Overview - Nx

Nx

7 days30 days

Latest CI Pipeline Executions

Fuzzy

Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
3 hours ago by leosvelperez
Failed
34509
Generating fix→
90870b83 fix(misc): update ts-morph to remediate minimatch vulnerability Bump ts-morph to ^27.0.2 so publishable artifacts consume @ts-morph/common@0.28.1 and minimatch@10.2.1 transitively.
6 hours ago by leosvelperez
Failed
34444
d77d7aac chore(repo): add C build tools for jemalloc in Linux Docker builds
8 hours ago by leosvelperez
Failed
34509
Environment issue→
e88b1d10 fix(misc): bump catalog minimatch to 10.2.1 for GHSA-3ppc-4f35-3m26
9 hours ago by leosvelperez
Failed
34444
d77d7aac chore(repo): add C build tools for jemalloc in Linux Docker builds
10 hours ago by leosvelperez
Failed
34423
Generating fix→
1dfe1a8b fix(testing): move port lock files to tmpdir Port lock files were being written to the source directory, triggering the daemon file watcher and causing unnecessary project graph recomputations during Cypress test runs. Move lock files to os.tmpdir()/nx-cypress-port-locks/ to avoid this.
12 hours ago by leosvelperez
Failed
33862
Environment issue→
251f733c cleanup(core): address PR review feedback for daemon delegation - Split DbCache into abstract base + InProcessDbCache + DaemonDbCache - Split TaskHistory into interface + InProcessTaskHistory + DaemonTaskHistory - Split RunningTasksService into interface + InProcessRunningTasksService + DaemonRunningTasksService - Split TaskDetailsRecorder into interface + InProcessTaskDetailsRecorder + DaemonTaskDetailsRecorder - Add getDbCache() factory, use from getCache() and report.ts - Remove performance marks/measures added in this branch - Remove null check from isJsonMessage (no handler returns null) - Fix handleGetRunningTasks serializing a Promise instead of awaiting - Fix handleAdd/RemoveRunningTask not awaiting async methods - Fix InProcessDbCache.putResult not null-coalescing terminalOutput - Rename daemon client methods to natural names
12 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
15 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
15 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
15 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
16 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
16 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
16 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
17 hours ago by leosvelperez
Succeeded
master
678cd321 fix(core): replace buggy ignore-files trie with correct path-component gitignore matching (#34447) ## Current Behavior The `nx watch` file watcher uses the `ignore-files` and `watchexec-filterer-ignore` crates to handle `.gitignore` matching. These crates use a trie-based approach that has a bug with path-component matching — certain gitignore patterns (e.g., prefix patterns) don't match correctly, causing files that should be ignored to trigger unnecessary watch events. ## Expected Behavior Gitignore patterns are matched correctly using per-directory `Gitignore` instances from the `ignore` crate — the same crate already used by the file walker. Each `.gitignore` file is scoped to its directory, and matching is done deepest-first so that nested gitignores take priority — matching standard git behavior. ### What changed - Replaced `ignore-files` + `watchexec-filterer-ignore` with direct use of `ignore::gitignore::{Gitignore, GitignoreBuilder}`, aligning the watcher with the approach already used by the file walker - Each `.gitignore` is now compiled as a standalone instance tied to its parent directory - Gitignore evaluation walks deepest-first; first match wins - `.nxignore` matching now uses `matched_path_or_any_parents` for correct ancestor checking - `create_filter` is now synchronous (no longer `async`) since the new approach doesn't need async I/O - Removed 2 crate dependencies (`ignore-files`, `watchexec-filterer-ignore`)
19 hours ago by leosvelperez
Failed
master
4f9be499 fix(core): reduce terminal output duplication and allocations in task runner (#34427) ## Current Behavior Terminal output in the task runner is accumulated via repeated string concatenation (`terminalOutput += chunk`). Each `+=` on a growing string causes V8 to allocate a new, larger string and copy the old contents, resulting in O(n²) allocation behavior for tasks with large output. Additionally, `PseudoTtyProcess.onExit` didn't pass `terminalOutput` to its callbacks, forcing callers like `TaskOrchestrator` to duplicate output accumulation logic with a separate `onOutput` listener. ## Expected Behavior - Terminal output is collected in `string[]` arrays and joined once at the end, reducing intermediate allocations from O(n²) to O(n) - `PseudoTtyProcess.onExit` now passes `terminalOutput` as a second argument, matching the signature of other `RunningTask` implementations - `TaskOrchestrator` no longer needs a special code path for `PseudoTtyProcess` — unified `onExit` handling for all task types - `tui-summary-life-cycle` accumulates output in chunks during execution and stores the finalized string on task completion, allowing chunk arrays to be GC'd - `SeriallyRunningTasks` and `RunningNodeProcess` similarly switched to chunk-based accumulation - `BatchProcess` and `NodeChildProcessWithNonDirectOutput` lazily join and cache their terminal output
20 hours ago by leosvelperez
Canceled
master
91b350ef fix(core): skip stale recomputations and prevent lost file changes in daemon (#34424) ## Current Behavior When file changes arrive rapidly, the daemon triggers multiple concurrent project graph recomputations that all run to completion — wasting CPU/memory on redundant work and returning stale results. Additionally, after processing file changes, the daemon clears all tracked files indiscriminately. Files that changed mid-recomputation are silently lost and never reflected in the project graph until another unrelated file change arrives. ## Expected Behavior Stale recomputations detect when a newer one has started and exit early, chaining to the newer promise so callers always get the freshest result. File change tracking now uses versioned maps. Each batch of file watcher events gets a unique version, and only files matching the snapshotted version are cleared after processing. Files that changed mid-recomputation are preserved and picked up by the next cycle.
22 hours ago by leosvelperez
Canceled
master
91b350ef fix(core): skip stale recomputations and prevent lost file changes in daemon (#34424) ## Current Behavior When file changes arrive rapidly, the daemon triggers multiple concurrent project graph recomputations that all run to completion — wasting CPU/memory on redundant work and returning stale results. Additionally, after processing file changes, the daemon clears all tracked files indiscriminately. Files that changed mid-recomputation are silently lost and never reflected in the project graph until another unrelated file change arrives. ## Expected Behavior Stale recomputations detect when a newer one has started and exit early, chaining to the newer promise so callers always get the freshest result. File change tracking now uses versioned maps. Each batch of file watcher events gets a unique version, and only files matching the snapshotted version are cleared after processing. Files that changed mid-recomputation are preserved and picked up by the next cycle.
22 hours ago by leosvelperez
Canceled
34423
849a43df fix(testing): move port lock files to tmpdir Port lock files were being written to the source directory, triggering the daemon file watcher and causing unnecessary project graph recomputations during Cypress test runs. Move lock files to os.tmpdir()/nx-cypress-port-locks/ to avoid this.
22 hours ago by leosvelperez
Canceled
34423
caf8654b fix(testing): move port lock files to tmpdir Port lock files were being written to the source directory, triggering the daemon file watcher and causing unnecessary project graph recomputations during Cypress test runs. Move lock files to os.tmpdir()/nx-cypress-port-locks/ to avoid this.
22 hours ago by leosvelperez