search

COSIMA / cosima-cookbook

Framework for indexing and querying ocean-sea ice model output.
https://cosima-recipes.readthedocs.io/en/latest/
Apache License 2.0
58 stars 25 forks source link

Explorer not working with analysis3-23.01 #321

Open aekiss opened 1 year ago

aekiss commented 1 year ago

With analysis3-23.01, when I try

import cosima_cookbook as cc
session = cc.database.create_session()
dbx = cc.explore.DatabaseExplorer(session=session)
dbx

it fails with

Error displaying widget: model not found

but it works fine with analysis3-22.10. This is with gadi_jupyter, if that matters.

angus-g commented 1 year ago

I don't seem to be able to reproduce this (on ARE)...

schmidt-christina commented 1 year ago

With analysis3.23.04 and analysis3.23.07 it displays the widget but if I click on a experiment it doesn't display the variables or not the up-to-date versions. For example panant-005-zstar-ACCESSyr2 doesn't show any variables but I know they exist, the permissions are correct and I can load them with cc.querying.getvar() . For panant-01-zstar-ACCESSyr2 it shows variables but not all of them (salt_flux is missing).

@aidanheerdegen

aidanheerdegen commented 1 year ago

Yes I noticed some inconsistencies had crept in. For me when I used DatabaseExplorer the keyword and variable tabs weren't tabs, they were just the words jammed together without the window decoration. When I resized the Jupyter output cell (collapsed and uncollapsed) it suddenly rendered ok, and since then it has rendered ok because the output cell always renders with a scrollbar. I guess it remembered to render is this way.

I also noticed the ExperimentExplorer doesn't show the variables correctly by default. However selecting a model shows them all, and then they show up fine when those selectors are changed (like hiding coordinates).

A missing variable is a bit more problematic @schmidt-christina. I might have a chance to check that, but can't guarantee it.