Adds geometry dirty nodes by chunhtai · Pull Request #180375 · flutter/flutter

chunhtai · 2025-12-29T23:50:52Z

Highlevel:

instead of clearing the entire geometry from the node that markSemanticsNeedsUpdate up to the first semantics boundary. and recalculate everything in between. This code now only clear the geometry for node that are markSemanticsNeedsUpdate. This requires us to keep a separate list for geomtry dirty node. Thus this change.

Pre-launch Checklist

I read the Contributor Guide and followed the process outlined there for submitting PRs.
I read the Tree Hygiene wiki page, which explains my responsibilities.
I read and followed the Flutter Style Guide, including Features we expect every widget to implement.
I signed the CLA.
I listed at least one issue that this PR fixes in the description above.
I updated/added relevant documentation (doc comments with ///).
I added new tests to check the change I am making, or this PR is test-exempt.
I followed the breaking change policy and added Data Driven Fixes where supported.
All existing and new tests are passing.

If you need help, consider asking for advice on the #hackers-new channel on Discord.

Note: The Flutter team is currently trialing the use of Gemini Code Assist for GitHub. Comments from the gemini-code-assist bot should not be taken as authoritative feedback from the Flutter team. If you find its comments useful you can update your code accordingly, but if you are unsure or disagree with the feedback, please feel free to wait for a Flutter team member's review for guidance on which automated comments should be addressed.

chunhtai · 2026-01-13T23:37:11Z

before

after

YEEEEEESSSSSSSS, cut half of total semantics calculation from the framework side

gemini-code-assist

Code Review

This pull request introduces a more granular mechanism for tracking semantics geometry updates by adding a _nodesNeedingSemanticsGeometryUpdate set to PipelineOwner. This should improve performance by avoiding unnecessary geometry recalculations. The changes are well-contained and include necessary test updates. I have one suggestion to improve the readability of a complex loop.

justinmc

Makes sense to me at a high level, but @LongCatIsLooong should definitely review and follow up on his old ideas (#166173 (comment)). Otherwise LGTM 👍

LongCatIsLooong

Hmm could you add a comment somewhere to explain when a RenderObject gets added to _nodesNeedingSemanticsGeometryUpdate? For a RenderObject author what should I do to make sure the semantics geometry is updated, when I skip layout and only do repaint (I think you can safely do that to re-align a child today since the child subtree's layout does not change at all). From the looks of it the RenderObject author must call markNeedsSemanticsUpdate when they re-align the child?

LongCatIsLooong · 2026-01-21T21:09:53Z

+                (RenderObject object) =>
+                    !object._needsLayout &&
+                    object.owner == this &&
+                    !object._semantics.parentDataDirty,


Is this for skipping dirty RenderObjectSemantics that will need geometry updates anyways because the parent data also needs updating? I'm thinking maybe instead of boolean flags we should be using an int/enum for indicating the dirtiness?

so it is slightly complicated, if at this point the parent data is still dirty, it is a blocked branch. so parentDataDirty doesn't imply it is blocked, it depends on the phase of update as well. I will add some more comment

for a RenderObjectSemantics to find out if it is in a blocked branch, does it only have to check its own parentDataDirty or it has to check all the way up to the root node? I assume its the former based on the implementation here, so even for a "blocked" branch we still recursively update parentDataDirty for all descendant nodes that are going to be blocked?

does it only have to check its own parentDataDirty

yes

It is a delicate condition that makes this true, the node in the list should have their parentdata clear in _RenderObjectSemantics.markNeedsUpdate is called. if it is still dirty here, it means the updateChild hasn't update the the parentdata of this node, and that can only happen if it is blocked.

This is controlled by this line in _RenderObjectSemantics.markNeedsUpdate

if (node != renderObject && producedSemanticsNode != null && node._semantics.parentDataDirty) { // If `this` node has already been added to [owner._nodesNeedingSemantics] // remove it as it is no longer guaranteed that its semantics // node will continue to be in the tree. If it still is in the tree, the // ancestor `node` added to [owner._nodesNeedingSemantics] at the end of // this block will ensure that the semantics of `this` node actually gets // updated. // (See semantics_10_test.dart for an example why this is required). renderObject.owner!._nodesNeedingSemantics.remove(renderObject); }

This is where the block node be excluded from updateChildren if it is blocked (if it is not blocked, the parent of this semantics boundary will call updateChildren on this subtree)

so even for a "blocked" branch we still recursively update parentDataDirty for all descendant nodes that are going to be blocked?

For a node in a blocked branch that got added to this list, its parentData must be dirty at this point due to the reason above, so the situation you mentioned should never happen

Thanks for the explanation! Interestingly when I comment out that if block all tests still pass (I had to use --update-goldens tho since golden tests never pass without that on my macbook).

I think some of the google tests may fail, that is why I know about this blocked node shenanigan.

In the long term we should remove the block node logic as it cause state loss for live region which is not good. There is currently an effort to remove blockSemantics usage and hopefully deprecates it.

chunhtai · 2026-01-21T22:59:17Z

Added doc comment on _nodesNeedingSemanticsGeometryUpdate not sure if that is the best place, let me know if you think otherwise.

From the looks of it the RenderObject author must call markNeedsSemanticsUpdate when they re-align the child?

This is already the case though, I am not changing the behavior.

LongCatIsLooong

So from reading this my observations are:

Every RenderObject will be added to _nodesNeedingSemanticsGeometryUpdate when it's markedNeedsSemanticsUpdate. But at that point we can't skip adding them to the set because the semantics tree structure has not been updated and we do not know which of them will form semantics nodes and who their parents will be after updateChildren.

For that reason, it is after updateChildren, that we traverse the _nodesNeedingSemanticsGeometryUpdate to figure out which branches need geometry updates and reset geometry to null to evict the geometry cache and to indicate that these branches will need to be processed for geometry update.

After that step we "Conduct the geometry update". This is done by locating geometry update roots, from the nodes in _nodesNeedingSemanticsGeometryUpdate, and calling ensureGeometry on these "root" nodes. ensureGeometry walks down the tree and if one of its immediate child still has geometry cache it would skip that branch.

(Reminder to myself that I still have a question regarding sibling merge groups).

LongCatIsLooong · 2026-01-28T21:05:14Z

+        // In both cases, the renderobject will be added to the list of nodes needing
+        // geometry update.
+        //
+        // This specific casecan happen if a renderobject decided to conditionally include


nit: missing space between "case" and "can". By this sentence do you mean that the visitChildrenForSemantics of that RenderObject having a branching condition which doesn't call markNeedsSemanticsUpdate when changed? Is there an example?

Since geometry is only reset to null after updateChildren in the ensureGeometry step (barring clear), so this is for a brand new subtree that has never been incorporated into the tree?

I think this may be an old code that no longer needed, that me remove it and see

ok turns out this is a corner case

I added the code comment explain the situation, I will also paste it here

// This specific case can happen if a parent somwhere above this node decided to // conditionally include this node. // // For example, render object A->B->C, where A is the root and all of them // creates semantics nodes. // // Frame 1: B and C both mark layout dirty, and B suddenly decide not to // inlcude C in the semantics tree. // // In this case, C's geometry will remain dirty because it is considered as // blocked. // // Frame 2: B is marked dirty again and decide C should be included // in the semantics tree. We will run into this situation where C's geometry // is dirty but it is not in the _nodesNeedingSemanticsGeometryUpdate.

LongCatIsLooong · 2026-01-28T21:23:28Z

+        bool isBlocked = target.parentDataDirty;
+        // Find the first render object semantics that forms a semantics node
+        // and whose geometry is not dirty as the anchor point to start the geometry update.
+        while (!isBlocked &&


Would it be easier /faster if each RenderObjectSemantics knows who the parent RenderObjectSemantics that formed the semantics boundary is?

I think we could, but since this is easy enough and efficient enough to find it on the spot, I try avoid adding more cache to this class unless necessary. That will probably create some corner cases.

LongCatIsLooong · 2026-01-28T21:27:03Z

+                (RenderObject object) =>
+                    !object._needsLayout &&
+                    object.owner == this &&
+                    !object._semantics.parentDataDirty,


for a RenderObjectSemantics to find out if it is in a blocked branch, does it only have to check its own parentDataDirty or it has to check all the way up to the root node? I assume its the former based on the implementation here, so even for a "blocked" branch we still recursively update parentDataDirty for all descendant nodes that are going to be blocked?

chunhtai · 2026-01-29T23:48:55Z

So from reading this my observations are:

Every RenderObject will be added to _nodesNeedingSemanticsGeometryUpdate when it's markedNeedsSemanticsUpdate. But at that point we can't skip adding them to the set because the semantics tree structure has not been updated and we do not know which of them will form semantics nodes and who their parents will be after updateChildren.

For that reason, it is after updateChildren, that we traverse the _nodesNeedingSemanticsGeometryUpdate to figure out which branches need geometry updates and reset geometry to null to evict the geometry cache and to indicate that these branches will need to be processed for geometry update.

After that step we "Conduct the geometry update". This is done by locating geometry update roots, from the nodes in _nodesNeedingSemanticsGeometryUpdate, and calling ensureGeometry on these "root" nodes. ensureGeometry walks down the tree and if one of its immediate child still has geometry cache it would skip that branch.

(Reminder to myself that I still have a question regarding sibling merge groups).

Yes the entire statement is accurate

LongCatIsLooong · 2026-02-09T22:00:27Z

+                (RenderObject object) =>
+                    !object._needsLayout &&
+                    object.owner == this &&
+                    !object._semantics.parentDataDirty,


Thanks for the explanation! Interestingly when I comment out that if block all tests still pass (I had to use --update-goldens tho since golden tests never pass without that on my macbook).

LongCatIsLooong · 2026-02-09T22:09:10Z

+          continue;
+        }
+        for (final _RenderObjectSemantics child in node._semantics._children) {
+          child.geometry = null;


Also why only removing the geometry cache on child nodes? In this case if node also happens to be a relayout boundary that changed its size (but its parent wasn't notified because it was a relayout boundary), if we only remove geometry from node's children, then in the "conduct the geometry update" step, ensureGeometry will be called on node but not on its semantics parent? But in this case I think it should be called on the semantics parent because node's geometry will need updating right?

in this case I think it should be called on the semantics parent because node's geometry will need updating right?

this is already the case. the target after the while loop must either be a root node, or shouldformsemanticsnode while geometry is not dirty

ensureGeometry() is called on the immediate parent, that form a semantics node, of the node in _nodesNeedingSemanticsGeometryUpdate.

ensureGeometry() is called on the immediate parent, that form a semantics node

for (final node in nodesToProcessGeometry) { _RenderObjectSemantics target = node._semantics; bool isBlocked = target.parentDataDirty; while (!isBlocked &&Collapse !target.isRoot && (target.geometryDirty || !target.shouldFormSemanticsNode)) { target = target.parent!; isBlocked = target.parentDataDirty; } if (!isBlocked) { target.ensureGeometry(); } }

if node.shouldFormSemanticsNode is true, we won't go into the while loop at all I think? In that case the final target will be node instead of node.parent? But if node's sized changed, it should be node.parent.ensureGeometry() instead of node.ensureGeometry?

interesting, I think you are right that this is a corner case. IIRC before the whole renderobjectsemantics refactoring, it is already like this that only the parent semantics node would ever update the geometry of the current node. I guess we just never hit this corner case.

let me look into this and get back to you

fixed the corner case by clear the geometry cache if the node shouldformSemanticsnode= true.
also added tests in
semantics_layout_sized_test.dart

LongCatIsLooong · 2026-02-09T22:28:43Z

+              // represents a group of render objects in the subtree, where node that
+              // forms a semantics node is in its _children.
+              for (final _RenderObjectSemantics nodeInSubtree in child._children) {
+                assert(nodeInSubtree.shouldFormSemanticsNode);


hmm interesting, so the merge-up tree is guaranteed to have a max depth of 2?

all the immediate semantics node are cached for non-semantics-node-forming renderbojectsemantics until it reach the a renderbojectsemantics that will form a semantics node.

It is this way to avoid walking the intermediate nodes to collect all renderbojectsemantics that will form semantics node

LongCatIsLooong · 2026-02-09T23:15:02Z

          continue;
        }
-        node._semantics.ensureGeometry();
+        if (!node._semantics.contributesToSemanticsTree) {


So here contributesToSemanticsTree is checked instead of shouldFormSemanticsNode because _children is only updated in updateChildren if contributesToSemanticsTree is true? And that's less children that needs to be marked dirty for geometry?

LongCatIsLooong

LGTM with questions regarding the RelayoutBoundary + shouldFormSemanticsNode case (see #180375 (comment)).

Roll Flutter from 2ec61af30672 to 195ae7b3a122 (36 revisions) flutter/flutter@2ec61af...195ae7b 2026-03-10 [email protected] [web] Updates to the README (flutter/flutter#176292) 2026-03-10 [email protected] [Impeller] Do not log VK_ERROR_SURFACE_LOST_KHR errors returned by vkAcquireNextImageKHR (flutter/flutter#183338) 2026-03-10 [email protected] DlDeferredImageGPUImpeller::ImageWrapper texture thread safety improvements (flutter/flutter#183429) 2026-03-10 [email protected] refactor: remove material imports from system_context_menu_test, table_test (flutter/flutter#183368) 2026-03-10 [email protected] refactor: remove material from baseline_test, slivers_test (flutter/flutter#183367) 2026-03-10 [email protected] Migrated circle sdf to euclidean 2d derivative (flutter/flutter#183184) 2026-03-10 [email protected] Roll Packages from c717018 to ee460d6 (2 revisions) (flutter/flutter#183451) 2026-03-10 [email protected] Roll Skia from 68f24903b592 to 4b35832cc7ea (1 revision) (flutter/flutter#183449) 2026-03-10 [email protected] Roll Skia from 8eda160375ea to 68f24903b592 (1 revision) (flutter/flutter#183444) 2026-03-10 [email protected] Roll Skia from e7b957806202 to 8eda160375ea (2 revisions) (flutter/flutter#183440) 2026-03-10 [email protected] Roll Fuchsia Linux SDK from ox-II3wPpa818y78_... to 8C_qfgWgoNhkV0_Mn... (flutter/flutter#183437) 2026-03-10 [email protected] Roll Skia from 887df915a95e to e7b957806202 (1 revision) (flutter/flutter#183434) 2026-03-10 [email protected] Remove material from scrollable_test.dart (flutter/flutter#181429) 2026-03-09 [email protected] Update 'a: text input' globs (flutter/flutter#183405) 2026-03-09 [email protected] Roll Skia from 3aa442e0624a to 887df915a95e (7 revisions) (flutter/flutter#183423) 2026-03-09 [email protected] Adds geometry dirty nodes (flutter/flutter#180375) 2026-03-09 [email protected] Roll Dart SDK from 5b1aa4f8db14 to ebef6c849489 (1 revision) (flutter/flutter#183417) 2026-03-09 [email protected] Add Alexander Dmitriev to AUTHORS (flutter/flutter#183357) 2026-03-09 [email protected] Update org triage (flutter/flutter#183254) 2026-03-09 [email protected] [Android] Avoid plugin auto-registration during FlutterFragmentActivity recreate timing window (flutter/flutter#182963) 2026-03-09 [email protected] Standardize skia includes. (flutter/flutter#183404) 2026-03-09 [email protected] Add a note to `arcTo` docs about the 2PI drawing angle issue (flutter/flutter#183293) 2026-03-09 [email protected] Roll pub packages (flutter/flutter#183411) 2026-03-09 [email protected] [ Tool ] Fix --print-dtd not working for web targets (flutter/flutter#183171) 2026-03-09 [email protected] Add await or ignore lint to flutter_driver unawaited callsites (flutter/flutter#183334) 2026-03-09 [email protected] Cleanup MockVulkan when the vulkan instance is destroyed (flutter/flutter#183324) 2026-03-09 [email protected] Updates tests commit (flutter/flutter#183301) 2026-03-09 [email protected] Implements `ClipRSuperellipse` on backdrop filter on platform view (flutter/flutter#182643) 2026-03-09 [email protected] Roll Skia from ea4fd0e1472a to 3aa442e0624a (1 revision) (flutter/flutter#183397) 2026-03-09 [email protected] Roll Dart SDK from 050b2f646425 to 5b1aa4f8db14 (1 revision) (flutter/flutter#183393) 2026-03-09 [email protected] Roll Packages from fe3de64 to c717018 (10 revisions) (flutter/flutter#183396) 2026-03-09 [email protected] Roll Skia from a83dca75c531 to ea4fd0e1472a (1 revision) (flutter/flutter#183389) 2026-03-09 [email protected] Roll Skia from 254e44141861 to a83dca75c531 (7 revisions) (flutter/flutter#183388) 2026-03-09 [email protected] Roll Fuchsia Linux SDK from giLoee6arX5CRHuRh... to ox-II3wPpa818y78_... (flutter/flutter#183386) 2026-03-09 [email protected] Roll Dart SDK from 1604910613c7 to 050b2f646425 (1 revision) (flutter/flutter#183383) 2026-03-08 [email protected] Roll Skia from af994ae4d990 to 254e44141861 (2 revisions) (flutter/flutter#183377) If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/flutter-packages Please CC [email protected] on the revert to ensure that a human is aware of the problem. To file a bug in Packages: https://github.com/flutter/flutter/issues/new/choose To report a problem with the AutoRoller itself, please file a bug: ...

fixes flutter#166173 Highlevel: instead of clearing the entire geometry from the node that markSemanticsNeedsUpdate up to the first semantics boundary. and recalculate everything in between. This code now only clear the geometry for node that are markSemanticsNeedsUpdate. This requires us to keep a separate list for geomtry dirty node. Thus this change. ## Pre-launch Checklist - [ ] I read the [Contributor Guide] and followed the process outlined there for submitting PRs. - [ ] I read the [Tree Hygiene] wiki page, which explains my responsibilities. - [ ] I read and followed the [Flutter Style Guide], including [Features we expect every widget to implement]. - [ ] I signed the [CLA]. - [ ] I listed at least one issue that this PR fixes in the description above. - [ ] I updated/added relevant documentation (doc comments with `///`). - [ ] I added new tests to check the change I am making, or this PR is [test-exempt]. - [ ] I followed the [breaking change policy] and added [Data Driven Fixes] where supported. - [ ] All existing and new tests are passing. If you need help, consider asking for advice on the #hackers-new channel on [Discord]. **Note**: The Flutter team is currently trialing the use of [Gemini Code Assist for GitHub](https://developers.google.com/gemini-code-assist/docs/review-github-code). Comments from the `gemini-code-assist` bot should not be taken as authoritative feedback from the Flutter team. If you find its comments useful you can update your code accordingly, but if you are unsure or disagree with the feedback, please feel free to wait for a Flutter team member's review for guidance on which automated comments should be addressed.  [Contributor Guide]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#overview [Tree Hygiene]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md [test-exempt]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#tests [Flutter Style Guide]: https://github.com/flutter/flutter/blob/main/docs/contributing/Style-guide-for-Flutter-repo.md [Features we expect every widget to implement]: https://github.com/flutter/flutter/blob/main/docs/contributing/Style-guide-for-Flutter-repo.md#features-we-expect-every-widget-to-implement [CLA]: https://cla.developers.google.com/ [flutter/tests]: https://github.com/flutter/tests [breaking change policy]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#handling-breaking-changes [Discord]: https://github.com/flutter/flutter/blob/main/docs/contributing/Chat.md [Data Driven Fixes]: https://github.com/flutter/flutter/blob/main/docs/contributing/Data-driven-Fixes.md

gaaclarke · 2026-03-16T18:13:55Z

@chunhtai this seems to have created a regression in one of the benchmarks (bench_material3_semantics.canvaskit.Semantics.ensureGeometry.average). Was this intentional?

https://flutter-flutter-perf.luci.app/e/?begin=1773084609&enable_chart_tooltip=true&end=1773684735&keys=Xa03992dafdbf5eb439f02fb696af44a2&plotSummary=true&selected=commit%3D2806880%26name%3D%252Carch%253Dintel%252Cbranch%253Dmaster%252Cconfig%253Ddefault%252Cdevice_type%253Dnone%252Cdevice_version%253Dnone%252Chost_type%253Dlinux%252Csub_result%253Dbench_material3_semantics.canvaskit.Semantics.ensureGeometry.average%252Ctest%253Dweb_benchmarks_ddc_hot_reload%252C%26tableCol%3D2%26tableRow%3D1475&use_titles=true&xbaroffset=2806880

chunhtai · 2026-03-16T18:22:00Z

it was expected to regress a bit, but not this much, taking a look

chunhtai · 2026-03-16T18:50:14Z

After taking a closer look, this jump in the ensuregeometry is minimal compare to the total frame, and is still in the expected range. (barely see any increase for total frame time which was ~600,000us)

This pr however reduced the run time in more common scrolling scenario, and cut the total frame time from 17,000us to 14,000us.

I think this is still a trade off worth making since the bench_material3_semantics is the worst case scenario where user close and open a new page every frame.

I also have future plan to further optimize this part, but I will fix forward instead of revert and reland.

Let me know if you think otherwise @gaaclarke

gaaclarke · 2026-03-16T19:51:50Z

Let me know if you think otherwise @gaaclarke

Your take sounds reasonable to me. Thanks for looking into it.

…r#11222) Roll Flutter from 2ec61af30672 to 195ae7b3a122 (36 revisions) flutter/flutter@2ec61af...195ae7b 2026-03-10 [email protected] [web] Updates to the README (flutter/flutter#176292) 2026-03-10 [email protected] [Impeller] Do not log VK_ERROR_SURFACE_LOST_KHR errors returned by vkAcquireNextImageKHR (flutter/flutter#183338) 2026-03-10 [email protected] DlDeferredImageGPUImpeller::ImageWrapper texture thread safety improvements (flutter/flutter#183429) 2026-03-10 [email protected] refactor: remove material imports from system_context_menu_test, table_test (flutter/flutter#183368) 2026-03-10 [email protected] refactor: remove material from baseline_test, slivers_test (flutter/flutter#183367) 2026-03-10 [email protected] Migrated circle sdf to euclidean 2d derivative (flutter/flutter#183184) 2026-03-10 [email protected] Roll Packages from c717018 to ee460d6 (2 revisions) (flutter/flutter#183451) 2026-03-10 [email protected] Roll Skia from 68f24903b592 to 4b35832cc7ea (1 revision) (flutter/flutter#183449) 2026-03-10 [email protected] Roll Skia from 8eda160375ea to 68f24903b592 (1 revision) (flutter/flutter#183444) 2026-03-10 [email protected] Roll Skia from e7b957806202 to 8eda160375ea (2 revisions) (flutter/flutter#183440) 2026-03-10 [email protected] Roll Fuchsia Linux SDK from ox-II3wPpa818y78_... to 8C_qfgWgoNhkV0_Mn... (flutter/flutter#183437) 2026-03-10 [email protected] Roll Skia from 887df915a95e to e7b957806202 (1 revision) (flutter/flutter#183434) 2026-03-10 [email protected] Remove material from scrollable_test.dart (flutter/flutter#181429) 2026-03-09 [email protected] Update 'a: text input' globs (flutter/flutter#183405) 2026-03-09 [email protected] Roll Skia from 3aa442e0624a to 887df915a95e (7 revisions) (flutter/flutter#183423) 2026-03-09 [email protected] Adds geometry dirty nodes (flutter/flutter#180375) 2026-03-09 [email protected] Roll Dart SDK from 5b1aa4f8db14 to ebef6c849489 (1 revision) (flutter/flutter#183417) 2026-03-09 [email protected] Add Alexander Dmitriev to AUTHORS (flutter/flutter#183357) 2026-03-09 [email protected] Update org triage (flutter/flutter#183254) 2026-03-09 [email protected] [Android] Avoid plugin auto-registration during FlutterFragmentActivity recreate timing window (flutter/flutter#182963) 2026-03-09 [email protected] Standardize skia includes. (flutter/flutter#183404) 2026-03-09 [email protected] Add a note to `arcTo` docs about the 2PI drawing angle issue (flutter/flutter#183293) 2026-03-09 [email protected] Roll pub packages (flutter/flutter#183411) 2026-03-09 [email protected] [ Tool ] Fix --print-dtd not working for web targets (flutter/flutter#183171) 2026-03-09 [email protected] Add await or ignore lint to flutter_driver unawaited callsites (flutter/flutter#183334) 2026-03-09 [email protected] Cleanup MockVulkan when the vulkan instance is destroyed (flutter/flutter#183324) 2026-03-09 [email protected] Updates tests commit (flutter/flutter#183301) 2026-03-09 [email protected] Implements `ClipRSuperellipse` on backdrop filter on platform view (flutter/flutter#182643) 2026-03-09 [email protected] Roll Skia from ea4fd0e1472a to 3aa442e0624a (1 revision) (flutter/flutter#183397) 2026-03-09 [email protected] Roll Dart SDK from 050b2f646425 to 5b1aa4f8db14 (1 revision) (flutter/flutter#183393) 2026-03-09 [email protected] Roll Packages from fe3de64 to c717018 (10 revisions) (flutter/flutter#183396) 2026-03-09 [email protected] Roll Skia from a83dca75c531 to ea4fd0e1472a (1 revision) (flutter/flutter#183389) 2026-03-09 [email protected] Roll Skia from 254e44141861 to a83dca75c531 (7 revisions) (flutter/flutter#183388) 2026-03-09 [email protected] Roll Fuchsia Linux SDK from giLoee6arX5CRHuRh... to ox-II3wPpa818y78_... (flutter/flutter#183386) 2026-03-09 [email protected] Roll Dart SDK from 1604910613c7 to 050b2f646425 (1 revision) (flutter/flutter#183383) 2026-03-08 [email protected] Roll Skia from af994ae4d990 to 254e44141861 (2 revisions) (flutter/flutter#183377) If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/flutter-packages Please CC [email protected] on the revert to ensure that a human is aware of the problem. To file a bug in Packages: https://github.com/flutter/flutter/issues/new/choose To report a problem with the AutoRoller itself, please file a bug: ...

fixes flutter#166173 Highlevel: instead of clearing the entire geometry from the node that markSemanticsNeedsUpdate up to the first semantics boundary. and recalculate everything in between. This code now only clear the geometry for node that are markSemanticsNeedsUpdate. This requires us to keep a separate list for geomtry dirty node. Thus this change. ## Pre-launch Checklist - [ ] I read the [Contributor Guide] and followed the process outlined there for submitting PRs. - [ ] I read the [Tree Hygiene] wiki page, which explains my responsibilities. - [ ] I read and followed the [Flutter Style Guide], including [Features we expect every widget to implement]. - [ ] I signed the [CLA]. - [ ] I listed at least one issue that this PR fixes in the description above. - [ ] I updated/added relevant documentation (doc comments with `///`). - [ ] I added new tests to check the change I am making, or this PR is [test-exempt]. - [ ] I followed the [breaking change policy] and added [Data Driven Fixes] where supported. - [ ] All existing and new tests are passing. If you need help, consider asking for advice on the #hackers-new channel on [Discord]. **Note**: The Flutter team is currently trialing the use of [Gemini Code Assist for GitHub](https://developers.google.com/gemini-code-assist/docs/review-github-code). Comments from the `gemini-code-assist` bot should not be taken as authoritative feedback from the Flutter team. If you find its comments useful you can update your code accordingly, but if you are unsure or disagree with the feedback, please feel free to wait for a Flutter team member's review for guidance on which automated comments should be addressed.  [Contributor Guide]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#overview [Tree Hygiene]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md [test-exempt]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#tests [Flutter Style Guide]: https://github.com/flutter/flutter/blob/main/docs/contributing/Style-guide-for-Flutter-repo.md [Features we expect every widget to implement]: https://github.com/flutter/flutter/blob/main/docs/contributing/Style-guide-for-Flutter-repo.md#features-we-expect-every-widget-to-implement [CLA]: https://cla.developers.google.com/ [flutter/tests]: https://github.com/flutter/tests [breaking change policy]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#handling-breaking-changes [Discord]: https://github.com/flutter/flutter/blob/main/docs/contributing/Chat.md [Data Driven Fixes]: https://github.com/flutter/flutter/blob/main/docs/contributing/Data-driven-Fixes.md

This reverts commit 1797fe2.

fixes flutter#166173 Highlevel: instead of clearing the entire geometry from the node that markSemanticsNeedsUpdate up to the first semantics boundary. and recalculate everything in between. This code now only clear the geometry for node that are markSemanticsNeedsUpdate. This requires us to keep a separate list for geomtry dirty node. Thus this change. ## Pre-launch Checklist - [ ] I read the [Contributor Guide] and followed the process outlined there for submitting PRs. - [ ] I read the [Tree Hygiene] wiki page, which explains my responsibilities. - [ ] I read and followed the [Flutter Style Guide], including [Features we expect every widget to implement]. - [ ] I signed the [CLA]. - [ ] I listed at least one issue that this PR fixes in the description above. - [ ] I updated/added relevant documentation (doc comments with `///`). - [ ] I added new tests to check the change I am making, or this PR is [test-exempt]. - [ ] I followed the [breaking change policy] and added [Data Driven Fixes] where supported. - [ ] All existing and new tests are passing. If you need help, consider asking for advice on the #hackers-new channel on [Discord]. **Note**: The Flutter team is currently trialing the use of [Gemini Code Assist for GitHub](https://developers.google.com/gemini-code-assist/docs/review-github-code). Comments from the `gemini-code-assist` bot should not be taken as authoritative feedback from the Flutter team. If you find its comments useful you can update your code accordingly, but if you are unsure or disagree with the feedback, please feel free to wait for a Flutter team member's review for guidance on which automated comments should be addressed.  [Contributor Guide]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#overview [Tree Hygiene]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md [test-exempt]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#tests [Flutter Style Guide]: https://github.com/flutter/flutter/blob/main/docs/contributing/Style-guide-for-Flutter-repo.md [Features we expect every widget to implement]: https://github.com/flutter/flutter/blob/main/docs/contributing/Style-guide-for-Flutter-repo.md#features-we-expect-every-widget-to-implement [CLA]: https://cla.developers.google.com/ [flutter/tests]: https://github.com/flutter/tests [breaking change policy]: https://github.com/flutter/flutter/blob/main/docs/contributing/Tree-hygiene.md#handling-breaking-changes [Discord]: https://github.com/flutter/flutter/blob/main/docs/contributing/Chat.md [Data Driven Fixes]: https://github.com/flutter/flutter/blob/main/docs/contributing/Data-driven-Fixes.md

…r#11222) Roll Flutter from 2ec61af30672 to 195ae7b3a122 (36 revisions) flutter/flutter@2ec61af...195ae7b 2026-03-10 [email protected] [web] Updates to the README (flutter/flutter#176292) 2026-03-10 [email protected] [Impeller] Do not log VK_ERROR_SURFACE_LOST_KHR errors returned by vkAcquireNextImageKHR (flutter/flutter#183338) 2026-03-10 [email protected] DlDeferredImageGPUImpeller::ImageWrapper texture thread safety improvements (flutter/flutter#183429) 2026-03-10 [email protected] refactor: remove material imports from system_context_menu_test, table_test (flutter/flutter#183368) 2026-03-10 [email protected] refactor: remove material from baseline_test, slivers_test (flutter/flutter#183367) 2026-03-10 [email protected] Migrated circle sdf to euclidean 2d derivative (flutter/flutter#183184) 2026-03-10 [email protected] Roll Packages from c717018 to ee460d6 (2 revisions) (flutter/flutter#183451) 2026-03-10 [email protected] Roll Skia from 68f24903b592 to 4b35832cc7ea (1 revision) (flutter/flutter#183449) 2026-03-10 [email protected] Roll Skia from 8eda160375ea to 68f24903b592 (1 revision) (flutter/flutter#183444) 2026-03-10 [email protected] Roll Skia from e7b957806202 to 8eda160375ea (2 revisions) (flutter/flutter#183440) 2026-03-10 [email protected] Roll Fuchsia Linux SDK from ox-II3wPpa818y78_... to 8C_qfgWgoNhkV0_Mn... (flutter/flutter#183437) 2026-03-10 [email protected] Roll Skia from 887df915a95e to e7b957806202 (1 revision) (flutter/flutter#183434) 2026-03-10 [email protected] Remove material from scrollable_test.dart (flutter/flutter#181429) 2026-03-09 [email protected] Update 'a: text input' globs (flutter/flutter#183405) 2026-03-09 [email protected] Roll Skia from 3aa442e0624a to 887df915a95e (7 revisions) (flutter/flutter#183423) 2026-03-09 [email protected] Adds geometry dirty nodes (flutter/flutter#180375) 2026-03-09 [email protected] Roll Dart SDK from 5b1aa4f8db14 to ebef6c849489 (1 revision) (flutter/flutter#183417) 2026-03-09 [email protected] Add Alexander Dmitriev to AUTHORS (flutter/flutter#183357) 2026-03-09 [email protected] Update org triage (flutter/flutter#183254) 2026-03-09 [email protected] [Android] Avoid plugin auto-registration during FlutterFragmentActivity recreate timing window (flutter/flutter#182963) 2026-03-09 [email protected] Standardize skia includes. (flutter/flutter#183404) 2026-03-09 [email protected] Add a note to `arcTo` docs about the 2PI drawing angle issue (flutter/flutter#183293) 2026-03-09 [email protected] Roll pub packages (flutter/flutter#183411) 2026-03-09 [email protected] [ Tool ] Fix --print-dtd not working for web targets (flutter/flutter#183171) 2026-03-09 [email protected] Add await or ignore lint to flutter_driver unawaited callsites (flutter/flutter#183334) 2026-03-09 [email protected] Cleanup MockVulkan when the vulkan instance is destroyed (flutter/flutter#183324) 2026-03-09 [email protected] Updates tests commit (flutter/flutter#183301) 2026-03-09 [email protected] Implements `ClipRSuperellipse` on backdrop filter on platform view (flutter/flutter#182643) 2026-03-09 [email protected] Roll Skia from ea4fd0e1472a to 3aa442e0624a (1 revision) (flutter/flutter#183397) 2026-03-09 [email protected] Roll Dart SDK from 050b2f646425 to 5b1aa4f8db14 (1 revision) (flutter/flutter#183393) 2026-03-09 [email protected] Roll Packages from fe3de64 to c717018 (10 revisions) (flutter/flutter#183396) 2026-03-09 [email protected] Roll Skia from a83dca75c531 to ea4fd0e1472a (1 revision) (flutter/flutter#183389) 2026-03-09 [email protected] Roll Skia from 254e44141861 to a83dca75c531 (7 revisions) (flutter/flutter#183388) 2026-03-09 [email protected] Roll Fuchsia Linux SDK from giLoee6arX5CRHuRh... to ox-II3wPpa818y78_... (flutter/flutter#183386) 2026-03-09 [email protected] Roll Dart SDK from 1604910613c7 to 050b2f646425 (1 revision) (flutter/flutter#183383) 2026-03-08 [email protected] Roll Skia from af994ae4d990 to 254e44141861 (2 revisions) (flutter/flutter#183377) If this roll has caused a breakage, revert this CL and stop the roller using the controls here: https://autoroll.skia.org/r/flutter-packages Please CC [email protected] on the revert to ensure that a human is aware of the problem. To file a bug in Packages: https://github.com/flutter/flutter/issues/new/choose To report a problem with the AutoRoller itself, please file a bug: ...

github-actions Bot added the framework flutter/packages/flutter repository. See also f: labels. label Dec 29, 2025

chunhtai force-pushed the issues/166173 branch 2 times, most recently from 1b7d1ca to fbd95d6 Compare January 8, 2026 21:34

chunhtai force-pushed the issues/166173 branch from dadd476 to 5b5c6db Compare January 12, 2026 23:18

github-actions Bot added a: accessibility Accessibility, e.g. VoiceOver or TalkBack. (aka a11y) f: scrolling Viewports, list views, slivers, etc. labels Jan 13, 2026

chunhtai force-pushed the issues/166173 branch from 1d90f77 to d081513 Compare January 13, 2026 22:02

github-actions Bot removed a: accessibility Accessibility, e.g. VoiceOver or TalkBack. (aka a11y) f: scrolling Viewports, list views, slivers, etc. labels Jan 13, 2026

chunhtai force-pushed the issues/166173 branch from d081513 to 4d83a82 Compare January 13, 2026 23:57

chunhtai requested a review from LongCatIsLooong January 14, 2026 17:38

chunhtai changed the title ~~[WIP]Adds geometry dirty nodes~~ Adds geometry dirty nodes Jan 14, 2026

chunhtai marked this pull request as ready for review January 14, 2026 17:39

gemini-code-assist Bot reviewed Jan 14, 2026

View reviewed changes

Comment thread packages/flutter/lib/src/rendering/object.dart Outdated

chunhtai requested a review from justinmc January 14, 2026 18:13

justinmc approved these changes Jan 17, 2026

View reviewed changes

Comment thread packages/flutter/lib/src/rendering/object.dart Outdated

Comment thread packages/flutter/lib/src/rendering/object.dart Outdated

LongCatIsLooong reviewed Jan 21, 2026

View reviewed changes

chunhtai force-pushed the issues/166173 branch from 927903d to 775417a Compare January 21, 2026 23:00

chunhtai requested a review from LongCatIsLooong January 21, 2026 23:07

LongCatIsLooong reviewed Jan 28, 2026

View reviewed changes

chunhtai force-pushed the issues/166173 branch from 775417a to 2402bbc Compare January 29, 2026 23:46

chunhtai requested a review from LongCatIsLooong January 30, 2026 16:08

LongCatIsLooong reviewed Feb 9, 2026

View reviewed changes

chunhtai requested a review from LongCatIsLooong February 9, 2026 23:49

LongCatIsLooong approved these changes Feb 10, 2026

View reviewed changes

chunhtai force-pushed the issues/166173 branch from 669cd0d to a33e54e Compare February 11, 2026 23:40

github-actions Bot added the a: accessibility Accessibility, e.g. VoiceOver or TalkBack. (aka a11y) label Feb 11, 2026

chunhtai added the autosubmit Merge PR when tree becomes green via auto submit App label Mar 9, 2026

auto-submit Bot added this pull request to the merge queue Mar 9, 2026

Merged via the queue into flutter:master with commit 1797fe2 Mar 9, 2026
73 of 74 checks passed

flutter-dashboard Bot removed the autosubmit Merge PR when tree becomes green via auto submit App label Mar 9, 2026

engine-flutter-autoroll mentioned this pull request Mar 10, 2026

Roll Flutter from 2ec61af30672 to 807eb3f183f1 (28 revisions) flutter/packages#11217

Closed

engine-flutter-autoroll added a commit to engine-flutter-autoroll/packages that referenced this pull request Mar 10, 2026

1797fe27c Adds geometry dirty nodes (flutter/flutter#180375)

48cab91

engine-flutter-autoroll mentioned this pull request Mar 10, 2026

Roll Flutter from 2ec61af30672 to da6c6b16d60d (29 revisions) flutter/packages#11218

Closed

engine-flutter-autoroll added a commit to engine-flutter-autoroll/packages that referenced this pull request Mar 10, 2026

1797fe27c Adds geometry dirty nodes (flutter/flutter#180375)

8301fae

engine-flutter-autoroll added a commit to engine-flutter-autoroll/packages that referenced this pull request Mar 10, 2026

1797fe27c Adds geometry dirty nodes (flutter/flutter#180375)

42c4feb

This was referenced Mar 10, 2026

Roll Flutter from 2ec61af30672 to 9db591bda44a (31 revisions) flutter/packages#11219

Closed

Roll Flutter from 2ec61af30672 to 0dce5322b7af (32 revisions) flutter/packages#11220

Closed

engine-flutter-autoroll added a commit to engine-flutter-autoroll/packages that referenced this pull request Mar 10, 2026

1797fe27c Adds geometry dirty nodes (flutter/flutter#180375)

c47fdc0

engine-flutter-autoroll mentioned this pull request Mar 10, 2026

Roll Flutter from 2ec61af30672 to 9b9c2d144430 (33 revisions) flutter/packages#11221

Closed

engine-flutter-autoroll added a commit to engine-flutter-autoroll/packages that referenced this pull request Mar 10, 2026

1797fe27c Adds geometry dirty nodes (flutter/flutter#180375)

f1b02f8

engine-flutter-autoroll mentioned this pull request Mar 10, 2026

Roll Flutter from 2ec61af30672 to 195ae7b3a122 (36 revisions) flutter/packages#11222

Merged

engine-flutter-autoroll added a commit to engine-flutter-autoroll/packages that referenced this pull request Mar 10, 2026

1797fe27c Adds geometry dirty nodes (flutter/flutter#180375)

feb94c4

Piinks mentioned this pull request Mar 13, 2026

Flutter Notable Commits - 2026 #181433

Open

chunhtai added a commit to chunhtai/flutter that referenced this pull request Mar 30, 2026

Revert "Adds geometry dirty nodes (flutter#180375)"

1786ac3

This reverts commit 1797fe2.

obiot mentioned this pull request Jul 23, 2026

[a11y] Null-check crash (_TypeError) in _RenderObjectSemantics.updateChildren — unguarded owner! at rendering/object.dart:5875 #189908

Open

4 tasks

Uh oh!

Conversation

chunhtai commented Dec 29, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pre-launch Checklist

Uh oh!

chunhtai commented Jan 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

Uh oh!

justinmc left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

LongCatIsLooong 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!

chunhtai Jan 29, 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!

Uh oh!

Uh oh!

chunhtai commented Jan 21, 2026

Uh oh!

LongCatIsLooong left a comment

Choose a reason for hiding this comment

Uh oh!

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!

chunhtai Jan 29, 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!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

chunhtai commented Jan 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

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!

LongCatIsLooong Feb 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!

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

Uh oh!

chunhtai commented Dec 29, 2025 •

edited

Loading

chunhtai commented Jan 13, 2026 •

edited

Loading

chunhtai Jan 29, 2026 •

edited

Loading

chunhtai Jan 29, 2026 •

edited

Loading

chunhtai commented Jan 29, 2026 •

edited

Loading

LongCatIsLooong Feb 10, 2026 •

edited

Loading

chunhtai Feb 10, 2026 •

edited

Loading

LongCatIsLooong left a comment •

edited

Loading