Keep attrs in call to astype #2070
Conversation
gajomi
commented
Apr 21, 2018
gajomi
commented
- Possible route to close Keeping attributes when using DataArray.astype #2049 (discussion needed on strategy with variable, docstrings)
- Tests added (for all bug fixes or enhancements)
- Tests passed (for all non-documentation changes)
| self.data = duck_array_ops.astype(self.data, | ||
| dtype, | ||
| casting = casting, | ||
| copy = copy) |
There was a problem hiding this comment.
E251 unexpected spaces around keyword / parameter equals
| @@ -1458,6 +1458,10 @@ def test_array_interface(self): | |||
| bar = Variable(['x', 'y'], np.zeros((10, 20))) | |||
| assert_equal(self.dv, np.maximum(self.dv, bar)) | |||
|
|
|||
| def test_astype_attrs(self): | |||
| mda = self.mda.astype(bool) | |||
There was a problem hiding this comment.
It would be good to test the datatype converting behavior of astype() here, too, at least with a few minimal cases. I don't know if we have test coverage for this already.
|
Hey @gajomi, do you have time to pick this back up? Looks like it just needs a few small changes. |
|
If anyone would like to pick this up (@gajomi if you're still around!), please do; it's very close and would be a welcome addition |
|
Oh my... It's been some time since I have logged in to github... I will have to take a look at this over the weekend to see if I can remember where I was on this PR |
|
Just ran into this issue. If it's really just a matter of expanding tests and resolving conflicts of @gajomi's PR to the current codebase I can take this on, but I don't want to take their credit. Is there an easy way for a non-owner to modify a PR or should I submit a new one? |
|
Hi @dnowacki-usgs. Feel free to take all the credit! I have ended up swamped at work the last few months and just haven't had the time to get back into this yet. I am guessing easiest way to move forward would be to fork my last commit and open up a new pull request. |
* Recreate @gajomi's #2070 to keep attrs when calling astype() * Accept keep_attrs flag, use apply_ufunc in variable.py, expand tests * Ignore casting if we get a TypeError (indicates sparse <= 0.10.0) * add to whats new * fix dtype test * improve error handling and check if sparse array * test and docstring updates * catch sparse error a bit more elegantly * check using sparse_array_type Co-authored-by: Keewis <[email protected]>
