ENH,TST: Bump stacklevel and add tests for warnings #7148
Conversation
|
Seems the stacklevel bump does not create new warnings on its own, so if wanted, I could split of the tests. |
|
Sorry, no, creates an extra warning on Python 2. (whatever the reason it survives in Python 3, did not check). Are the slow tests not run on travis (or not on python 3)? Since the new test should fail. |
11b10b7 to
a30a2dd
Compare
|
Note that this is a superset of gh-6424. |
|
Interesting failures ;) |
|
☔ The latest upstream changes (presumably #7181) made this pull request unmergeable. Please resolve the merge conflicts. |
a30a2dd to
d9413cc
Compare
|
☔ The latest upstream changes (presumably #7194) made this pull request unmergeable. Please resolve the merge conflicts. |
|
@seberg Needs rebase. |
|
The test part of this, needs the suppressed warnings thing merged anyway. Can split it off (and do the rebase) though if you like. |
|
@seberg |
|
If everyone still likes changing stack level, I guess clean it up and put it in. The test added also checks the other PR though, so will do that first unless you prefer to split the stacklevel stuff off. |
d9413cc to
e244ea8
Compare
Enforces that stacklevel is used in warnings.warn and "ignore" is not used in filterwarnings or simplefilter for most of numpy.
e244ea8 to
68ea0c7
Compare
|
OK, rebased against master and (new) tests passing. |
|
Ah, another 39 files with modifications. Homu will come calling. Thanks Sebastian ;) |
|
Hehe :). |
This is related (and relies upon the suppressed warnings pull request). The tests will not be successfully (one of them actually checks things only fixed in the other one, and I would be surprised if the stacklevel bump does not create test failures on its own.