xfork 0.27

submitting cpu-bound tasks to processes and io-bound tasks to threads

Write a classic sequential program. Then convert it into a parallel one.

Why?

It runs faster.

What if not?

Don’t use it.

How?

for image in images:
    create_thumbnail(image)       # original

for image in images:
    fork(create_thumbnail, image) # parallelized explicitly

for image in images:
    create_thumbnail(image)       # parallelized implicitly (read below)

What about return values?

As usual:

result = fork(my_func, *args, **kwargs)

It’s a proxy object that behaves almost exactly like the real return value of my_func. Furthermore, it evaluates only if needed; also in combination with operators (like +, - etc.).

What happens to exceptions?

Their original (sequential) tracebacks are preserved. That should make debugging easier. However, don’t try to catch exceptions. You better want to exit and see them.

Speaking of threads …

and processes? fork will take care of that for you.

You can assist fork by decorating your functions (not decorating defaults to cpu_bound):

@io_bound
def call_remote_webservice():
    # implementation

@cpu_bound
def fib(n):
    # naive implementation of Fibonacci numbers

@unsafe # don't fork; run sequentially
def weird_side_effects(*args, **kwargs):
    # implementation

Parallelize implicitly?

If you don’t like the fork calling syntax, you can convert certain functions into stand-alone forks.

Use with caution.

@io_bound_fork
def create_thumbnail_by_webservice(image):
    # implementation

@cpu_bound_fork
def create_thumbnail_by_bare_processing_power(image):
    # implementation

# the following two lines spawn two forks
create_thumbnail_by_webservice(image1)
create_thumbnail_by_bare_processing_power(image2)

Conclusion

Good

• easy to give it a try / easy way from sequential to parallel and back

• results evaluate lazily

• sequential tracebacks are preserved

• it’s thread-safe / cascading forks possible

• compatible with Python 2 and 3

Bad

• weird calling syntax (no syntax support)

• type(result) == ResultProxy

• not working with lambdas due to PickleError

• needs fix:

  • fix exception handling somehow

  • not working with coroutines (asyncio) yet