numpy.random.randint complaints about missing positional argument.

[thakkardharmik]

When the dtype is specified in the call to numpy.random.randint, we should set the default value for low based on the datatype. Currently the call does not support creating random int just by specifying the datatype of the random value.

RAND_MAPPING = {
    'TINYINT': np.random.randint(dtype=np.int8),
    'SMALLINT': np.random.randint(dtype=np.int16)
}
for k in range(20):
    print(RAND_MAPPING['TINYINT'])

Error stack -

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-79-67fec5a70990> in <module>()
     12 
     13 RAND_MAPPING = {
---> 14     'TINYINT': np.random.randint(dtype=np.int8),
     15     'SMALLINT': np.random.randint(dtype=np.int16)
     16 }

mtrand.pyx in mtrand.RandomState.randint()

TypeError: randint() takes at least 1 positional argument (0 given)

We should be able to generate random number just specifying datatype.

[bashtage]

The default value for low is 0. The one parameter configuration sets high.

If a 0 positional parameter version was possible, it would have to sample on [0,iinfo(dtype).max], not [iinfo(dtype).min, iinfo(dtype).max].

[ihnorton]

related to #9573