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
I've just put in place two work-arounds due to failing coordinate transform tests.
The first is in odc.algo for choosing a Transform Path, which is choosing a different transform depending on something like the GDAL/Proj version.
The other is in odc.apps.dc_tools during the STAC transform. The code is for transform Geographic coordinates into a Projected Coordinate system, and (IMO choosing a bad rounding value), which is resulting in off by 1 values in the transform. I've wrapped these coordinate values in pytest.approx with an epsilon of 1.
I'm not happy with either of these, and the root causes should be addressed and understood.
I've just put in place two work-arounds due to failing coordinate transform tests.
The first is in
odc.algo
for choosing a Transform Path, which is choosing a different transform depending on something like the GDAL/Proj version.The other is in
odc.apps.dc_tools
during the STAC transform. The code is for transform Geographic coordinates into a Projected Coordinate system, and (IMO choosing a bad rounding value), which is resulting in off by 1 values in the transform. I've wrapped these coordinate values inpytest.approx
with an epsilon of 1.I'm not happy with either of these, and the root causes should be addressed and understood.
From #524
I think the STAC geometry transform is from #93, which did include some improvements, but I don't think addressed all the potential problems.
The text was updated successfully, but these errors were encountered: