On Thu, Sep 22, 2005 at 11:49:24AM -0500, Kevin Keith wrote:
> It is a CGI (shell) script that is calling a C program. I found the issue -
> which was external to Postgres.
So what was it? Even if it wasn't a PostgreSQL problem it might
be educational to learn what happened.
> I still have one remaining question - if the COPY to FILE command fails,
> does it return an error? If the program uses:
> int i;
> i = PQexec("copy from <<table_name>> to <<fully_qualified_file_name>>");
>
> and the table has (for example) 500 records, but only 350 were copied to
> the file, what is returned in i?
PQexec() returns a PGresult *, not an int. You can pass that PGresult *
to PQresultStatus() to find out the command's status; if the command
failed you can pass the PGresult * to PQresultErrorMessage() to get
a descriptive error message. See the libpq documentation for more
information.
On success, COPY returns a result status of PGRES_COMMAND_OK; I'd
expect it to return something else if it failed. For example, if
the disk is full then COPY fails with PGRES_FATAL_ERROR. If you
have a case where COPY reports success but writes an incomplete
file then please post the code -- that way we can see if the bug
is in your code or in PostgreSQL.
--
Michael Fuhr