PyPy has LOTS of problems with 3rd party libraries. If you want to deploy it in production you'll have to check that each one of them does exactly what you need it to and oftentimes you are surprised how things are broken.
We are using PyPy for some of our services (where it is doing about x3 faster than CPython), while for some others (Django UI - at least the way we are using it) we found that PyPy is actually slower, so we are sticking with CPython.
Unfortunately PyPy team has not even made it their priority to test PyPy with Django. It is one thing to have a cookie-cutter test suite that measures simple use cases and it is entirely different matter to test how well it can run the whole stack of apps on top of it.
I do not understand this kind of comment.
Even if Open Source is a great thing, some people just want to use the tool. They are not interested or do not have the time to participate in the project.
Much open source exists because people in the event of scratching their own itch, happen to release something as it might be of use to others.
When others contribute it helps make the solution more generic/robust as it is guided towards meeting multiple requirement sets.
Do you have any idea how shit it was writing software in the 1990s without the breadth of tools we have today that are open source and permissively licensed?
PyPy is a smallish project with very limited funding. If you try it out and it doesn't help you complete your goals, find another way. That might be to make it better somehow and if you don't have the resources, find an other way to meet your business goals.
Please stop replying to people like this. It's extremely discouraging to people, and not helpful to the PyPy project (of which I'm one of the developers).
People have a right to have a problem with our software without trying to fix it themselves.
I am very thankful to PyPy developers for all the help they provided though as I said it did not address all of our concerns, which is why our PyPy deployment is more limited than what we wish it could be.
We are not Google or Facebook. We cannot invest resources in hacking internals of every low-level component of our stack. We are happy to contribute feedback and test suggested solutions, but we have to stay focused on our own product and business.
We are using PyPy for some of our services (where it is doing about x3 faster than CPython), while for some others (Django UI - at least the way we are using it) we found that PyPy is actually slower, so we are sticking with CPython.
Unfortunately PyPy team has not even made it their priority to test PyPy with Django. It is one thing to have a cookie-cutter test suite that measures simple use cases and it is entirely different matter to test how well it can run the whole stack of apps on top of it.