0c8442a8d1
Phase 7 of the whole-project review (findings 3.2 and 3.3). Service.mu is the lock the Fyne main thread takes on every Jobs() and Runtime() call, so anything blocking inside it makes a UI refresh wait on the disk. Three things did: - Every SaveJobs/SaveConfig was a marshal, fsync, and rename under mu. Writes are now prepared under the lock (Store.PrepareSaveJobs / PrepareSaveConfig snapshot the payload and target path) and run after it is released. deferSaveLocked takes saveMu while mu is still held, so writes still reach the file in the order their snapshots were taken and an older snapshot can never land on top of a newer one. - executeRun ran runner.CleanupLogs under mu after every run. It needs only the values already snapshotted into runEnv, so it now runs after the unlock — including when the job is gone, since the run still wrote a log file that retention covers. - adoptJobsLocked ran runner.SeedStats under mu, reached from UpdateSettings on the UI thread. Seeding moved out into applySeededStatsLocked; UpdateSettings now reads the new jobs file and seeds its statistics before taking the lock, and re-checks the "no jobs-file switch while running" guard once it has it. SeedStats also opened every log file twice — once to find the job, again to read the result. readLogSummary reads job_id, state, and duration in one pass, so each log is opened once. StartOnly runs were built with exec.CommandContext on the app's lifecycle context. os/exec keeps a watcher goroutine alive until Wait returns or the context is done, and StartOnly never calls Wait, so one goroutine leaked per run and would then try to kill a process whose handle startJobOnly had already released. The invocation now uses context.Background(), whose nil Done channel means no watcher is started at all. Regression tests: TestRunJobStartOnlyLeavesNoContextWatcher (fails with 5 leaked goroutines on the old code), TestConcurrentJobOperationsLeaveTheFileMatchingMemory, and TestUpdateSettingsSeedsAdoptedJobsFromLogs. STANDARDS gains the no-I/O-under-mu rule and the "a StartOnly process outlives GoSentry" entry. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
293 lines
15 KiB
Markdown
293 lines
15 KiB
Markdown
# GoSentry Architecture
|
|
|
|
This document shows the current component interaction model. GoSentry is a
|
|
single desktop process: the GUI, application service, scheduler, storage, and
|
|
command runner live in one application. They communicate through typed events
|
|
and well-defined interfaces rather than shared mutable state.
|
|
|
|
## Package Map
|
|
|
|
```
|
|
cmd/gosentry entry point — starts the UI
|
|
src/
|
|
domain/ pure value types: Job, Config, RunRecord, Schedule, JobRuntime
|
|
app/ Service — sole owner of job/runtime state; emits typed Events
|
|
scheduler/ pure timing loop; calls app.Service.RunDue on every tick
|
|
runner/ shell command execution + log file writing + cleanup
|
|
storage/ JSON persistence (gosentry.json, jobs.json)
|
|
platform/
|
|
autostart/ Manager interface + Windows (shortcut) and Linux (XDG) impls
|
|
desktop/ desktop entry + icon under XDG data home (Linux only)
|
|
filemanager/ open a folder in the desktop file manager
|
|
winproc/ hidden-window startup flags (Windows only)
|
|
ui/ Fyne windows, tabs, and dialogs; reads service via Events
|
|
```
|
|
|
|
## Component Diagram
|
|
|
|
```mermaid
|
|
flowchart LR
|
|
user["Desktop user"]
|
|
ui["src/ui - Fyne windows, tabs, dialogs"]
|
|
svc["src/app Service - sole owner of job + runtime state"]
|
|
store["src/storage Store - JSON config and jobs"]
|
|
sched["src/scheduler Scheduler - pure timing loop"]
|
|
runner["src/runner - shell command execution"]
|
|
autostart["src/platform/autostart Manager - Windows shortcut / Linux XDG"]
|
|
config["gosentry.json - application settings"]
|
|
jobs["jobs.json - job definitions"]
|
|
logs["logs_dir - per-run command output logs"]
|
|
shell["Platform shell - cmd.exe /C or sh -c"]
|
|
|
|
user -->|"edits jobs, settings, runs commands"| ui
|
|
ui -->|"CreateJob, UpdateJob, DeleteJob, RunNow, UpdateSettings, AutostartStatus, …"| svc
|
|
svc -->|"SaveJobs, SaveConfig, LoadJobs, LoadConfig"| store
|
|
store -->|"read/write"| config
|
|
store -->|"read/write"| jobs
|
|
|
|
svc -->|"Start(RunDue)"| sched
|
|
sched -->|"RunDue(now)"| svc
|
|
svc -->|"RunJob"| runner
|
|
runner -->|"execute command"| shell
|
|
runner -->|"write stdout/stderr log"| logs
|
|
runner -->|"RunRecord"| svc
|
|
svc -->|"emit JobChanged / RunRecorded / JobsLoaded / ErrorOccurred"| ui
|
|
ui -->|"display jobs, history, status"| user
|
|
|
|
svc -->|"Set / Status via Manager"| autostart
|
|
```
|
|
|
|
## Platform layer
|
|
|
|
GoSentry ships one binary per target OS. Platform-specific code is not a
|
|
workaround for missing cross-platform support — it **is** the cross-platform
|
|
strategy: shared interfaces and call sites, with OS-specific implementations
|
|
selected at **compile time** (`*_windows.go`, `//go:build linux`, and similar).
|
|
Runtime `runtime.GOOS` checks appear only for small UI details (see below), not
|
|
for autostart, file-manager integration, or command invocation.
|
|
|
|
Callers (`app.Service`, `ui`, `runner`) depend on the shared API; they do not
|
|
branch on the operating system.
|
|
|
|
| Package / file | Windows | Linux | Other (`!windows && !linux`) |
|
|
| --- | --- | --- | --- |
|
|
| `platform/autostart` | Startup-folder `.lnk` shortcut | XDG `~/.config/autostart/gosentry.desktop` | Stub — `Set` returns an error when enabled |
|
|
| `platform/desktop` | no-op | Installs `.desktop` + icon under XDG data home | no-op |
|
|
| `platform/filemanager` | `explorer` | `xdg-open` | Unsupported — `Open` returns an error |
|
|
| `platform/winproc` | `CREATE_NO_WINDOW` / `HideWindow` on child processes | no-op | no-op |
|
|
| `runner/invocation_*` | `cmd.exe /S /C` with Windows-safe quoting | `sh -c` | `sh -c` (same as Linux) |
|
|
|
|
**Why separate implementations are required**
|
|
|
|
- **Autostart** — each OS defines its own login startup mechanism (shortcut,
|
|
XDG Autostart, LaunchAgents on macOS). There is no portable API in Go, Fyne, or
|
|
the standard library; a third-party helper would still wrap the same per-OS
|
|
code behind an interface.
|
|
- **Opening a folder** — the desktop shell exposes no shared “reveal in file
|
|
manager” call; each platform invokes its registered handler (`explorer`,
|
|
`xdg-open`, `open` on macOS).
|
|
- **Command shell** — users expect OS-native semantics (`cmd.exe` batch files,
|
|
`%VAR%`, and path rules on Windows; POSIX `sh` on Linux). A single shell for
|
|
all platforms would break commands on one side or the other.
|
|
- **Hidden console window** — launching a child process from a GUI app can flash
|
|
a console on Windows only; Linux and macOS do not need equivalent flags.
|
|
|
|
**Deliberate platform choices (not OS API limits)**
|
|
|
|
- **Window and tray icons** — Fyne accepts icons on every platform, but Windows
|
|
renders the notification area and titlebar from multi-size `.ico` resources
|
|
(embedded via `packaging/windows/gosentry.rc`), while Linux StatusNotifier
|
|
trays scale better from a larger PNG. `ui/run.go` and `ui/tray.go` branch on
|
|
`runtime.GOOS` for asset selection only.
|
|
- **Sample job commands** in `storage/store.go` — demo `echo` lines differ only
|
|
because shell quoting rules differ; real jobs are user-authored per platform.
|
|
|
|
**Adding new platform code**
|
|
|
|
- Put OS integration in `src/platform/<name>/` with a small shared API, or use
|
|
`*_GOOS.go` files in the owning package when the surface is a single function
|
|
(as in `runner/invocation_*`).
|
|
- Do not scatter `runtime.GOOS` through `app.Service` or UI business logic.
|
|
- Unsupported platforms get an explicit stub (return an error or no-op) rather
|
|
than silently doing nothing — see `autostart_other.go` and
|
|
`filemanager_other.go`.
|
|
|
|
macOS autostart and file-manager handlers are not implemented yet; see
|
|
[ROADMAP.md](ROADMAP.md) for blocked or deferred cross-platform work (for
|
|
example window-maximized detection, which would need per-OS native calls).
|
|
|
|
## Main Flows
|
|
|
|
1. Startup:
|
|
`cmd/gosentry` calls `ui.Run`, which creates an `app.Service`, opens the
|
|
store, loads `gosentry.json` and `jobs.json`, subscribes the UI to service
|
|
events, builds the main window, and calls `Service.Start` to begin the
|
|
scheduler loop. On every launch the service seeds per-job run-time statistics
|
|
from existing log files so the details panel reflects accumulated history
|
|
immediately (see §Statistics below).
|
|
|
|
2. Editing settings or jobs:
|
|
The UI calls mutating methods on `app.Service` (e.g. `CreateJob`,
|
|
`UpdateJob`, `UpdateSettings`). The Service validates the request, updates
|
|
its in-memory state, persists through `storage.Store`, and emits a typed
|
|
`Event`. The UI's observer receives the event and refreshes the relevant
|
|
widget on the main thread via `fyne.Do`.
|
|
|
|
`UpdateSettings` has one extra step: when the configured jobs file changes
|
|
and a file already exists at the new path, that file is authoritative. The
|
|
Service loads it, calls `adoptJobsLocked` to rebuild the jobs slice, runtime
|
|
map, schedule cache, and next-run times around it, applies the statistics
|
|
seeded from the new logs directory, and emits `JobsLoaded` plus a broad
|
|
`JobChanged`. A path with no file behind it receives the current jobs instead.
|
|
Adoption drops all runtime state, so it is refused while a job is running.
|
|
Reading the new file and seeding its statistics both happen before `mu` is
|
|
taken (the no-I/O-under-`mu` rule in [STANDARDS.md](STANDARDS.md)), so the
|
|
running-job check is re-evaluated under the lock before anything is replaced.
|
|
|
|
3. Scheduled run:
|
|
`scheduler.Scheduler` fires a tick every second. On each tick it calls
|
|
`Service.RunDue(now)`. The Service checks which enabled, non-paused jobs are
|
|
due, marks each as running, and launches `runner.RunJob` in a goroutine.
|
|
|
|
4. Manual run:
|
|
`Run now` in the UI calls `Service.RunNow`. The Service checks that the job
|
|
exists, is not already running, and (in sequential mode) that no other job is
|
|
running, then executes `runner.RunJob` with the `Manual` trigger. Manual runs
|
|
are allowed even while the scheduler is globally paused.
|
|
|
|
5. Command execution:
|
|
`runner.RunJob` builds the platform-specific invocation, executes the
|
|
command through the platform shell under the caller-supplied timeout, captures
|
|
stdout and stderr, writes one timestamped `.log` file, and returns a
|
|
`domain.RunRecord` containing
|
|
`DurationMS` (wall-clock milliseconds from start to finish; for `StartOnly`
|
|
fire-and-forget jobs it measures launch latency — the time to spawn the
|
|
process — since there is no exit to wait for).
|
|
|
|
6. History update:
|
|
When a run goroutine completes, `Service` updates the job's runtime
|
|
(including the statistics aggregate) under `mu`, then — after releasing it —
|
|
runs log cleanup and emits `RunRecorded`. Nothing is saved: a run changes only
|
|
`JobRuntime`, which is never persisted. The UI observer appends the record to the History tab.
|
|
History rows exist only for the current process session; restarting the app
|
|
clears the table (aggregate stats in the details panel are still seeded from
|
|
log files).
|
|
|
|
7. Autostart:
|
|
`UpdateSettings` in the Service calls `autostart.Manager.Set`. The Manager
|
|
interface has two implementations: Windows writes a `.lnk` shortcut to the
|
|
user Startup folder; Linux writes an XDG Autostart `.desktop` file. When
|
|
`KeepRunningInTray` is enabled the entry passes `--start-in-tray`; when it is
|
|
off the entry launches the executable without that flag so the main window
|
|
opens after sign-in.
|
|
|
|
8. Error surfacing:
|
|
Background errors (failed JSON saves, cleanup errors) are emitted as
|
|
`ErrorOccurred` events and displayed in the UI status area, rather than
|
|
being silently discarded.
|
|
|
|
## Key Domain Concepts
|
|
|
|
### Per-job overlap policy
|
|
|
|
`domain.Job` carries an `OverlapPolicy` field (`json:"overlap_policy,omitempty"`).
|
|
When non-empty it overrides the global `Config.OverlapPolicy` for that job alone.
|
|
Empty means inherit the global default. `app.Service.RunDue` resolves the
|
|
effective policy per job: it uses `job.OverlapPolicy` when set, otherwise falls
|
|
back to `store.Config.OverlapPolicy`. `normalizeJob` in `app/operations.go` leaves
|
|
the field empty on new jobs so the inherit semantics are preserved.
|
|
|
|
Under the `"queue"` policy, each occurrence that fires while a run is still
|
|
in flight increments `JobRuntime.PendingRuns`. When the current run finishes,
|
|
`executeRun` drains the counter by starting one deferred run per completion until
|
|
`PendingRuns` reaches zero.
|
|
|
|
### Per-job command timeout
|
|
|
|
`domain.Job` carries a `TimeoutSeconds *int` field
|
|
(`json:"timeout_seconds,omitempty"`), following the same inherit pattern as the
|
|
overlap policy. It is a **pointer** because the setting has three states that
|
|
must stay distinguishable on disk:
|
|
|
|
| `Job.TimeoutSeconds` | jobs.json | Meaning |
|
|
| --- | --- | --- |
|
|
| `nil` | field absent | inherit `Config.DefaultTimeoutSeconds` |
|
|
| `0` | `"timeout_seconds": 0` | no timeout, does **not** inherit |
|
|
| `> 0` | `"timeout_seconds": 45` | per-job limit in seconds |
|
|
|
|
The global `Config.DefaultTimeoutSeconds` (default **0**, i.e. no timeout) is
|
|
written unconditionally — no `omitempty` — for the same reason: `0` there is a
|
|
deliberate choice, not a missing value, and `storage.loadOrCreateConfig` must not
|
|
normalize it away. `app.Service.effectiveTimeout`
|
|
resolves the effective duration under `mu` and `startRunLocked` snapshots it into
|
|
`runEnv.timeout`. `runner.RunJob(ctx, job, trigger, logsDir, timeout)` takes the
|
|
resolved duration as an argument, so the runner stays ignorant of the global
|
|
config: a positive duration applies the timeout via `context.WithTimeout` and
|
|
reports `Timed out after <timeout>` on expiry; a non-positive duration runs
|
|
without a deadline, bounded only by `ctx` (app shutdown). `StartOnly` jobs are
|
|
built on `context.Background()` instead — neither the timeout nor app shutdown
|
|
applies to them — and so measure launch latency only.
|
|
|
|
### Run-time statistics
|
|
|
|
`domain.JobRuntime` holds a rolling aggregate updated after each run:
|
|
|
|
| Field | Meaning |
|
|
|-------|---------|
|
|
| `RunCount` | total runs recorded |
|
|
| `FailCount` | runs that exited non-zero |
|
|
| `LastDurationMS` | wall-clock time of the most recent run (launch latency for `StartOnly`) |
|
|
| `AvgDurationMS` | mean over all runs with a recorded duration |
|
|
| `MaxDurationMS` | longest recorded run |
|
|
|
|
`runner.RunJob` measures the wall-clock start→finish and sets `DurationMS` on
|
|
the returned `RunRecord`. `runner/logfile.go` writes a `duration` line into the
|
|
log file header alongside the existing `state` line.
|
|
|
|
On startup, `runner.SeedStats` scans log files (matched primarily by the
|
|
`job_id` header, with a sanitized-name filename fallback for legacy logs,
|
|
bounded by `Config.MaxLogFiles`) and folds the parsed `state`/`duration`
|
|
headers into a `runner.SeededStats` map. `NewService` applies those seeds to
|
|
the runtime map before the first scheduler tick, so the details panel shows
|
|
accumulated run history immediately after a restart.
|
|
Older log files that pre-date the `duration` header are tolerated: the run is
|
|
counted but the timing is skipped.
|
|
|
|
`JobRuntime.Logs` (per-run `RunRecord` entries shown in the History tab) is
|
|
**session-only**: it is not written to `jobs.json` and is not rebuilt from
|
|
`.log` files on startup. Log files on disk feed aggregate counters via
|
|
`SeedStats` only. See [STANDARDS.md](STANDARDS.md).
|
|
|
|
### Persisted global pause
|
|
|
|
`domain.Config` carries a `Paused bool` field (`json:"paused,omitempty"`).
|
|
`app.Service.SetGlobalPause` writes the new value into `store.Config` and calls
|
|
`SaveConfig`, so the paused state survives a restart. `NewService` initialises
|
|
`s.paused` from `store.Config.Paused` and applies the paused next-run text to
|
|
all runtimes before the first tick, ensuring the UI shows the correct state from
|
|
the moment the window opens.
|
|
|
|
### `jobs_view.go` file structure
|
|
|
|
The size guideline for a file in this project is ~250 lines.
|
|
`src/ui/jobs_view.go` is split across three files along these seams; the view
|
|
file itself has grown back over the guideline since — see the split item in
|
|
[ROADMAP.md](ROADMAP.md), which tracks every file currently over it:
|
|
|
|
| File | Contents |
|
|
|------|----------|
|
|
| `jobs_view.go` | `newJobsView` — list, toolbar, button wiring, and layout |
|
|
| `jobs_view_details.go` | `detailsPanel` struct — widget creation, `update`, `clear`, `container` |
|
|
| `jobs_view_helpers.go` | Pure helpers — `filteredJobIndexes`, `folderOptions`, `filterValue`, `indexOfID`, `lastJobLogs`, `nextJobListView`, `viewToggleText` |
|
|
|
|
### `settings_view.go` file structure
|
|
|
|
`src/ui/settings_view.go` is split across three files the same way, once its
|
|
own size passed the guideline:
|
|
|
|
| File | Contents |
|
|
|------|----------|
|
|
| `settings_view.go` | `settingsView` — field construction, save, load, validate; the Theme label translation helpers |
|
|
| `settings_view_layout.go` | `newSettingsLayout`, `settingsSection`, `settingsRow` — the two-column arrangement and the button row |
|
|
| `settings_view_helpers.go` | Pure helpers — `fyneVersion`, `mustParseURL`, `settingsFolderPath`, `openFolder`, `chooseFile`/`chooseJSONFile`, `chooseFolder` (`chooseFile` also backs `job_dialog.go`'s command browser) |
|