BUG #16149: Prepared COPY queries always report 0 parameters when described - Mailing list pgsql-bugs

From PG Bug reporting form
Subject BUG #16149: Prepared COPY queries always report 0 parameters when described
Date
Msg-id 16149-531659f1ce965744@postgresql.org
Whole thread Raw
Responses Re: BUG #16149: Prepared COPY queries always report 0 parameters when described
List pgsql-bugs
The following bug has been logged on the website:

Bug reference:      16149
Logged by:          Steven Fackler
Email address:      sfackler@gmail.com
PostgreSQL version: 12.1
Operating system:   Debian Buster
Description:

When a Postgres backend describes a prepared `COPY ... TO STDOUT query`, it
always reports 0 query parameters regardless of how many are actually
present. Here's a simple example program that demonstrates the issue:

```
#include <libpq-fe.h>
#include <assert.h>
#include <stdio.h>

int main() {
    PGconn *conn = PQconnectdb("host=localhost port=5433 user=postgres");
    assert(PQstatus(conn) == CONNECTION_OK);

    PGresult *result = PQprepare(conn, "a", "COPY (SELECT $1::TEXT) TO STDOUT",
0, NULL);
    assert(PQresultStatus(result) == PGRES_COMMAND_OK);

    result = PQdescribePrepared(conn, "a");
    assert(PQresultStatus(result) == PGRES_COMMAND_OK);
    printf("nparams: %d\n", PQnparams(result));

    result = PQexecPrepared(conn, "a", 0, NULL, NULL, NULL, 0);
    assert(PQresultStatus(result) == PGRES_FATAL_ERROR);
    printf("error: %s\n", PQresultErrorField(result,
PG_DIAG_MESSAGE_PRIMARY));
```

When run, it prints the following:

```
nparams: 0
error: there is no parameter $1
```

If you change the query to just the inner `SELECT $1::TEXT`, the number of
parameters is correctly reported, but interestingly the error message
changes:

```
nparams: 1
error: bind message supplies 0 parameters, but prepared statement "a"
requires 1
```

From some quick Googling, I did see this StackOverflow post[1] stating that
COPY queries don't support parameters, but if that's the case it seems like
an error should be reported at the preparation stage. I also don't see
anything about that in the documentation for COPY[2], though I may have
missed it! I see the same behavior on Postgres 11.1 as well, if that's
relevant.

[1]: https://stackoverflow.com/a/22963085
[2]: https://www.postgresql.org/docs/12/sql-copy.html


pgsql-bugs by date:

Previous
From: Jeff Janes
Date:
Subject: Re: BUG #16148: Query on Large table hangs in ETL flows and gives outof memory when run in pgAdmin4
Next
From: Justin Pryzby
Date:
Subject: Re: BUG #16111: Unexpected query compilation error “negative bitmapset member not allowed”