-
-
Notifications
You must be signed in to change notification settings - Fork 1.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
convolve.py: use arr[tuple(seq)]
instead of arr[seq]
-- with numpy 1.14.0
#7816
Comments
Thanks for reporting this, @NimSed ! Could you please provide a minimal example to reproduce the warning? Looks like this might still be a problem in astropy/astropy/convolution/convolve.py Line 709 in 1f34264
I wonder why CI did not catch this... @adamginsburg ? |
This is a recent change in numpy, with the intent to make indexing with a list always return the items in that list, i.e., that only with a tuple you get indexes for each axis. One can add |
I've seen this on travis a few days ago, but didn't get to the point to pin down why an error wasn't raised, so the draft issue was never opened. |
I've recently had to make the same change in other projects. |
I'm actually surprised at the number of backward-incompatible changes Numpy is making these days! |
There has been quite a bit of discussion on that for each case, but, yes, it is true, mostly because there are now a couple of developers who have started to go through and clean up... The most painful, oddly, was the work to make representations of floats consistent, since that breaks so many doctests. Anyway, it is not THAT bad - the number of |
@mhvk - I'm just jealous given that I fancy doing that for some parts of astropy ;) |
May be that's totally doable for 4.0? ;) |
Maybe a topic for the coordination meeting - under what conditions do we envisage ever breaking backward compatibility in future? The reason Python 3 was bad is because they packaged all the breaks in a single release. In a way, Numpy's approach is less impactful - just break one thing per release ;) |
Yes, numpy is working their way on the same approach as setuptools, sphinx and conda where one awaits the new release with a lump in the stomach to see what's broken this time. Not sure which one is better, from the users' point of view, this is more invisible than having all the change at once I guess. |
I dunno. I have mixed feeling on both from a downstream maintainer's point of view. 🙈 |
Long live our numpy-dev travis test runs! 🎆 |
Some times you just gotta break stuff 😁 |
There are still a few of this very same warning turn up in |
Hi.
I get such warnings when using numpy 1.14.0:
Rolling back to numpy 1.10 seems to solve the problem.
The text was updated successfully, but these errors were encountered: