feat/smooth-scrolling

3addcc6

marcocondrache force-pushed the feat/smooth-scrolling branch from e9bd4e5 to 3addcc6 Compare April 21, 2026 07:41

Uh oh!

Conversation

marcocondrache commented Dec 14, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

niekdomi commented Dec 15, 2025

Uh oh!

marcocondrache commented Dec 15, 2025

Uh oh!

Uh oh!

ttytm left a comment

Choose a reason for hiding this comment

Uh oh!

marcocondrache commented Dec 16, 2025

Uh oh!

niekdomi commented Mar 1, 2026

Uh oh!

MrSubidubi left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

marcocondrache Mar 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

marcocondrache Mar 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This comment was marked as spam.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This comment was marked as spam.

seanstrom commented Apr 20, 2026

Uh oh!

MrSubidubi left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

marcocondrache commented Apr 21, 2026

Uh oh!

Reviewers

Assignees

marcocondrache commented Dec 14, 2025 •

edited

Loading

marcocondrache Mar 10, 2026 •

edited

Loading

marcocondrache Mar 10, 2026 •

edited

Loading