Open haimivan opened 4 years ago
Thanks for the report. This shouldn't have happened, but equally I'm not sure compiling this code makes sense? The @jit
decorator has the forceobj=True
kwarg supplied which will force the compiler to compile in object mode
, there's then an object mode block defined, but the compiler is already in object mode, so I'm not sure what this is trying to achieve?
To learn more about nopython
vs object mode
see: http://numba.pydata.org/numba-doc/latest/user/5minguide.html#what-is-nopython-mode
Hi @stuartarchibald ,
I came across this issue when fiddling around with https://github.com/numba/numba/issues/5093 .
I am with you that it does not make really sense, but my thought was that it maybe might help to analyse that issue or others.
Thanks, whilst unusual, am labeling this as a bug as LoweringErrors should not end up in user space.
Hi,
this snippet gets called several times in my main Numba processing function:
Using the directives
@nb.jit(boundscheck=True, debug=True, forceobj=True)
leads to this error (at least sometimes):Please see also https://github.com/numba/numba/issues/5093
Unfortunately, I cannot share my code. Extracting it to a minimal reproducable code snippet did not work neither.