improve result fetching experience #593
Conversation
msokoloff1
commented
Jun 3, 2022
msokoloff1
commented
- Fetch the result file for the user
- Try to match AnnotationImport.statuses and .errors as close as possible.
msokoloff1
force-pushed
the
ms/fetch-result
branch
2 times, most recently
from
fab754d to
a645024
Compare
msokoloff1
force-pushed
the
ms/fetch-result
branch
from
5bfa80f to
e0e8265
Compare
| return data.get('error') | ||
| if self.status == "FAILED": | ||
| result = self._fetch_remote_json() | ||
| return result['error'] |
There was a problem hiding this comment.
I was thinking that this should never be None since we are specifically checking that the action is a Json import.
labelbox/schema/task.py
Outdated
| self.wait_till_done(timeout_seconds=600) | ||
| if self.status == "IN_PROGRESS": | ||
| raise ValueError( | ||
| "Job status still in `IN_PROGRESS`. The result is not available. Increase timeout or contact support." |
There was a problem hiding this comment.
Increase timeout or contact support. since this method is being called from errors and result but those don't accept a timeout param, increasing the timeout would be harder for a user. Should we include a timeout param in the other methods, or just ask them to contact support?
There was a problem hiding this comment.
Also since this is lru_cache, does that mean that this will cache the first result, and always return the same result for any subsequent calls while a script runs? Would it make more sense since we don't need to evict any items from the cache, to just use @cache instead?
https://docs.python.org/3/library/functools.html#functools.cache
There was a problem hiding this comment.
They can wait_till_done before calling this with a larger timeout. I'll update the message to make this clear. Also the cache does not save the result if an error is thrown.
