[welder]

I like how simple it is, but I didn't like the threading.lock usage in your client because of overhead and bottlenecks.[0] Maybe it's not a problem, but my API doesn't use threading so locking isn't necessary. I prefer to instrument my own API with a Python timing decorator, or similar. I also want to instrument my WakaQ background tasks so having a decorator or context manager would be useful there.

[0] https://stackoverflow.com/a/30115176/1290627

I like your pricing, simplicity, and ease of setup (minus being able to instrument my WakaQ tasks). I'll probably try it out!

[itssimon]

Thanks for your feedback!

Being able to instrument background tasks is a great idea that I would find useful for my own projects as well. I'll consider adding that to the roadmap!

The threading.Lock usage is a bit of a precaution that might not always be necessary. However, in cases where it's not necessary, the lock will always be uncontended and then be trivial to acquire. I'll think about whether that's worth optimizing.