https://www.nixtla.io logo
Join Slack
Powered by
Hi - looks like I'm getting seg faults with a re-i...
# mlforecast
b
Hi - looks like I'm getting seg faults with a re-install today. I suspect numpy 2.0 is the issue. Anyone else come across this?
j
I'm currently getting segfaults on windows with numpy 2.0. Are you using windows?
b
No, mac. I've got a MRE, shall I make an issue?
j
Oh, can you check if it's the same as this?
Basically installing mlforecast from pip on a conda env
b
Don't think so as I don't use conda. This is pip inside a docker container with no conda
j
Ok, yes please open an issue
b
So it happens when the input is polars but not for pandas. Issue here: https://github.com/Nixtla/mlforecast/issues/354
j
Thanks a lot! That line takes a polars datetime series and runs 
to_numpy
. I'll try to reproduce the issue with just that
Yeah it's coming from polars, the following produces a segfault:
Copy code
import datetime

import polars as pl

pl.Series([datetime.datetime.now()]).to_numpy()
Do you want to open the issue in the polars repo?
b
j
Thanks!
b
np
Big fan of your work @José Morales
j
Thank you for the kind words!
Open in Slack
PreviousNext
Powered by