OK. Writing a dataflow to read a CSV file. Connection Manager Editor says format is delimited, text qualifer is <none>. And that's just fine, but when run, a bunch of rows were generating errors. Why?
Well, turns out one of the fields contained values sometimes ending with a percent sign, "31%" adjacent to our delimiter - which fwiw is actually a pipe "|" not a comma. Well, it seems like the percent was acting like an escape character. Is
that correct? Eliminating the percent fixed the import. Changing to %% fixed the import.
And, it seemed that changing the field specification (in the connection manager Advanced tab) to "text qualified = False" also fixed it. But - why did it think that the field was text
qualified, when at the connection (file) level, we told it "no text
qualifier"? Well, qualifie*r* is not the same as qualifie*d*, right? Ugh. Is this a bug? (2008 R2). Is it a feature??? Should I make it a point in all flat file CSV connections
to override all fields to "text qualified = False" - would that have any other side-effects?
View Complete Post