Weird warnings when using @jit on version 0.50

Reporting a bug

• I am using the latest released version of Numba (most recent is visible in the change log (https://github.com/numba/numba/blob/master/CHANGE_LOG).
• I have included below a minimal working reproducer (if you are unsure how to write one see http://matthewrocklin.com/blog/work/2018/02/28/minimal-bug-reports).

When using Numba version 0.38:

Python 3.6.5 |Anaconda, Inc.| (default, Apr 29 2018, 16:14:56) 
[GCC 7.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
>>> import numba
>>> numba.__version__
'0.38.0'
>>> from numba import jit
>>> @jit
... def _ndiff(x):
...     s = x.size
...     r = np.empty(s, dtype=np.bool) 
...     for i in range(1, s): 
...         r[i] = x[i] - x[i-1] 
...     return r
... 
>>> arr = np.random.choice([0, 1], size=1000000).astype(np.bool)
>>> %timeit _ndiff(arr)
5.32 microseconds

Now the exact same code with version 0.50:

Python 3.8.3 (default, May 19 2020, 18:47:26) 
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
>>> import numba
>>> numba.__version__
'0.50.0'
>>> from numba import jit
>>> @jit
... def _ndiff(x):
...     s = x.size
...     r = np.empty(s, dtype=np.bool) 
...     for i in range(1, s): 
...         r[i] = x[i] - x[i-1] 
...     return r
... 
>>> arr = np.random.choice([0, 1], size=1000000).astype(np.bool)
>>> %timeit _ndiff(arr)
5.32 microseconds
<stdin>:1: NumbaWarning: 
Compilation is falling back to object mode WITH looplifting enabled because Function "_ndiff" failed type inference due to: No implementation of function Function(<built-in function empty>) found for signature:
 
 >>> empty(int64, dtype=Function(<class 'bool'>))
 
There are 2 candidate implementations:
  - Of which 2 did not match due to:
  Overload in function 'empty': File: <built-in>: Line <N/A>.
    With argument(s): '(int64, dtype=Function(<class 'bool'>))':
   No match.

During: resolving callee type: Function(<built-in function empty>)
During: typing of call at <stdin> (4)


File "<stdin>", line 4:
<source missing, REPL/exec in use?>

<stdin>:1: NumbaWarning: 
Compilation is falling back to object mode WITHOUT looplifting enabled because Function "_ndiff" failed type inference due to: cannot determine Numba type of <class 'numba.core.dispatcher.LiftedLoop'>

File "<stdin>", line 5:
<source missing, REPL/exec in use?>

/home/ubuntu/anaconda3/envs/py38/lib/python3.8/site-packages/numba/core/object_mode_passes.py:177: NumbaWarning: Function "_ndiff" was compiled in object mode without forceobj=True, but has lifted loops.

File "<stdin>", line 3:
<source missing, REPL/exec in use?>

  warnings.warn(errors.NumbaWarning(warn_msg,
/home/ubuntu/anaconda3/envs/py38/lib/python3.8/site-packages/numba/core/object_mode_passes.py:187: NumbaDeprecationWarning: 
Fall-back from the nopython compilation path to the object mode compilation path has been detected, this is deprecated behaviour.

For more information visit http://numba.pydata.org/numba-doc/latest/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit

File "<stdin>", line 3:
<source missing, REPL/exec in use?>

  warnings.warn(errors.NumbaDeprecationWarning(msg,
(again, a few microseconds)

Given the performance of the @jit version in both 0.38 and 0.50, it is ABSOLUTELY AND DEFINITELY not running in object mode. Removing @jit on a 1,000,000-long array makes the Python version take forever, while the version with @jit runs in a few microseconds for the same array size.

Why so many warnings claiming that the function is running in object mode, if this is manifestly not the case?

[stuartarchibald]

Thanks for the report. The code above won’t compiled with nopython=True because np.bool is not supported as a dtype (np.bool is bool whereas np.bool_ is the dtype). The code runs at about the same speed because under @jit because Numba spots that the loop can be compiled and so compiles it and swaps out the interpreted loop for an compiled one (Compilation is falling back to object mode WITH looplifting enabled...). See also http://numba.pydata.org/numba-doc/latest/user/5minguide.html#what-is-nopython-mode

[stuartarchibald]

Closing this issue as no new information has been provided. It is assumed that the above resolved the reported issue, if this is not the case please reopen the issue with new information. Many thanks.