[24.0] Raise appropriate exception if user forces a collection that is not populated with elements as input#18023
Conversation
|
also xref #17560 where I triggered this exception, by doing exactly that. (I am impatient waiting for the collection to upload, I "assume" (as a user) that Galaxy knows when the collection will be ready and won't start the job until the collection is ready, like it does with other datasets, and just wanted to queue my job so I could go back to my other tasks.) |
|
I think the natural solution there is #17393, where the request itself is persisted, until then no way around it. At a minimum you'll get a message saying we can't do anything. I'm looking to make fetch data collection uploads a bit smarter so that we create the collection in a usable state if you're not building a collection from an archive where we really have no handle on what the collection is going to look like. |
|
Also note that #17560 is a different in that your collection is populated, but we can't filter on metadata if we don't have metadata yet. If your collection is not populated it has no leaf elements. |
It was using the wrong id type, and since we don't prepopulate the collection we need to wait until discovery happens. It actually nicely illustrates the bug.
0e6c669 to
a671090
Compare
These aren't valid inputs and I think the only way to get here is by dragging such a collection into an input field, which bypasses all the validations happening in the dataset collection matcher.
Minor performance fix included for nested collections in said dataset collection matcher.
Fixes #18012
How to test the changes?
(Select all options that apply)
License