numba jit: "Type of variable <varname>.2 cannot be determined" for a fully specified variable. #128548
Closed as not planned
Description
Bug report
Bug description:
I have several functions that work fine with jit and njit. All variables are explicitly typed, such as
def afun(parm1:int, parm2:int)->list[int]:
varname:int=3
listname:list[int]=[0]In one function, the jit complains about a variable as having an undeterminable type, even though it has been explicitly typed:
Type of variable 'artp.2' cannot be determined, operation: call $1078load_global.2($binop_add1104.10, func=$1078load_global.2, args=[Var($binop_add1104.10, primes2025a.py:240)], kws=(), vararg=None, varkwarg=None, target=None), location: /home/dakra/./primes2025a.py (240)
where the code looks like:
from numba import njit
from numba import jit
@jit
def sieve12(upToNumm: int=100000000, pnprimorial:int=3) -> list[int]:
# early in the function similar to:
C0:int=0
modprimorialdo:list[int]=[1,5]
lmodprimorialdo:int=len(modprimorialdo)
ddocol:int=C0
fdorowcol:list[int]=[C0,C0]
artp:int=C0
# and later:
# there are assignment statements for useful values for these variables, and then:
artp:int=int(ddocol+fdorowcol[C0]*lmodprimorialdo))The full error report is :
Traceback (most recent call last):
File "/home/dakra/./primes2025a.py", line 810, in <module>
print("sieve12",n,pr,len(primesl:=sieve12(n,pr)), primesl[:10], primesl[-10:])
^^^^^^^^^^^^^
File "/home/dakra/.local/lib/python3.12/site-packages/numba/core/dispatcher.py", line 423, in _compile_for_args
error_rewrite(e, 'typing')
File "/home/dakra/.local/lib/python3.12/site-packages/numba/core/dispatcher.py", line 364, in error_rewrite
raise e.with_traceback(None)
numba.core.errors.TypingError: Failed in nopython mode pipeline (step: nopython frontend)
Type of variable 'artp.2' cannot be determined, operation: call $1078load_global.2($binop_add1104.10, func=$1078load_global.2, args=[Var($binop_add1104.10, primes2025a.py:240)], kws=(), vararg=None, varkwarg=None, target=None), location: /home/dakra/./primes2025a.py (240)
File "primes2025a.py", line 240:
def sieve12(upToNumm: int=100000000, pnprimorial:int=3)->list[int]:
<source elided>
artp:int=int(ddocol+fdorowcol[C0]*lmodprimorialdo)
^
In another case, the jit processing complained about one of the parameters.
What does it take to get the jit to recognize the explicit type specifications?
CPython versions tested on:
3.12
Operating systems tested on:
Linux