Hi,
I'm using the API in multithreading environment on Windows. Hence the question.
Is Shotgun API thread safe? Can I call the methods of same Shotgun object from different threads without encountering with data corruption, race condition, etc. or shall I rather treat the object as a shared resource?
Regards,
Roman Ignatov,
Shotgun Developer,
Axis Animation
My investigations on this lead me to believe that you will not suffer any data corruption, however the underlying requests can rather easily deadlock.
I am currently (at this very moment) building a Shotgun pool into my https://github.com/westernx/sgsession wrapper, as the Shotgun developers have recently stated that they will not be fixing this for the Python API version 3.
With my patch against shotgun_api3 (see: https://github.com/westernx/shotgun_api3/commit/dac6a58cb2389f0ec3b002e5eae4b7b80b025a84) my queries run in parallel very nicely, but this will not be officially supported.
Hello Mike, thanks a lot for the tip here. I have implemented it and I still getting deadlock.
I have tried to merge your proposed fix with the latest shotgun version, but I think I might have missed something.
Having a real non-blocking thread-safe shotgun_api3 would really help imho.
Dave.
Hi Dave,
I've ported it to the new master branch; see <https://github.com/mikeboers/shotgun_api3/tree/threading2>
The script I used to test it (which also uses <https://github.com/westernx/sgsession>) is at <https://gist.github.com/mikeboers/5d804639c46f051ef49e>.
Cheers,
Mike
