We've been hitting bunches of errors with the codecov ci step, obscuring the results of the tests which precede it.
The coverage analysis - and particularly, its upload to codecov.io - runs with every permutation of Python and Pydantic version.
It's a little unfortunate since we recently added plenty of code along the lines of "if Python > 3.9 do this" and "if pydantic v1 do that". But given the failures, think we're better off just running that analysis for one combination. Probably the "latest known working" versions - Python 3.11 and Pydantic 2.6+.
This would look very similar to the black/ formatting ci step.
dan-fritchman
commented
1 week ago
We've been hitting bunches of errors with the codecov ci step, obscuring the results of the tests which precede it.
The coverage analysis - and particularly, its upload to codecov.io - runs with every permutation of Python and Pydantic version.
It's a little unfortunate since we recently added plenty of code along the lines of "if Python > 3.9 do this" and "if pydantic v1 do that". But given the failures, think we're better off just running that analysis for one combination. Probably the "latest known working" versions - Python 3.11 and Pydantic 2.6+.
This would look very similar to the black/ formatting ci step.