Add tests for Myers false-positive merge conflicts by mtsgrd · Pull Request #13031 · gitbutlerapp/gitbutler

mtsgrd · 2026-03-25T10:45:55Z

Adds test coverage for the gitoxide Myers diff bug ↗ in the context of workspace management. When two branches make non-overlapping edits to the same file, the Myers diff algorithm produces split hunks with empty insertions that falsely collide during the three-way merge — causing branches to be unapplied from the workspace.

Test results against unpatched gitoxide

Test	Result	Outcome
Basic merge of non-overlapping edits	FAIL	Second branch falsely unapplied
Hero stack with non-overlapping edits	FAIL	Hero preserved, non-hero falsely unapplied
Multi-commit branches	FAIL	Same false unapply — deeper history doesn't help
Squash then workspace merge	FAIL	Squash succeeds, but merge still falsely unapplies the other branch
Amend then workspace merge	FAIL	Amend succeeds, but merge still falsely unapplies the other branch
Stack ordering symmetry	PASS	Bug is symmetric — second stack always unapplied regardless of order

All five failing tests should pass once the upstream fix ( GitoxideLabs/gitoxide#2476 ↗ ) lands.

Test plan

Verify all 6 tests compile and run: cargo test -p but-workspace --test workspace myers_false_conflict
Confirm 5 expected failures against current gitoxide
Re-run after upgrading gitoxide to confirm all tests pass

vercel · 2026-03-25T10:45:58Z

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
gitbutler-web	Ignored	Preview	Mar 28, 2026 1:54pm

Copilot

Pull request overview

Adds regression test coverage in but-workspace for a known gitoxide/gix Myers diff false-positive conflict that can cause otherwise independent stacks to be marked conflicting and unapplied during workspace merges.

Changes:

Add a new Rust test module covering several merge scenarios (basic, hero stack, multi-commit, squash, amend, ordering symmetry).
Add two new scenario fixtures to reproduce the problematic non-overlapping edits in a single file (single-commit and multi-commit variants).
Register the new test module in the workspace commit test suite.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.

File	Description
crates/but-workspace/tests/workspace/commit/myers_false_conflict.rs	New regression tests exercising workspace merges that currently trigger Myers false conflicts.
crates/but-workspace/tests/workspace/commit/mod.rs	Wires the new test module into the existing commit test suite.
crates/but-workspace/tests/fixtures/scenario/myers-false-conflict-same-file.sh	New fixture repo reproducing non-overlapping edits in the same file.
crates/but-workspace/tests/fixtures/scenario/myers-false-conflict-multi-commit.sh	New fixture repo reproducing the same issue with multi-commit branches for squash/amend cases.

crates/but-workspace/tests/workspace/commit/myers_false_conflict.rs

Copilot

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

Copilot · 2026-03-25T15:19:52Z

crates/but-workspace/tests/workspace/commit/myers_false_conflict.rs

+    // Verify the extra file is in the merged tree
+    let commit = merge_out.workspace_commit_id.attach(&repo).object()?;
+    let tree_viz = visualize_tree(commit.peel_to_tree()?.id());
+    insta::assert_snapshot!(tree_viz);


insta::assert_snapshot!(tree_viz); will look for (or try to create) an external snapshot file under a snapshots/ directory. Since this PR doesn't add the corresponding .snap file, this test will fail in CI unless snapshots are updated. Consider switching to an inline snapshot (second argument) or add the generated snapshot file to the repo.

Suggested change

insta::assert_snapshot!(tree_viz);

assert!(

tree_viz.contains("extra-file.txt"),

"Merged tree should contain amended extra file, but got:\n{tree_viz}"

);

Copilot · 2026-03-25T15:19:53Z

crates/but-workspace/tests/workspace/commit/myers_false_conflict.rs

+///   1. A config file has three sections: `[alpha]`, `[bravo]`, `[charlie]`.
+///   2. User creates branch A and deletes the `[alpha]` section.
+///   3. User creates branch B and deletes the `[bravo]` section.


The reproduction comment describes sections like [alpha], [bravo], [charlie], but the actual fixture content and branch names use alpha_x, bravo_x, charlie_x. Aligning the comment with the fixture makes it easier to understand and maintain this test.

Suggested change

/// 1. A config file has three sections: `[alpha]`, `[bravo]`, `[charlie]`.

/// 2. User creates branch A and deletes the `[alpha]` section.

/// 3. User creates branch B and deletes the `[bravo]` section.

/// 1. A config file has three sections: `[alpha_x]`, `[bravo_x]`, `[charlie_x]`.

/// 2. User creates branch A and deletes the `[alpha_x]` section.

/// 3. User creates branch B and deletes the `[bravo_x]` section.

Copilot · 2026-03-25T15:19:53Z

crates/but-workspace/tests/workspace/commit/myers_false_conflict.rs

+mod utils {
+    use but_core::ref_metadata::{
+        StackId, WorkspaceCommitRelation::Merged, WorkspaceStack, WorkspaceStackBranch,
+    };
+    use but_meta::VirtualBranchesTomlMetadata;
+    use gix::refs::Category;
+
+    use crate::ref_info::with_workspace_commit::utils::{StackState, add_stack_with_segments};
+
+    pub fn add_stacks(
+        meta: &mut VirtualBranchesTomlMetadata,
+        short_stack_names: impl IntoIterator<Item = &'static str>,
+    ) {
+        for (idx, stack_name) in short_stack_names.into_iter().enumerate() {
+            add_stack_with_segments(
+                meta,
+                idx as u128 + 1,
+                stack_name,
+                StackState::InWorkspace,
+                &[],
+            );
+        }
+    }
+
+    pub fn to_stacks(
+        short_stack_names: impl IntoIterator<Item = &'static str>,
+    ) -> Vec<WorkspaceStack> {
+        short_stack_names
+            .into_iter()
+            .map(|short_name| WorkspaceStack {
+                id: StackId::generate(),
+                workspacecommit_relation: Merged,
+                branches: vec![WorkspaceStackBranch {
+                    ref_name: Category::LocalBranch
+                        .to_full_name(short_name)
+                        .expect("known good short ref name"),
+                    archived: false,
+                }],
+            })
+            .collect()
+    }
+}
+use utils::{add_stacks, to_stacks};
+


This file defines a local mod utils with add_stacks/to_stacks that duplicates the helper module already present in crates/but-workspace/tests/workspace/commit/mod.rs (inside from_new_merge_with_metadata). Consider reusing a shared helper (or renaming to avoid divergence) so future changes don’t need to be made in multiple places.

Suggested change

mod utils {

use but_core::ref_metadata::{

StackId, WorkspaceCommitRelation::Merged, WorkspaceStack, WorkspaceStackBranch,

};

use but_meta::VirtualBranchesTomlMetadata;

use gix::refs::Category;

use crate::ref_info::with_workspace_commit::utils::{StackState, add_stack_with_segments};

pub fn add_stacks(

meta: &mut VirtualBranchesTomlMetadata,

short_stack_names: impl IntoIterator<Item = &'static str>,

) {

for (idx, stack_name) in short_stack_names.into_iter().enumerate() {

add_stack_with_segments(

meta,

idx as u128 + 1,

stack_name,

StackState::InWorkspace,

&[],

);

}

}

pub fn to_stacks(

short_stack_names: impl IntoIterator<Item = &'static str>,

) -> Vec<WorkspaceStack> {

short_stack_names

.into_iter()

.map(|short_name| WorkspaceStack {

id: StackId::generate(),

workspacecommit_relation: Merged,

branches: vec![WorkspaceStackBranch {

ref_name: Category::LocalBranch

.to_full_name(short_name)

.expect("known good short ref name"),

archived: false,

}],

})

.collect()

}

}

use utils::{add_stacks, to_stacks};

use super::from_new_merge_with_metadata::utils::{add_stacks, to_stacks};

Caleb-T-Owens · 2026-03-25T15:36:27Z

Hey! I'm curious if you could elaborate on why your interested in testing at this particular place and level

mtsgrd · 2026-03-25T16:46:54Z

Hey sorry, I should have put this in draft. It's related to GitoxideLabs/gitoxide#2476, I'm trying to figure out how I got myself into a position to discover that problem in the first place.

It's half baked atm, and I'm omw to dinner now.. but I should have some conclusion from these test cases by tomorrow.

Add test fixtures and 7 workspace-level tests that reproduce the gitoxide Myers diff bug where non-overlapping edits to the same file are falsely reported as conflicts. The tests cover single-file edits, multi-commit branches, amend-then-merge, and squash-then-merge scenarios. Five tests fail as expected, documenting the bug until the upstream fix (gitoxide#2476) lands.

github-actions bot added the rust Pull requests that update Rust code label Mar 25, 2026

mtsgrd marked this pull request as ready for review March 25, 2026 10:46

mtsgrd requested review from Byron, Caleb-T-Owens and krlvi as code owners March 25, 2026 10:46

Copilot AI review requested due to automatic review settings March 25, 2026 10:46

Copilot started reviewing on behalf of mtsgrd March 25, 2026 10:47 View session

Copilot AI reviewed Mar 25, 2026

View reviewed changes

mtsgrd force-pushed the fix/myers-false-conflict-tests branch from f966237 to d582c07 Compare March 25, 2026 14:59

Copilot AI review requested due to automatic review settings March 25, 2026 15:08

mtsgrd force-pushed the fix/myers-false-conflict-tests branch from d582c07 to c62de4c Compare March 25, 2026 15:08

Copilot started reviewing on behalf of mtsgrd March 25, 2026 15:08 View session

Copilot AI reviewed Mar 25, 2026

View reviewed changes

mtsgrd marked this pull request as draft March 25, 2026 16:45

mtsgrd force-pushed the fix/myers-false-conflict-tests branch from c62de4c to a8f6a7f Compare March 28, 2026 13:44

mtsgrd force-pushed the fix/myers-false-conflict-tests branch from a8f6a7f to f318f28 Compare March 28, 2026 13:54

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add tests for Myers false-positive merge conflicts#13031

Add tests for Myers false-positive merge conflicts#13031
mtsgrd wants to merge 1 commit intomasterfrom
fix/myers-false-conflict-tests

mtsgrd commented Mar 25, 2026 •

edited

Loading

Uh oh!

vercel bot commented Mar 25, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 25, 2026

Uh oh!

Copilot AI Mar 25, 2026

Uh oh!

Copilot AI Mar 25, 2026

Uh oh!

Caleb-T-Owens commented Mar 25, 2026 •

edited

Loading

Uh oh!

mtsgrd commented Mar 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

-    insta::assert_snapshot!(tree_viz);
+    assert!(
+        tree_viz.contains("extra-file.txt"),
+        "Merged tree should contain amended extra file, but got:\n{tree_viz}"
+    );

Conversation

mtsgrd commented Mar 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Test results against unpatched gitoxide

Test plan

Uh oh!

vercel bot commented Mar 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 25, 2026

Choose a reason for hiding this comment

Uh oh!

Caleb-T-Owens commented Mar 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mtsgrd commented Mar 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

mtsgrd commented Mar 25, 2026 •

edited

Loading

vercel bot commented Mar 25, 2026 •

edited

Loading

Caleb-T-Owens commented Mar 25, 2026 •

edited

Loading