fix: limit chardet < 7.0.0 and pandas <3.0.0#1205
fix: limit chardet < 7.0.0 and pandas <3.0.0#1205JGSweets wants to merge 6 commits intocapitalone:mainfrom
Conversation
| Seems more needs to be evaluated in chardet |
| looks like they default to windows-1252 more than utf-8 which is why the new error / change is occurring |
| @JGSweets is it possible to resolve the failed checks? |
I could alter the tests, but I don't feel like that is the ideal solution. looking at their newest version, it seems they changed behavior that results in files we originally would label as In addition, it did not seem like there was an easy way to chunk results into the detector and get a full result rather than a single candidate. |
| In addition, i think there's a larger issue now at play given |
| fixed tests similar to the previous pr which have approximate issues |
JGSweets changed the title | @shania-m should be good to go now! |
| @shania-m @ryanSoley any updates here? |
| self.assertAlmostEqual( | ||
| expected_diff["statistics"].get("t-test").get("welch").pop("p-value"), | ||
| profile_diff["statistics"].get("t-test").get("welch").pop("p-value"), | ||
| places=10, |
There was a problem hiding this comment.
QQ- I see some of these assertions are at 2 earlier in the file and others are at 10. Do you mind explaining the difference?
There was a problem hiding this comment.
Sure! The places indicate the precision of the almost equal, in this case 10 decimal places. So the intention here to be as close as possible despite floating point precision issues (possibly due to python versions).
I cannot say why the previous ones were only set to to two places.
There was a problem hiding this comment.
@shania-m if you have any other questions, lmk!
2 participants
this pr: