np.trim_zeros #5543
np.trim_zeros #5543
Conversation
|
Thanks very much for the PR, I've queued it for review. |
| def impl(a, trim='fb'): | ||
| a_ = np.asarray(a) | ||
| first = 0 | ||
| if 'f' in trim: |
There was a problem hiding this comment.
From this line it seems the numpy implementation works with upper and lower case letters for trim
>>> np.trim_zeros(np.array([0, 1, 2, 3, 0, 0]), trim="FB")
array([1, 2, 3])
numba/tests/test_np_functions.py
Outdated
| yield np.zeros(1) | ||
| yield np.array([1, 2, 3]) | ||
| yield np.array([0, 1, 2, 3]) | ||
| yield np.array([0, 1, 2, 0, 0]) |
There was a problem hiding this comment.
Could we test arrays that have non-zero values that are not ints (e.g. np.array([0., 1., 2., np.nan, 0.]))?
We could also test other data types that work with the numpy version (e.g. np.array(["Hello", "world"])).
| yield np.array([0, 1, 2, 0, 0]) | ||
|
|
||
| pyfunc = np_trim_zeros | ||
| cfunc = jit(nopython=True)(pyfunc) |
There was a problem hiding this comment.
Can we also test when trim != "fb"?
|
Nice work! You'll probably want to update the numpy supported features documentation as part of your PR as well. |
|
I added support for both lower and upper case arguments and some new test cases as you mentioned. |
stuartarchibald
added
4 - Waiting on author
|
Hi, i'm working on a project that uses trim_zeros and numba, do you know why this wasn't added? |
|
@ionikha Thanks for the ping, I think it just fell by the wayside and never got a review from a maintainer (even though @EPronovost very kindly gave it a review). I've added this to the agenda for the dev meeting tomorrow to see if we can kickstart it. |
|
Assignment from triage meeting: @gmarkall to merge |
gmarkall
added
4 - Waiting on reviewer
|
@gmarkall, if needed, we can switch and I can merge |
|
@gmarkall @guilhermeleobas -- can I ask for an update here? What is the status of this? |
|
I'll put this on my near-term todo list so that it can get done. |
|
Is replaced by #9074 |
esc
added
abandoned
Add support for
np.trim_zeros. It currently does not handle nested list or tuple input well.#4074