First fixes for NumPy 2.0 #1867
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
NumPy 1.25 introduced numpy.exceptions so use this for exceptions when possible (needed for NumPy 2.0).
This is needed for NumPy 2.0 but is hopefully backwards-compatible for the NumPy versions we care about. We could have used numpy.double as the replacement, but I think it's good to be explicit about the size of the type (e.g. 64-bit).
In most cases we just use np.bytes_ (needed for NumPy 2.0) along
with a note to check whether we still need this (historically we
have had issues from crates returning bytes rather than strings,
but it's not clear if this is still present).
One case I just removed it as it was only for a warning message and
it's not clear why it was being used.
There are several minor code cleanups related to this: use of
isinstance() rather than "type(..) in ...".
A few minor drive-by code clean ups to change
if blah:
do_something
else:
raise an_error
to
if not blah:
raise an_error
do_something
Codecov Report
@@ Coverage Diff @@
## main #1867 +/- ##
=======================================
Coverage 81.23% 81.24%
=======================================
Files 73 73
Lines 26460 26463 +3
Branches 3962 3962
=======================================
+ Hits 21496 21499 +3
Misses 4803 4803
Partials 161 161
Continue to review full report in Codecov by Sentry.
|
hamogu
approved these changes
Aug 31, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
Internal code changes to improve support for NumPy 2.0.
Details
Spurned on by
let's see what changes we can make now to future-proof ourselves.
With the dvelopment build of NumPy 2.0 installed
and removing
matlpotlibandastropy(as the released versions of these I had, which are admittedly old, also have problems with NumPy 2.0) then most tests pass with this change. We do need to work out what to do with the change in string representation - e.g. #1864 (comment) - but this forms a nice limited-scope set of changes.Let's see how it copes with some of the older NumPy versions we have in our CI runs! It's hard to check against numpy 2.0 since we don't have a CI run that provides this.
There will likely be more changes needed for our compiled extensions, but that can come later (and I don't know how to make
pyproject.tomluse a development build of NumPy rather thanoldest-supported-numpy!).Changes
np.exceptionsto importVisibleDeprecationWarning, falling back to the old location if not availablenp.exceptionswas added in NumPy 1.25np.float64rather thannp.float_np.doubleinstead, but I think I like the choice of an explicit fixed-size datatype (since we have to deal with compiled code)float_andfloat64become synonyms but I assume it holds for all recent NumPy versionsnp.bytes_rather thannp.string_string_andbytes_become synonyms but I assume it holds for all recent NumPy versionsisinstancerather than an explicittypecheck