Improve readability of State in taskrunner

README.md

@@ -23,12 +23,12 @@ with Local() as e:
 
 - Semaphore queue `semaphore.py` / `semaphore_subprocess.py`
 
-```
+```python
 with semaphore_queue("user@remote_host:path/to/semaphore.queue"):
     do_stuff_while_holding_semaphore()
 ```
 
-- Tasks / Taskrunner `tasks.py` / `taskrunner.py`
+- Tasks `tasks.py`
 
 ```python
 class Setup(Task):
@@ -43,6 +43,26 @@ class Test(Task):
             e.sh(f"echo Testing commit {self.state.commit}")
 ```
 
+- Taskrunner `taskrunner.py`
+
+```python
+# Create a simple State object
+state = State()
+
+# or create a State object with additional context for the tasks
+state = State(
+    commit="1234ca3d41a23d4e4f923",
+    branch="some/branch",
+    repo_name="MyRepo",
+)
+
+# Run Tasks directly
+run_tasks([Task1, Task2], state)
+
+# or run Tasks from a separate file
+run_all_tasks_in_file("tasks.py", state)
+```
+
 # Docs
 
 TODO

minimalci/taskrunner.py

@@ -109,13 +109,14 @@ def run(self) -> None:
     parser.add_argument('--file', help='Tasks file', default="tasks.py")
     args = parser.parse_args()
 
-    state = State()
-    state.commit = args.commit
-    state.branch = args.branch
-    state.repo_name = args.repo_name
-    state.log_url = args.log_url
-    state.logdir = Path(args.logdir)
-    state.identifier = args.identifier
+    state = State(
+        commit=args.commit,
+        branch=args.branch,
+        repo_name=args.repo_name,
+        log_url=args.log_url,
+        identifier=args.identifier,
+        logdir=Path(args.logdir),
+    )
 
     # Kill executor processes and cleanly exit on SIGTERM/SIGINT
     set_sigterm_sigint_global_kill_signal_handler()

minimalci/tasks.py

@@ -7,6 +7,7 @@
 from pathlib import Path
 import enum
 from dataclasses import dataclass
+import secrets
 
 from minimalci.executors import ProcessError, global_kill_signal
 from minimalci import semaphore, util
@@ -75,16 +76,22 @@ def load(cls, filepath: Path) -> "StateSnapshot":
 
 
 class State():
-    def __init__(self) -> None:
-        self.tasks: List[Task] = []
-        self.logdir = Path()
-
+    def __init__(
+        self,
+        commit: str = "",
+        branch: str = "",
+        repo_name: str = "",
+        log_url: str = "",
+        identifier: str = "",
+        logdir: Optional[Path] = None,
+    ) -> None:
         self.commit = ""
         self.branch = ""
         self.repo_name = ""
         self.log_url = ""
-        self.identifier = ""
-
+        self.identifier = identifier if identifier else secrets.token_hex(16)
+        self.logdir = logdir if logdir else Path()
+        self.tasks: List[Task] = []
         self.started = time.time()
         self.finished: Optional[float] = None
 

server/server.py

@@ -489,10 +489,11 @@ def start_taskrunner_in_docker(commit: str, branch: str) -> str:
     external_workdir = config.EXTERNAL_DATA_MOUNT_POINT / workdir.relative_to(config.DATA_PATH)
 
     # Create empty state to indicate build has been initiated
-    state = State()
-    state.branch = branch
-    state.commit = commit
-    state.identifier = identifier
+    state = State(
+        branch=branch,
+        commit=commit,
+        identifier=identifier,
+    )
     state.snapshot().save(logdir / config.STATEFILE)
 
     command = [