It looks like Redshift is looking for a Vec<i64> instead of a Vec<i32> for the parameter.

Even though the column is a integer which according to the docs is
INTEGER, INT, or INT4 ?

Oh wait, I was responding to the initial error of Err(Error(Conversion(WrongType(Type(Int8Array))))). What changed to go from that error to the "invalid array flags" one?

I had the column as biginteger so I changed it to integer which it should have been. Sorry I should have maybe created another issue.

The Redshift docs claim that arrays aren't supported at all, but I'm not sure why a query using them would fail in that way: https://docs.aws.amazon.com/redshift/latest/dg/c_unsupported-postgresql-datatypes.html.

That's weird as I can issue
SELECT id, name FROM sensors WHERE id = ANY(ARRAY[596304,597992,610978,597998]) from my database ide and it works fine.

Also I am using ANY here as I don't think the IN clause works with rust-postgres ?

It works just fine - it just takes a sequence of separate values rather than an array of values.

One possibility is that Redshift is forked off of Postgres 8.1 or older, where they didn't support null array elements (support was added in 2005!). We should be able to make this work by only setting the null flag if there are actually null values.

I have tried the IN clause now as well. (hacked as a single str)

ie
let results : postgres::Result<Rows> = conn.query("SELECT id, name FROM sensors WHERE id IN (596304,597992,610978,597998)", &[]);

From rust-postgres I get the same error of invalid array flags

If that exact query generates that error, then that seems like a bug on the Redshift side of things. We're not producing any arrays in that case, we're just sending a string to Redshift.

Odd. If I copy and paste that query into my ide (datagrip) it works fine.
I must be doing something silly in the code then.

Out of curiosity is there away to pass an array to a sequence so I can do something like

conn.query("SELECT id, name FROM sensors WHERE id IN ($1)", &[596304,597992,610978,597998]);

Nope - #326 (comment)

Ok I got the In clause working.

So to recap in redshift

SELECT id, name FROM sensors WHERE id = ANY(ARRAY[596304,597992,610978,597998])
fails
but

let results : postgres::Result<Rows> = conn.query("SELECT id, name FROM sensors WHERE id IN (596304,597992,610978,597998)", &[]);

works.

Thanks

Wait, so even ARRAY[...] literals in the query string are broken with Redshift? I though it was just transferring array values as parameters that didn't work.