-
Notifications
You must be signed in to change notification settings - Fork 4.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ENH: Fix Pandas 0.19.2 compatibility issues #1975
Conversation
5a0ddce
to
ce63a3d
Compare
Hi Miguel, This looks like a great WIP, and seems like the fix for some issues you might be running into. Just wanted to message you in regards to some stuff here. So the Quantopian platform runs on pandas 0.18, and as a result, for us to make pandas 0.19 required would mean that many of our users would have go change their algorithms to be compatible with panda 0.19. I think the most valuable thing here is to support pandas 0.19, and not make it required. That way the users on Quantopian are not affected by this change and so that you (along with other Zipline users) can continue to use Zipline. Just figured I'd bring this up before you put in a bunch of other awesome work into this, and so there are less complications along the way. |
@Peque : just curious: have you tried to bump it to 0.20? |
d60b506
to
cbe5ff4
Compare
For those cases in which other type of warnings (i.e.: RuntimeWarning, that could be thrown by NumPy) should not affect the tests results.
@freddiev4 Does it look better now? |
@tibkiss Yeah, that resulted in new errors popping out, so I guess it is better to go step by step. 😊 |
@freddiev4 Friendly ping. 😊 |
Considering we merged #2194 should we close this @richafrank? In favor of #2243 |
Agreed |
Pandas 0.19 fixes some issues with Python 3.6 compatibility. It also adds new features and performance improvements. 0.19 is already 1 year old, so I guess it is already well-tested.
Note:
tests/test_examples.py
as I was not able torebuild_example_data
. Guess Benchmark downloading is broken #1965 is related.check_categorical=False
is the best way to fix the tests. But I think a better fix would require deeper work (i.e.: remove the use ofNone
/NaN
in categorical data, as Pandas already warns about its future deprecation).