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
Part of switching to celery tasks for this will mean the API endpoint will have to return different things. We probably want something truly asynchronous and that doesn't depend on all output dataset objects existing, jobs created, etc...
I'm not sure what that looks like exactly yet, but if we had an async, task-driven endpoint for this I think it would be more feasible to add functionality like passing in S3 URIs, DRS URIs (#11819), etc... during tool submission and resolving them immediately before tool submission via task chaining. We could do that first in the celery task worker but ultimately plan to stash them in the job definition to run on the worker (#10873). Designing this API though would let us decompose this task into pieces and write a GUI that can be agnostic about where and when URIs are resolved.
This is related to what is outlined in #11787 toward the end.
The text was updated successfully, but these errors were encountered:
Part of switching to celery tasks for this will mean the API endpoint will have to return different things. We probably want something truly asynchronous and that doesn't depend on all output dataset objects existing, jobs created, etc...
I'm not sure what that looks like exactly yet, but if we had an async, task-driven endpoint for this I think it would be more feasible to add functionality like passing in S3 URIs, DRS URIs (#11819), etc... during tool submission and resolving them immediately before tool submission via task chaining. We could do that first in the celery task worker but ultimately plan to stash them in the job definition to run on the worker (#10873). Designing this API though would let us decompose this task into pieces and write a GUI that can be agnostic about where and when URIs are resolved.
This is related to what is outlined in #11787 toward the end.
The text was updated successfully, but these errors were encountered: