John Chilton

Thanks for the rebase! Pushing the milestone but hopefully we can make some substantial progress this next release cycle.

One sticking point at a time - I think User. _calculate_or_set_disk_usage is going to be a problem here right? Like as soon as the user's disk usage is recalculated -...

If we're just adopting Pulsar states outright in Galaxy - I'm not sure the benefit of renaming them. I trust you though @natefoo 👍

Ugh... I'm sorry to throw you under the bus but ``change_state`` is so expensive and this adds new ones for states that will only exist for a very brief period...

> An SQL trigger maybe? Better than a Python loop (assuming we don't need to reload all the dataset association data into sql alchemy anyway...) but we still would have...

What if every dataset association had a direct link to the job that specifies its state? Our in-python ``creating_job`` logic is definitely not going to cut it.

> although this duplicates the job_to_output_* tables It is encoding slightly different information (in lieu of stuff like copied datasets, etc..) - so I wouldn't really call it duplicating those...

```diff diff --git a/lib/galaxy_test/base/data/test_workflow_batch.ga b/lib/galaxy_test/base/data/ index 32de906b38..8a4792542f 100644 --- a/lib/galaxy_test/base/data/test_workflow_batch.ga +++ b/lib/galaxy_test/base/data/test_workflow_batch.ga @@ -57,7 +57,7 @@ "tool_errors": null, "tool_id": "addValue", "tool_state": "{\"__page__\": 0, \"__rerun_remap_job_id__\": null, \"__job_re - "tool_version": "1.0.0", +...

I've done so much work to preserve those buggy things. I like this idea a lot. I wonder what the deal is is with the ``|__identifier`` attributes - any chance...

Moving to 21.09 also 😅, please ping again though if it wrapped up sooner. This is good and important work.