mirror of
https://github.com/Brandon-Rozek/website.git
synced 2024-11-25 09:36:31 -05:00
42 lines
1.2 KiB
Markdown
42 lines
1.2 KiB
Markdown
---
|
|
title: "Quick Python: Concurrent Futures"
|
|
date: 2020-04-11T20:40:28-04:00
|
|
draft: false
|
|
tags: ["Python"]
|
|
---
|
|
|
|
Another way to perform concurrency in python is to use the `concurrent.futures` module.
|
|
|
|
```python
|
|
from concurrent.futures import ThreadPoolExecutor, ProcessPoolExecutor
|
|
def add(x, y):
|
|
return x + y
|
|
with ThreadPoolExecutor(max_workers=4) as executor:
|
|
future = executor.submit(add, 1, 2)
|
|
result = future.result(timeout=30) # unit: seconds
|
|
```
|
|
|
|
If `max_workers=None` then it will default to the number of processors on the machine multiplied by 5.
|
|
|
|
If `timeout=None` then there is no time limit applied.
|
|
|
|
You can also apply a function to a list or iterables
|
|
|
|
```python
|
|
def double(x):
|
|
return 2 * x
|
|
with ThreadPoolExecutor() as executor:
|
|
future = executor.map(function_handle, [1, 2, 3])
|
|
result = future.result()
|
|
```
|
|
|
|
Instead of threads, it is also possible to spawn processes to side-step the global interpreter lock. The documentation warns that only picklable objects can be executed and returned though.
|
|
|
|
```python
|
|
def add(x, y):
|
|
return x + y
|
|
with ProcessPoolExecutor() as executor:
|
|
future = executor.submit(add, 1, 2)
|
|
result = future.result()
|
|
```
|
|
|