Details
-
Bug
-
Resolution: Unresolved
-
Major
-
Goldfish GA
-
Untriaged
-
0
-
Unknown
-
Analytics Sprint 40, Analytics Sprint 41, Analytics Sprint 42
Description
I'm attaching a file containing 10 airline documents, each of which is supplemented with a field called `myKey` which contains a UUID.
Copy that file to s3. I used the bucket `goldfish-test-data`.
Create a link called `g3` to the bucket containing the file.
Create three standalone collections:
- CREATE COLLECTION `Default`.`Default`.`pk_string` PRIMARY KEY (myKey: string);
- CREATE COLLECTION `Default`.`Default`.`pk_uuid` PRIMARY KEY (myKey: uuid);
- CREATE COLLECTION `Default`.`Default`.`pk_auto` PRIMARY KEY (myKey: uuid) autogenerate;
Try to copy the above file to each of the standalone collections:
copy into pk_string from `goldfish-test-data` at g3 path "" with
{ "format": "json", "include": ["autogenerate.json"]};
copy into pk_uuid from `goldfish-test-data` at g3 path "" with
{ "format": "json", "include": ["autogenerate.json"]};
copy into pk_auto from `goldfish-test-data` at g3 path "" with
{ "format": "json", "include": ["autogenerate.json"]};
The first copy works correctly, all 10 documents are loaded.
The second copy results in only 1 document in the pk_uuid collection, and the uuid from the `myKey` field is rewritten to a random value.
The third copy results in the following error message:
{ "code": 23069, "msg": "Field type string cannot be promoted to type uuid" }
The second copy should work, but doesn't. This is clearly a bug. There should be at least 10 documents added. In addition, it is an open question how to handle data which already has values in the primary key field.
I believe that the third copy should work. At least give me an option to indicate that it should try to promote a string value from a JSON file to a uuid.