DM-40120: Add without_datastore flag to Butler #867
Conversation
This can be used if you know you are only querying a registry and not interested in whether the datastore is usable or not.
Optional means that the user will be uncertaint whether they have a datastore or not. Skip means that the user is guaranteed that datastore will not work.
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## main #867 +/- ##
==========================================
+ Coverage 87.72% 87.76% +0.03%
==========================================
Files 274 274
Lines 35937 36057 +120
Branches 7537 7562 +25
==========================================
+ Hits 31526 31645 +119
Misses 3241 3241
- Partials 1170 1171 +1
☔ View full report in Codecov by Sentry. |
| ): | ||
| # Name ourselves with the timestamp the datastore | ||
| # was created. | ||
| self.name = f"{type(self).__name__}@{time.time()}" |
There was a problem hiding this comment.
Why do we need the timestamp? Two NullDatastore instances should be identical (it could even be a singleton, though I don't think it needs to be), and I'd hope that means we don't need to care about conflicts.
There was a problem hiding this comment.
Good point, although since this datastore should never result in a dataset being stored anywhere there's no requirement for it to be a fixed name. The time is really there to let you know when you created that butler and it will appear in stringified output of Butler. Whether anyone cares is a question.
timj
changed the title
This makes it match the other datastores when the dataset is not present.
This allows
butler = Butler(uri, skip_datastore=True)to return a Butler instance that has no connection to a real datastore. I've also modified script implementations to add the flag where appropriate.Checklist
doc/changes