feat(cmd): add --default-working-dir flag to start command
#1459
Conversation
WalkthroughAdds a --default-working-dir option and propagates it from CLI (start/enqueue) through the spec loader and builder into runtime, enabling a default working directory for DAGs without an explicit workingDir and passing that value into sub-DAG execution commands. Changes
Sequence DiagramsequenceDiagram
autonumber
participant User
participant CLI as CLI (start/enqueue)
participant Loader as Spec Loader
participant Builder as Spec Builder
participant Executor as SubDAG Executor
Note over CLI,Loader: CLI reads --default-working-dir flag
User->>CLI: invoke start/enqueue with --default-working-dir
CLI->>Loader: loadDAGWithParams(..., WithDefaultWorkingDir(value))
Loader->>Loader: set LoadOptions.defaultWorkingDir
Loader->>Builder: create BuildOpts with DefaultWorkingDir
Builder->>Builder: buildWorkingDir: choose explicit > DefaultWorkingDir > file dir
Loader-->>CLI: return DAG spec (with resolved workingDir)
CLI->>Executor: launch sub-DAG
Executor->>Executor: buildCommand() adds --default-working-dir if workDir present
Executor-->>User: sub-DAG process started with args (including default working dir)
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes
Poem
Pre-merge checks and finishing touches❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
📜 Recent review detailsConfiguration used: CodeRabbit UI Review profile: CHILL Plan: Pro 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
Comment |
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (3)
internal/cmd/flags.go (1)
160-165:defaultWorkingDirFlagdefinition is consistent and clearFlag name/description match the new behaviour, and leaving
bindViperunset keeps it strictly CLI-scoped, which seems appropriate here. If you ever want config-driven defaults, you could later addbindViper: true, but it’s not required for this PR.internal/core/spec/builder.go (1)
81-86: DefaultWorkingDir threading and precedence look correctAdding
DefaultWorkingDirtoBuildOptsand checking it inbuildWorkingDirbetween explicitspec.WorkingDirand file-based fallback gives the intended override semantics for inherited working dirs, matching the new tests. Since callers (CLI/sub-DAG runner) already pass a concrete working directory, skipping env/path expansion here is reasonable; if you ever expose more user-facing inputs, you might normalise that value before constructingBuildOptsrather than here.Also applies to: 644-680
internal/cmd/start.go (1)
60-60:--default-working-diris correctly wired into DAG loadingAdding
defaultWorkingDirFlagtostartFlagsand threading"default-working-dir"throughloadDAGWithParamsintospec.WithDefaultWorkingDircleanly connects the CLI flag toBuildOpts.DefaultWorkingDirfor both root and sub-DAG runs. The comment mentions sub-DAG inheritance (which is the main use case) but this also intentionally affects root DAGs when the flag is set, as covered byTestWithDefaultWorkingDir.Also applies to: 369-377
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (8)
internal/cmd/enqueue.go(1 hunks)internal/cmd/flags.go(1 hunks)internal/cmd/start.go(2 hunks)internal/core/spec/builder.go(2 hunks)internal/core/spec/loader.go(4 hunks)internal/core/spec/loader_test.go(2 hunks)internal/integration/workingdir_test.go(2 hunks)internal/runtime/executor/dag_runner.go(1 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.go
📄 CodeRabbit inference engine (AGENTS.md)
**/*.go: Backend entrypoint incmd/orchestrates the scheduler and CLI; runtime, persistence, and service layers sit underinternal/*(for exampleinternal/runtime,internal/persistence)
Keep Go filesgofmt/goimportsclean; use tabs, PascalCase for exported symbols (SchedulerClient), lowerCamelCase for locals, andErr...names for package-level errors
Repository linting relies ongolangci-lint; prefer idiomatic Go patterns, minimal global state, and structured logging helpers ininternal/common
Files:
internal/core/spec/builder.gointernal/cmd/start.gointernal/integration/workingdir_test.gointernal/core/spec/loader_test.gointernal/runtime/executor/dag_runner.gointernal/cmd/flags.gointernal/core/spec/loader.gointernal/cmd/enqueue.go
**/*_test.go
📄 CodeRabbit inference engine (AGENTS.md)
**/*_test.go: Co-locate Go tests as*_test.go; favour table-driven cases and cover failure paths
Usestretchr/testify/requireand shared fixtures frominternal/testinstead of duplicating mocks
Files:
internal/integration/workingdir_test.gointernal/core/spec/loader_test.go
🧠 Learnings (1)
📚 Learning: 2025-12-04T10:34:17.051Z
Learnt from: CR
Repo: dagu-org/dagu PR: 0
File: AGENTS.md:0-0
Timestamp: 2025-12-04T10:34:17.051Z
Learning: Applies to **/*_test.go : Co-locate Go tests as `*_test.go`; favour table-driven cases and cover failure paths
Applied to files:
internal/core/spec/loader_test.go
🧬 Code graph analysis (4)
internal/cmd/start.go (2)
internal/common/logger/context.go (1)
Errorf(75-77)internal/core/spec/loader.go (1)
WithDefaultWorkingDir(124-128)
internal/core/spec/loader_test.go (1)
internal/core/spec/loader.go (3)
LoadYAML(169-183)WithDefaultWorkingDir(124-128)Load(145-166)
internal/runtime/executor/dag_runner.go (2)
internal/common/config/context.go (1)
ConfigFileUsed(26-31)internal/core/dag.go (1)
DAG(33-135)
internal/core/spec/loader.go (1)
internal/core/spec/builder.go (1)
BuildFlag(56-56)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Test on ubuntu-latest
🔇 Additional comments (5)
internal/cmd/enqueue.go (1)
32-32: Enqueue now correctly exposes--default-working-dirIncluding
defaultWorkingDirFlaginenqueueFlagskeeps enqueue’s DAG loading behaviour aligned withstart(both can now driveWithDefaultWorkingDirvialoadDAGWithParams). Looks good.internal/core/spec/loader_test.go (1)
3-13: NewTestWithDefaultWorkingDirthoroughly exercises precedence rulesThe added test cases (no file context, default vs file directory, explicit workingDir overriding the default) align with the new loader/builder semantics and should prevent regressions. Use of
t.Parallel,require, andassertis consistent with the existing suite and prior testing guidelines. Based on learnings, this is solid coverage.Also applies to: 549-617
internal/integration/workingdir_test.go (1)
20-22: Integration test expectations now match inherited/override semanticsThe updated description and
assert.Containschecks forcall_child_with_wdvscall_child_no_wdcorrectly encode the new behaviour: explicit subDAGworkingDiroverrides, while a missingworkingDirinherits the parent’s working directory for local runs. This gives good end-to-end coverage of the CLI + loader + executor path.Also applies to: 90-104
internal/runtime/executor/dag_runner.go (1)
119-134: SubDAG executor now correctly passes--default-working-dirtostartConditionally appending
--default-working-dir=<workDir>and moving the DAG path after config arguments keeps the command line well-formed while enabling the loader to resolve a default working directory for child DAGs. Combined withcmd.Dir = workDir, this should give the desired inheritance semantics for local sub-DAG execution.internal/core/spec/loader.go (1)
29-37: Loader-level defaultWorkingDir option is wired cleanly with one minor noteExtending
LoadOptionswithdefaultWorkingDir, addingWithDefaultWorkingDir, and threading that intoBuildOpts.DefaultWorkingDirin bothLoadandLoadYAMLgives a consistent API surface and matches the builder/test expectations.The repository uses keyed fields consistently in
BuildOptscomposite literals. Two emptyBuildOpts{}literals exist inloader_test.go(lines 200, 210), which are safe from field reordering concerns. All other instances explicitly name their fields.
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #1459 +/- ##
==========================================
- Coverage 59.85% 59.80% -0.06%
==========================================
Files 188 188
Lines 21238 21316 +78
==========================================
+ Hits 12713 12749 +36
- Misses 7202 7234 +32
- Partials 1323 1333 +10
... and 8 files with indirect coverage changes Continue to review full report in Codecov by Sentry.
🚀 New features to boost your workflow:
|
Summary by CodeRabbit
New Features
Tests
✏️ Tip: You can customize this high-level summary in your review settings.