How to avoid hitting PREPARE limit while using DESCRIBE OUTPUT query?

[vspinu]

Problem Statement

For moderately complex PREPARE queries I am seeingtrino.exceptions.Http502Error: error 502: bad gateway:

This is how the query is run as part of the ibis project:

            cur.execute(f"PREPARE {name} FROM {query}")
            try:
                cur.execute(f"DESCRIBE OUTPUT {name}")
                info = cur.fetchall()
            finally:
                cur.execute(f"DEALLOCATE PREPARE {name}")

Searching for similar issues on trino and trino-client side it turns out that it's a known problem caused by the header size limit of 4KB

• server issues (1, 2) suggests a partial solution by altering the http-server limits:

http-server.max-request-header-size=64kB
http-server.max-response-header-size=64kB

I cannot find an option to configure the header size on the client side. (In fact I don't know if this is the problem caused by the limit on the client or server side)

• client issue Disable prepared statements by default trino-python-client#306 suggests
  • using legacy_prepared_statements=False (which does not solve the issue for me)
  • EXECUTE IMMEDIATE which does not work with DESCRIBE OUTPUT as the latter requires a prepared named statement.

Question

What is the current recommended solution for this problem (ideally on the client side). Particularly, is there a way to execute DESCRIBE OUTPUT immediately without going through the PREPARE step?

Thanks!