updated thresholds/values to work with PROJ 9.2 #2163
Conversation
|
This probably needs a check on PROJ(pyproj) versions within the tests to pass the minimum versions CI as well. |
kirillrdy
force-pushed
the
main
branch
4 times, most recently
from
ce0151a
to
28cabb1
Compare
| assert_almost_equal(np.array(eqdc.y_limits), | ||
| (-10001965.729313632, 17558791.85156368), | ||
| decimal=7) | ||
| if pyproj.__proj_version__ >= '9.2.0': |
There was a problem hiding this comment.
What are your thoughts about changing the number of decimal places we compare to here? It seems like if we can compare to 2-3+ decimal places that should be good enough for Cartopy's purposes. We are really relying on proj for all the "official" transformations.
There was a problem hiding this comment.
I don't have any strong opinion about number of decimals. I do find it strange that cartopy supports different versions of proj that would lead to very different results (eg boundary of UK)
I would either not test values from proj at all, since they change between versions OR only support one version of proj that is considered "correct", most likely latest version of proj.
that's just my 2 cents no need to change anything for this PR
There was a problem hiding this comment.
I agree with everything you say here :)
kirillrdy
force-pushed
the
main
branch
2 times, most recently
from
54f7541
to
cee33ee
Compare
|
i see tests are failing, but passing in my setup, I have an other look |
|
should be fixed now |
| assert_almost_equal(np.array(eqdc.y_limits), | ||
| (-10001965.729313632, 17558791.85156368), | ||
| decimal=7) | ||
| if pyproj.__proj_version__ >= '9.2.0': |
There was a problem hiding this comment.
I agree with everything you say here :)
Rationale
fixes #2145
Implications
I cant speak for correctness of values
from release log https://proj.org/news.html#release-notes
it looks like EPSG and ESRI databases got updated so its probably related