Open griffbad opened 5 months ago
Thanks for bringing this to my attention!
The whole extra_data
field needs to be reworked. We are moving away from using .attrs
and instead just providing extra data in "non-spatial" arrays that are still managed by requests. This way there's no unexplained data being passed around.
In the mean time until .attrs
is completely removed, I think this seems like a reasonable patch for the current behavior. I would only make sure that the value stored in .attrs
remains a dictionary instead of being turned into a list as it is done here. Something like just assuming the set of keys in each batch do not overlap and then turning them into a larger dict with a union across the keys.
This PR will be closed in one week due to inactivity. Please update the PR if necessary.
This PR will be closed in one week due to inactivity. Please update the PR if necessary.
This PR will be closed in one week due to inactivity. Please update the PR if necessary.
This PR will be closed in one week due to inactivity. Please update the PR if necessary.
This PR will be closed in one week due to inactivity. Please update the PR if necessary.
This PR will be closed in one week due to inactivity. Please update the PR if necessary.
This PR will be closed in one week due to inactivity. Please update the PR if necessary.
took me a while to figure out why my attrs data (initially found from using extra_data in SpecifiedLocation) was not preserved at end of pipeline. This was it.
Pre-Merge Checklist:
vX.Y-dev
)patch-X.Y.Z
)