Bob Pesner

Sorry for the delayed thought, but I should also mention that I wrote a c program to make the same calls directly to unixODBC and they all worked normally.

Mark, I should have mentioned that the tester.cmd program fails during the connect phase because the call to SQLGetInfo doesn't return a datatype list.

Mark, Thanks for looking at this. This, to me, is a weird one. After more extensive testing, I've discovered two bugs that each occur in two places in the Rexx/SQL code. Both places involve calls to unixODBC functions. In one pair, a local c variable that is being passed by value is being overwritten with 0. In the other pair, the result is an error in the PostgreSQL ODBC driver. All these points in code are in cli-odbc/rexxcli.c Here are the details (my environment is described below): On a call...

I ran into the same problem on Fedora 31 with rexxsql 26B3 trying to issue an sql insert statement against a PostgreSQL database using ODBC. I traced the problem to two lines in the SQLCOMMAND function in cli-odbc/rexxcli.c. After the statement provided to the function is executed, the function calls FreeArgv to free the statement at various lines. By the time it gets to the call to FreeArgv at line 2826, the statement is already null, so I added an if to skip the call if stmt is null. And a few...