I am attempting to use an externally defined function inside of a numba jitted function, through use of numba's overload (docs here) but the Python kernel dies when I run it. Here is my minimal example to reproduce: from numpy.polynomial.legendre import leggauss as lg # My externally defined function import numba as nb from numba.extending import overload @overload(lg) def implementlg(n): if not isinstance(n, (int, nb.types.Integer, nb.int32)): raise nb.errors.TypingError("must be int") def impli(n): return lg(n) return impli @nb.jit(nopython=True) def tmp(): return lg(15) tmp() I would guess my problem has something to do with the typing in my implementlg function, but without a stack trace I am at a loss. Continue reading...
