Add option to limit the number of workers

python/cpython#89240
Any use of mp.Pool with a number of workers greater than 60 fails
on Windows machines.

Twofold solution is introduced:
- add option for users to explicitly change number of workers
through -w/--workers, this is similar to solution used in black
- add limitation in the code for Windows machines

Co-Authored-By: Blank Spruce <32396809+BlankSpruce@users.noreply.github.com>

.gersemirc.example

@@ -6,3 +6,4 @@ line_length: 80
 list_expansion: favour-inlining
 quiet: false
 unsafe: false
+workers: 8

CHANGELOG.md

@@ -1,5 +1,9 @@
 # Changelog
 
+## [0.11.0] 2024-10-11
+### Added
+- Number of workers spawned for formatting multiple files can be changed with `-w/--workers`. By default it will be number of CPUs available in the system but limited to 60 for Windows machines due to [this](https://github.com/python/cpython/issues/89240).
+
 ## [0.10.0] 2023-12-22
 ### Added
 - configuration schema that can be used with yaml LSP server, see: [JSON Schema](https://json-schema.org/) and #12

README.md

@@ -62,6 +62,9 @@ configuration:
                         entities will be formatted in such way that sublists will be
                         completely expanded once expansion becomes necessary at all.
                         [default: favour-inlining]
+  -w INTEGER, --workers INTEGER
+                        Number of workers used to format multiple files in parallel.
+                        [default: number of CPUs on this system]
 ```
 
 ### [pre-commit](https://pre-commit.com/) hook
@@ -71,7 +74,7 @@ You can use gersemi with a pre-commit hook by adding the following to `.pre-comm
 ```yaml
 repos:
 - repo: https://github.com/BlankSpruce/gersemi
-  rev: 0.10.0
+  rev: 0.11.0
   hooks:
   - id: gersemi
 ```

gersemi/__main__.py

@@ -132,6 +132,17 @@ def create_argparser():
     [default: {Configuration.list_expansion.value}]
             """,
     )
+    configuration_group.add_argument(
+        "-w",
+        "--workers",
+        metavar="INTEGER",
+        dest="workers",
+        type=int,
+        help=f"""
+    {conf_doc['workers']}
+    [default: number of CPUs on this system]
+        """,
+    )
 
     parser.add_argument(
         dest="sources",
@@ -171,7 +182,7 @@ def main():
     configuration = make_configuration(args)
     mode = get_mode(args)
 
-    sys.exit(run(mode, configuration, args.sources))
+    sys.exit(run(mode, configuration, args.workers, args.sources))
 
 
 if __name__ == "__main__":

gersemi/__version__.py

@@ -4,4 +4,4 @@
 __license__ = "MPL 2.0"
 __title__ = "gersemi"
 __url__ = "https://github.com/BlankSpruce/gersemi"
-__version__ = "0.10.0"
+__version__ = "0.11.0"

gersemi/configuration.py

@@ -3,13 +3,24 @@
 from enum import Enum
 from hashlib import sha1
 from itertools import chain
+import multiprocessing
 import os
 from pathlib import Path
+import sys
 import textwrap
 from typing import Iterable, Optional
 import yaml
 
 
+def number_of_workers():
+    result = multiprocessing.cpu_count()
+    if sys.platform == "win32":
+        # https://bugs.python.org/issue26903
+        # https://github.com/python/cpython/issues/89240
+        return min(result, 60)
+    return result
+
+
 def doc(text: str) -> str:
     return " ".join(textwrap.dedent(text).splitlines()).strip()
 
@@ -111,15 +122,22 @@ class Configuration:
         ),
     )
 
+    workers: int = field(
+        default=number_of_workers(),
+        metadata=dict(
+            title="Workers",
+            description="Number of workers used to format multiple files in parallel.",
+        ),
+    )
+
     def summary(self):
         hasher = sha1()
         hasher.update(repr(self).encode("utf-8"))
         return hasher.hexdigest()
 
 
-def make_default_configuration_file():
-    default_configuration = Configuration()
-    d = asdict(default_configuration)
+def make_configuration_file(configuration):
+    d = asdict(configuration)
     d["list_expansion"] = d["list_expansion"].value
     result = f"""
 # yaml-language-server: $schema=https://raw.githubusercontent.com/BlankSpruce/gersemi/master/gersemi/configuration.schema.json
@@ -129,6 +147,10 @@ def make_default_configuration_file():
     return result.strip()
 
 
+def make_default_configuration_file():
+    return make_configuration_file(Configuration())
+
+
 def find_common_parent_path(paths: Iterable[Path]) -> Path:
     return Path(os.path.commonpath([path.absolute() for path in paths]))
 
@@ -168,6 +190,10 @@ def sanitize_list_expansion(list_expansion):
     )
 
 
+def sanitize_workers(workers):
+    return max(1, workers)
+
+
 def load_configuration_from_file(configuration_file_path: Path) -> Configuration:
     with enter_directory(configuration_file_path.parent):
         with open(configuration_file_path, "r", encoding="utf-8") as f:
@@ -178,6 +204,8 @@ def load_configuration_from_file(configuration_file_path: Path) -> Configuration
                 config["list_expansion"] = sanitize_list_expansion(
                     config["list_expansion"]
                 )
+            if "workers" in config:
+                config["workers"] = sanitize_workers(config["workers"])
         return Configuration(**config)
 
 
@@ -193,6 +221,8 @@ def override_configuration_with_args(
             value = normalize_definitions(value)
         if param == "list_expansion":
             value = sanitize_list_expansion(value)
+        if param == "workers":
+            value = sanitize_workers(value)
         setattr(configuration, param, value)
     return configuration
 

gersemi/configuration.schema.json

@@ -54,6 +54,12 @@
       "description": "Skip default sanity checks.",
       "title": "Unsafe",
       "type": "boolean"
+    },
+    "workers": {
+      "minimum": 1,
+      "description": "Number of workers used to format multiple files in parallel.",
+      "title": "Workers",
+      "type": "integer"
     }
   },
   "title": "Configuration",

gersemi/runner.py

@@ -137,10 +137,10 @@ def consume_task_result(task_result: TaskResult) -> Tuple[Path, int]:
     return path, return_code
 
 
-def create_pool(is_stdin_in_sources):
+def create_pool(is_stdin_in_sources, num_workers):
     if is_stdin_in_sources:
         return mp_dummy.Pool
-    return partial(mp.Pool, processes=mp.cpu_count())
+    return partial(mp.Pool, processes=num_workers)
 
 
 def filter_already_formatted_files(
@@ -163,12 +163,14 @@ def store_files_in_cache(
         cache.store_files(configuration_summary, files)
 
 
-def run(mode: Mode, configuration: Configuration, sources: Iterable[Path]):
+def run(
+    mode: Mode, configuration: Configuration, num_workers: int, sources: Iterable[Path]
+):
     configuration_summary = configuration.summary()
     requested_files = get_files(sources)
     task = select_task(mode, configuration)
 
-    pool_cm = create_pool(Path("-") in requested_files)
+    pool_cm = create_pool(Path("-") in requested_files, num_workers)
     with create_cache() as cache, pool_cm() as pool:
         files_to_format = list(
             filter_already_formatted_files(

tests/test_configuration.py

@@ -6,7 +6,7 @@
 from gersemi.configuration import (
     Configuration,
     ListExpansion,
-    make_default_configuration_file,
+    make_configuration_file,
 )
 
 try:
@@ -44,6 +44,10 @@ def generate(self, schema, mode=None):
         }
         result["properties"]["list_expansion"]["$ref"] = "#/$defs/ListExpansion"
         del result["properties"]["list_expansion"]["allOf"]
+
+        del result["properties"]["workers"]["default"]
+        result["properties"]["workers"]["minimum"] = 1
+
         return result
 
 
@@ -65,4 +69,6 @@ def test_example_file_in_repository_is_consistent_with_configuration_definition(
     with open(example_path, "r", encoding="utf-8") as f:
         example = f.read().strip()
 
-    assert example == make_default_configuration_file()
+    configuration = Configuration()
+    configuration.workers = 8
+    assert example == make_configuration_file(configuration)