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
In the alpha implementation, each source data of the input list (of source data) is attempted to be added and commited to the DB separately. When the repository receives a list of source data with duplicates, it will push only the first one and fail for all the rest. See TODO note in the repository function. The solution might be somehow repristining the sqla session (although trying to session.close() then with self.session.begin() will also fail with a different error).
Overall, a better solution would be managing duplicates somehow first and then commiting in batch, instead of one by one.
Also, the API test is very basic, could cover more cases for the beta.
Also, by default, the source data gets registered as status = "AVAILABLE", but we need a better way of managing this for the beta.
The text was updated successfully, but these errors were encountered:
alebg
changed the title
KnowledgeSourceRepository: new_source_data: fix "halferror" case for duplicated lfns
KnowledgeSourceAPI: new_source_data: fix "halferror" case for duplicated lfns in repository; improve API tests (more)
Nov 19, 2023
alebg
changed the title
KnowledgeSourceAPI: new_source_data: fix "halferror" case for duplicated lfns in repository; improve API tests (more)
KnowledgeSourceAPI: new_source_data: fix "halferror" case for duplicated lfns in repository; improve API tests (more); manage source data status
Nov 19, 2023
In the alpha implementation, each source data of the input list (of source data) is attempted to be added and commited to the DB separately. When the repository receives a list of source data with duplicates, it will push only the first one and fail for all the rest. See TODO note in the repository function. The solution might be somehow repristining the sqla session (although trying to session.close() then with self.session.begin() will also fail with a different error).
Overall, a better solution would be managing duplicates somehow first and then commiting in batch, instead of one by one.
Also, the API test is very basic, could cover more cases for the beta.
Also, by default, the source data gets registered as status = "AVAILABLE", but we need a better way of managing this for the beta.
The text was updated successfully, but these errors were encountered: