Add command right click functionality for window resizing like in i3wm by tymscar · Pull Request #1618 · nikitabobko/AeroSpace

tymscar · 2025-08-07T22:12:39Z

This PR adds the right click drag feature from i3wm.
The user should be able to command right-click and drag on any window, and it will resize in the correct direction. If you drag to the left from the left side, you make it bigger, for example.

I have gotten it quite smooth, and as far as I can tell, it actually circumvents the issue that arises when you manually resize a window over another one, and they switch spaces. I can't replicate that with this method. This method runs at the refresh rate of your fastest monitor, or if that fails, I default it to 60fps.

I have also made it so when you hold Command, it doesn't pass through the right-click to the window under it. That was annoying because while dragging, I would by mistake click on Back or something.

Fixes issue #206

tymscar · 2025-09-26T23:46:11Z

Hello!

Sorry for bothering @nikitabobko, I just wanted to know if theres anything else needed from me for this PR?

nikitabobko

Thanks for the PR. I do want to implement this feature at some point.

I was thinking of more generic mouse events where users could assign command for whatever mouse + keyboard events (as described here: #371), but we can start small and implement the mouse dragging/moving with the modifier the way you propose it in this PR, and the way it's done in i3

I have left my first iteration review comments. The most major problem is logic duplication. Please share the logic with existing func resizeWithMouse. Once/If you fix the comments, we can discuss the PR further

nikitabobko · 2025-10-02T09:46:38Z

Sources/AppBundle/GlobalObserver.swift

+                (1 << CGEventType.mouseMoved.rawValue) |
+                (1 << CGEventType.flagsChanged.rawValue),
+        )
+        if cmdRightTap == nil,


GlobalObserver.initObserver is called only once at startup. Why do you need cmdRightTap == nil check?

Sorry about that. I usually try to code more defensively. I didn't look to deep into how we reload the config, so I was thinking the whole thing might get reinitialised.

nikitabobko · 2025-10-02T09:50:14Z

Sources/AppBundle/mouse/resizeCmdRightDrag.swift

+@MainActor
+private func scheduleThrottledRefresh() {
+    if pendingDragRefreshTask != nil { return }
+    pendingDragRefreshTask = Task { @MainActor in
+        try? await Task.sleep(for: preferredFrameDuration())
+        runRefreshSession(.globalObserver("cmdRightMouseDragged"), optimisticallyPreLayoutWorkspaces: true)
+        pendingDragRefreshTask = nil
+    }
+}
+
+@MainActor
+private func preferredFrameDuration() -> Duration {
+    let maxFps = NSScreen.screens.map { $0.maximumFramesPerSecond }.max() ?? 60
+    let fps = max(maxFps, 1)
+    let nanosPerFrame = 1_000_000_000 / fps
+    return .nanoseconds(nanosPerFrame)
+}


Please take a look at how throttling is implemented in func movedObs and do the same.

I think there is no need for Task.sleep, and there is no need for complicated frames math

Oh yeah, that's so much cleaner!

nikitabobko · 2025-10-02T09:53:07Z

Sources/AppBundle/mouse/resizeCmdRightDrag.swift

+func onCmdRightMouseDragged() async {
+    guard let session = cmdRightResizeSession else { return }
+    guard let window = Window.get(byId: session.windowId) else { return }
+
+    let point = mouseLocation
+    let direction = edgeToDirection(session.edge)
+    let delta: CGFloat = (direction.orientation == .h) ? (point.x - session.startPoint.x) : (point.y - session.startPoint.y)
+    let diff: CGFloat = direction.isPositive ? delta : -delta
+
+    guard let (parent, _, neighborIndex, orientation) = resolveParentAndNeighbor(window, direction) else { return }
+    if abs(diff) < 1 { return }
+
+    window.parentsWithSelf.lazy
+        .prefix(while: { $0 !== parent })
+        .compactMap { node -> TreeNode? in
+            let p = node.parent as? TilingContainer
+            return (p?.orientation == orientation && p?.layout == .tiles) ? node : nil
+        }
+        .forEach { $0.setWeight(orientation, $0.getWeightBeforeResize(orientation) + diff) }
+
+    let sibling = parent.children[neighborIndex]
+    sibling.setWeight(orientation, sibling.getWeightBeforeResize(orientation) - diff)
+
+    currentlyManipulatedWithMouseWindowId = window.windowId


I read this code diagonally, so maybe I'm missing something, but my intuition says that the existing func resizeWithMouse can be reused

You might need to refactor func resizeWithMouse to make it more generic to handle both cases, but I don't want these obviously related logics being implemented two times in two different ways

Your custom logic doesn't handle more complicated cases:

Screen.Recording.2025-10-02.at.13.47.28.mov

nikitabobko · 2025-10-02T09:56:08Z

Sources/AppBundle/GlobalObserver.swift

+               callback: { _, type, event, _ in
+                   if !TrayMenuModel.shared.isEnabled { return Unmanaged.passUnretained(event) }
+                   let flags = event.flags
+                   let isCmd = flags.contains(.maskCommand)


The modifier must be configurable in the config

Smth like:

mouse-resize-modifier = 'alt' # Possible values: cmd, alt, ctrl, shift

AeroSpace's default MOD key is alt, so that should be the default for resizing as well

That makes sense. I personally always prefer command for this sort of things. Thanks for bringing this up!

nikitabobko · 2025-10-02T09:56:53Z

Sources/AppBundle/mouse/resizeCmdRightDrag.swift

+import Common
+
+
+private enum ResizeEdge { case left, right, up, down }


Can't CardinalDirection just be reused?

tymscar · 2025-10-03T00:52:27Z

Thanks so much for the review @nikitabobko !

So I spent quite a while trying all sorts of things out and I think I got something that reduces the duplication as much as I can.

You might know this better, but as far as I can tell, I think the way we resize with the mouse normally and how right-click drag should work is totally different.

In the case of the mouse resize, that's just normal macOS resizing, and then the accessibility framework updates us on what happened so we can sync up the weights. Whereas for the right-click drag one, that's fully custom, which means we have to handle the actual resizing logic. So what I ended up doing was I factored out the weights calculation and cleaned up the logic a bit, and now both methods use that, but I am not married to the idea. I am all ears if you think there's something cleaner.

Thanks for giving me the opportunity to learn more!

…ify drag refresh handling

…nd simplifying CmdRightResizeSession

tymscar · 2025-11-30T17:49:48Z

Hey @nikitabobko!

I have been using this fork now for four months daily, and I find it super helpful. So I spent some time fixing merge conflicts, rebased onto main, as well as fixed the issue with resizing diagonals you have told me about using that GIF.

…er compliance

jthomaschewski · 2025-11-30T23:18:10Z

@tymscar Thanks a lot for the PR! I’ve been running your branch for a week with no major issues.

Just one thing: Would it be possible to add support for resizing floating windows as well?
Currently it appears to work only for windows within split containers

tymscar · 2025-12-01T00:22:25Z

Thank you @jthomaschewski !

I don't really use floating windows, and I will be busy starting from tomorrow until middle of December, but I might take a look at it after that!

# Conflicts: # Sources/AppBundle/GlobalObserver.swift # Sources/AppBundle/config/Config.swift # Sources/AppBundle/config/parseConfig.swift # Sources/AppBundle/mouse/mouse.swift

Phase 4 tab detection fixes (5 blocking, 7 non-blocking): - Fix index out-of-bounds crash on tab promotion (clamp saved index) - Fix tryOnWindowDetected skipped for slot-restored tabs - Fix suspendedWindowSlots cross-reference leak in tryRestoreTabSlot - Fix getWindowLevel mid-cycle cache rebuild causing state inconsistency - Fix floating-point group keys (round bounds to integers) - Add O(1) reverse lookups for isBackgroundTab and tabGroupKey - Replace force-cast with guard-let in validateStillPopups - Make TabGroup struct private, add access-level documentation - Add explanatory comments for layoutReason and false-positive risks Phase 1-3 pre-existing bug fixes: - Fix dead code in GlobalObserver flagsChanged handler (PR nikitabobko#1618) The guard-isModifier prevented modifier-release events from reaching the switch, so releasing the resize modifier mid-drag never terminated the resize session. Moved flagsChanged before the guard. - Fix swapWindows using window1.ownIndex instead of window2.ownIndex in moveWithMouse.swift

nikitabobko force-pushed the main branch from 116f6a2 to 27140a6 Compare August 17, 2025 20:40

nikitabobko requested changes Oct 2, 2025

View reviewed changes

nikitabobko force-pushed the main branch 2 times, most recently from f324d2e to 7aff813 Compare November 24, 2025 01:31

tymscar added 5 commits November 30, 2025 14:24

Add command right click functionality for window resizing like in i3wm

9cdea66

Fix lint errors

2ee36e4

Add configurable mouse resize modifier for right-click window resizing

e8da454

Refactor CmdRightResizeSession to use CardinalDirection and simpl…

376e057

…ify drag refresh handling

Refactor mouse resizing logic by extracting weight adjustment logic a…

8f31af7

…nd simplifying CmdRightResizeSession

tymscar force-pushed the add-right-click-resize branch from f9cb42a to 8f31af7 Compare November 30, 2025 17:12

Fix cmd-right drag to schedule refresh for correct session updates

98358ff

tymscar force-pushed the add-right-click-resize branch from 33d72d5 to 98358ff Compare November 30, 2025 17:14

Fix right click drag resize not working with diagonals

df97f49

tymscar force-pushed the add-right-click-resize branch from f0bfc64 to df97f49 Compare November 30, 2025 17:52

Reorder modifiers (nonisolated private) in GlobalObserver for formatt…

2e5f763

…er compliance

tymscar requested a review from nikitabobko November 30, 2025 18:06

		import Common


		private enum ResizeEdge { case left, right, up, down }

Uh oh!

Conversation

tymscar commented Aug 7, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tymscar commented Sep 26, 2025

Uh oh!

nikitabobko left a comment

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!

tymscar commented Oct 3, 2025

Uh oh!

tymscar commented Nov 30, 2025

Uh oh!

jthomaschewski commented Nov 30, 2025

Uh oh!

tymscar commented Dec 1, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

tymscar commented Aug 7, 2025 •

edited

Loading