IssueTracker

Please add support for numba jit compilation in Python runtime on GCP cloud functions.

The python numba package allows for jit compilation of Python code, and is especially useful for optimising the speed of loops in Python. It is ordinarily installed via pip, and uses a function decorator to indicate which functions to optimise. For example:

import numba

@numba.njit()
def sum_test_njit(a, b):
  return a + b

print(sum_test_njit(2, 3)) # prints '5'

It would be extremely useful and potent to be able to use it from the Python runtime in GCP cloud functions. At this point in time it does not work from GCP cloud functions. The package is successfully installed, for example, when calling numba.__version__ it will show the version number of the installed package. However, when trying to use it to decorate functions it will throw an error for example:  

severity:  "ERROR"  
 textPayload:  "Traceback (most recent call last):
  File "/env/local/lib/python3.7/site-packages/google/cloud/functions_v1beta2/worker.py", line 281, in run_background_function
    _function_handler.invoke_user_function(event_object)
  File "/env/local/lib/python3.7/site-packages/google/cloud/functions_v1beta2/worker.py", line 171, in invoke_user_function
    return call_user_function(request_or_event)
  File "/env/local/lib/python3.7/site-packages/google/cloud/functions_v1beta2/worker.py", line 168, in call_user_function
    event_context.Context(**request_or_event.context))
  File "/user_code/main.py", line 25, in hello_pubsub
    print(f'numba working? {sum_test_njit(2, 3)}')
  File "/env/local/lib/python3.7/site-packages/numba/dispatcher.py", line 368, in _compile_for_args
    raise e
  File "/env/local/lib/python3.7/site-packages/numba/dispatcher.py", line 325, in _compile_for_args
    return self.compile(tuple(argtypes))
  File "/env/local/lib/python3.7/site-packages/numba/dispatcher.py", line 653, in compile
    cres = self._compiler.compile(args, return_type)
  File "/env/local/lib/python3.7/site-packages/numba/dispatcher.py", line 83, in compile
    pipeline_class=self.pipeline_class)
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 873, in compile_extra
    return pipeline.compile_extra(func)
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 367, in compile_extra
    return self._compile_bytecode()
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 804, in _compile_bytecode
    return self._compile_core()
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 791, in _compile_core
    res = pm.run(self.status)
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 253, in run
    raise patched_exception
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 245, in run
    stage()
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 678, in stage_nopython_backend
    self._backend(lowerfn, objectmode=False)
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 628, in _backend
    lowered = lowerfn()
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 615, in backend_nopython_mode
    self.flags)
  File "/env/local/lib/python3.7/site-packages/numba/compiler.py", line 991, in native_lowering_stage
    lower = lowering.Lower(targetctx, library, fndesc, interp)
  File "/env/local/lib/python3.7/site-packages/numba/lowering.py", line 93, in __init__
    self.env = Environment.from_fndesc(self.fndesc)
  File "/env/local/lib/python3.7/site-packages/numba/lowering.py", line 31, in from_fndesc
    mod = fndesc.lookup_module()
  File "/env/local/lib/python3.7/site-packages/numba/funcdesc.py", line 83, in lookup_module
    return sys.modules[self.modname]
KeyError: "Failed at nopython (nopython mode backend)\n'main'"