Python thread pool and execution graph
This library implements a framework for managing a fixed-size thread pool whose threads pull tasks from a queue and execute them. The tasks wrap a function and its arguments into a callable object. Tasks perform a primitive unboxing of arguments - if a positional or keyword argument is itself a task, then the parent task will remain in a blocking state until the subtask has been run. The library is best suited for long-running calculations that release the Global Interpreter Lock (GIL) or for blocking I/O that does the same.
The task framework captures the return value from a function, the exception raised during function execution or a cancellation exception if the task was cancelled. A caller can either issue a blocking call for the task value (which will reraise a raised exception) or can provide a callback that will be called when the task has been executed.
An example
import puuul
puuul.start() t1 = puuul.do(sum, [1, 2]) t2 = puuul.do(sum, [3, 4], start=1) t3 = puuul.do(cmp, t1, t2) print t3() # prints -1 print t2() # prints 1+3+4 = 8
print puuul.do(cmp, puuul.do(sum, [1, 2]), puuul.do(sum, [3, 4], start = 1))