Closed pnadolny13 closed 1 year ago
Is this a bug in the SDK? I think in general we should bias towards case-insensitive comparisons.
These streams should all land into the same table without error and without duplication:
MyStream
Mystream
mystream
And these tables should all be considered valid destination tables for a stream named "MyStream":
MyStream
Mystream
mystream
We can (optionally!) let the target decide how/if to normalize casing, but regardless of casing preferences of the destination system, we should find and load to existing tables even if casing is not a match.
I'd bet on it being something on the postgres target specifically. I agree it should work regardless of casing but I do think the case should follow whatever was initially set, maybe I care about something that doesn't matter here though.
A lot of this comes to the normalization stuff I ripped out. Maybe just tolower and a replace dashes and underscores. That would be an easy fix here but wouldn't fix this generically
Originally raised in https://meltano.slack.com/archives/C013EKWA2Q1/p1680488722168999
I was able to replicate the behavior. I sent some dummy data to the target using an uppercase stream name
Account
, like this:I see the table created in the database but then immediately get this error:
cc @visch @aaronsteers @edgarrmondragon