Para

0.0.8 verified Fri May 01 auth: no python

Para (v0.0.8) is a utility set that leverages Python's 'multiprocessing' module to distribute CPU-intensive tasks across multiple cores. It provides a simple API for parallel execution with built-in progress bar support and exception handling. Release cadence is irregular.

pip install para

Common errors

error AttributeError: Can't pickle local object 'my_function.<locals>.<lambda>' ↓

cause Lambda functions cannot be pickled by multiprocessing.

fix

Use a named top-level function instead of lambda.

error TypeError: map() got an unexpected keyword argument 'nprocesses' ↓

cause The parameter is 'nprocesses' (note the uppercase 'P'), not 'nprocesses'.

fix

Use correct keyword: map(func, data, nprocesses=4).

Warnings

gotcha The library is very early stage (v0.0.8); API may change. Not recommended for production. ↓

fix Consider alternatives like concurrent.futures, multiprocessing.Pool, or joblib.

gotcha map returns a generator, not a list. You must consume it (e.g., list(results)) to execute the tasks. ↓

fix Wrap call with list() or iterate explicitly.

gotcha If the function is not picklable (e.g., lambda, nested function, class method), multiprocessing will raise an error. Use functools.partial or top-level functions. ↓

fix Define the function at module level or use pathos.multiprocessing instead.

Imports

map
```
from para import map
```
The primary function is map, used to apply a function to an iterable in parallel.
map_unordered
```
from para import map_unordered
```
Returns results as they complete, useful for non-dependent tasks.
map_starmap
```
from para import map_starmap
```
Like map but unpacks argument tuples.

Quickstart

Parallel mapping of a CPU-intensive function across 4 processes.

from para import map

def square(x):
    return x * x

results = map(square, range(10), nprocesses=4)
print(list(results))
# [0, 1, 4, 9, 16, 25, 36, 49, 64, 81]