Smoke valid args for binary ufunc tests #18
Closed
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
Try to duck-type NumPy, only without array scalars, so that
* dtype('float32').type --> np.float32
and
* np.float32(3) --> array(3.0, dtype='float32)
IOW, the attempt is to only have zero-dim arrays (we have them anyway)
and have them everywhere where NumPy creates a scalar.
Try to duck-type NumPy, only without array scalars, so that
* dtype('float32').type --> np.float32
and
* np.float32(3) --> array(3.0, dtype='float32)
IOW, the attempt is to only have zero-dim arrays (we have them anyway)
and have them everywhere where NumPy creates a scalar.
So that current tests pass. This involves fixing a slew of small issues all around but hey.
This is a stub, will need to decide and clearly articulate what exactly is a scalar.
Structured dtypes, longdoubles and all that.
Also make test_numerictypes from numpy pass. Remove weird things, xfail the "scalar does not upcast array" test, fix up the rest.
The largest difference is that our type promotions are way simpler and more straighforward. No "weak promotion", no values.
|
These smoke tests certainly do not need to be generated by a script anymore, so please feel free to fix them! Also cf gh-17, which add a bunch of tests for this. |
ev-br
reviewed
Jan 9, 2023
| try: | ||
| other_instance = dtype(other) | ||
| except TypeError: | ||
| return False |
There was a problem hiding this comment.
Not sure why it is better than raising? (Uhm, that was me. Meaning this code is not right, long tern)
ev-br
reviewed
Jan 9, 2023
| return asarray(self._tensor >= asarray(other).get()) | ||
|
|
||
| def __le__(self, other): | ||
| return asarray(self._tensor <= asarray(other).get()) |
There was a problem hiding this comment.
NB: this will need to be redone similar to gh-17
ev-br
reviewed
Jan 9, 2023
| from ..testing import assert_allclose | ||
|
|
||
|
|
||
| def test_add(): | ||
| assert_allclose(np.add(0.5, 0.6), | ||
| add(0.5, 0.6), atol=1e-7, check_dtype=False) | ||
| np.add(0.5, 0.6), atol=1e-7, check_dtype=False) |
There was a problem hiding this comment.
assert_allclose(np.add(0.5, 0.5), np.add(0.5, 0.6))?
|
Looks like the test inputs have now been fixed, so closing. |
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.
This PR is WIP, and starts from #12 as it contains a critical fix.
The binary ufunc smoke tests were passing floats to functions that didn't accept them, which this PR fixes... kinda. Running
autogen/gen_ufuncs.pyas-is does not produce the same file structure as it exists fortorch_np, and moving things over still breaks due to invalid import references.So I can work on fixing these things generally, although IMO it looks like these smoke tests don't need to be generated by a script, but instead could be both generated and ran on runtime with
pytest.mark.parametrize. Mind if I have a go exploring that?