Embedded Boost::Python and C++ : run in parallel

Question

How do I run C++ and Boost::Python code in parallel without problems?

Eg in my game I'd want to execute Python code in parallel with C++ code; if the embedded Python interpreter's code executes a blocking loop, like while(True): pass, the C++ code would still be running and processing frames to render with its own loop.

I tried with boost::thread and std::thread but unless I joined these threads with the main thread the program would crash...

Any suggestions or examples?

I do, but how do I keep the c++ code from terminating? Using a `while(!terminated)` loop? — UberLambda, Aug 13 '13 at 15:38
That might work, but you might need a way of telling it to stop at some point — doctorlove, Aug 13 '13 at 15:40

score 2 · Accepted Answer · edited May 23 '17 at 12:16

2

Your idea to use a second thread for the Python interpreter should just work. Make sure you use the PyGILState_Ensure/Release mechanisms everywhere you wish to run code that will invoke any Python or Boost::Python code. You have more details on this other SO thread here.

edited May 23 '17 at 12:16

Community

1
1

answered Aug 14 '13 at 13:16

André Anjos

4,641
2
27
34

It works now, adding a single line! So I don't need to relase the lock at the end? – UberLambda Aug 14 '13 at 16:08
Happy it worked out for you. The recommendation is that you release it and re-acquire it every time you need it. – André Anjos Aug 15 '13 at 06:14
To complement, we use things like smart locks to do this. See here for an example: https://github.com/idiap/bob/blob/master/src/python/gil.cc – André Anjos Aug 15 '13 at 06:16
Upvoting the comment and the answer would be awesome then! ;-) – André Anjos Aug 15 '13 at 12:13

score 0 · Answer 2 · answered Aug 13 '13 at 15:31

0

You need to use the multiprocessing module in python so that you get a separate GIL for each python thread.

answered Aug 13 '13 at 15:31

Paul Evans

27,315
3
37
54

I'd prefer avoiding Python modules as much as possible, because of the need to redistribute them :) – UberLambda Aug 13 '13 at 15:55

Embedded Boost::Python and C++ : run in parallel

2 Answers2