Typing jit signature of readonly arrays inputs

[mroeschke]

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).

Is there a way to type the signature of readonly arrays? I did not find a way to specify readonly on the documentation page or StackOverflow (or maybe specifying array should work on readonly arrays as well?).

In [1]: import numba

In [2]: arr = np.array([1])

In [3]: arr.setflags(write=False)

In [4]: @numba.jit((numba.int64[:],))
   ...: def f(x):
   ...:     return x
   ...:

In [5]: f(arr)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-5-0e64fe0f2f72> in <module>
----> 1 f(arr)

/lib/python3.7/site-packages/numba/dispatcher.py in _explain_matching_error(self, *args, **kws)
    498         msg = ("No matching definition for argument type(s) %s"
    499                % ', '.join(map(str, args)))
--> 500         raise TypeError(msg)
    501
    502     def _search_new_conversions(self, *args, **kws):

TypeError: No matching definition for argument type(s) readonly array(int64, 1d, C)

In [6]: numba.__version__
Out[6]: '0.45.1'

[stuartarchibald]

Thanks for the report. Unfortunately there's no short cut for specifying an array that is read only, the full numba.types API calls have to be used. Here's an example:

In [1]: from numba import njit, types                                                                                           

In [2]: import numpy as np                                                                                                      

In [3]: arr = np.array([1])                                                                                                     

In [4]: arr.setflags(write=False)                                                                                               

In [5]: arr.flags                                                                                                               
Out[5]: 
  C_CONTIGUOUS : True
  F_CONTIGUOUS : True
  OWNDATA : True
  WRITEABLE : False
  ALIGNED : True
  WRITEBACKIFCOPY : False
  UPDATEIFCOPY : False

In [6]: @njit((types.Array(types.int64, 1, 'C', readonly=True),)) # this is an int64, 1D, C order, read only array
   ...: def f(x): 
   ...:     return x 
   ...:                                                                                                                         

In [7]: f(arr)                                                                                                                  
Out[7]: array([1])

In [8]: f.signatures                                                                                                            
Out[8]: [(readonly array(int64, 1d, C),)]

hope this helps.

[esc]

@mroeschke I will close this issue now as I believe this issue has been resolved. Thanks again for asking about this.

[mroeschke]

Great, thanks for your response!