You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The examples in gettingstarted.md (included on the landing page for API docs) do not pass doctesting, as they are intended to. Full input/output below.
The problems seem to be cases where the order of items served by os.listdir is different than expected. I.e., for the most part the folder structures have the same items, but different orders. I have verified that the testing passes on one machine (Windows) but fails on another (Mac).
Generally, this shouldn't rely on the behavior of os.listidir for returning a particular order. Not sure exactly what to do yet, but here is my brainstorming of a few potential approaches:
Add a sort=True to every call, then output should be reliably identical. The only downside is that this makes the documentation less clean IMO (adds an argument not relevant to the concepts being displayed)
Create a separate script for testing this code. I looked into this briefly, and seems possible but perhaps overly complicated. It also doesn't solve the cases were the actual items included are different (e.g. when the itemlimit is set).
Create an alternative version of this documentation which is doctestable.
Do nothing and accept that this file cannot be doctested, or at least that a certain amount of failures are expected.
I'm most okay with option 4 if there is some improvement of the test suite, following #17.
The examples in gettingstarted.md (included on the landing page for API docs) do not pass doctesting, as they are intended to. Full input/output below.
The problems seem to be cases where the order of items served by
os.listdir
is different than expected. I.e., for the most part the folder structures have the same items, but different orders. I have verified that the testing passes on one machine (Windows) but fails on another (Mac).Generally, this shouldn't rely on the behavior of
os.listidir
for returning a particular order. Not sure exactly what to do yet, but here is my brainstorming of a few potential approaches:sort=True
to every call, then output should be reliably identical. The only downside is that this makes the documentation less clean IMO (adds an argument not relevant to the concepts being displayed)itemlimit
is set).I'm most okay with option 4 if there is some improvement of the test suite, following #17.
Input:
Output:
The text was updated successfully, but these errors were encountered: