[Info-ingres] DBD IngresII.pm weirdness

[Chris Clark]

On Thursday, January 9, 2020 at 7:08:08 AM UTC-8, Martin Bowes wrote:
....
> Now the Perl program has a segmentation fault. It occurs at step 3 when it attempts to connect. Tracing shows that this at line 103 of the DBD/IngresII.pm where it executes:
....
> 
> If I move the dbmopen command to be the first of these three steps then all is OK and the program works correctly.
> 

It sounds like memory is getting trampled somewhere. Either in the IngresII or the dbm module. There are a bunch of different things you can try as diagnostics, with varying degrees of effort.

You could try using valgrind or similar (Dr Memory - https://drmemory.org/) and figure out whats going on (or even debugger).

Alternatives would be to try a different driver, e.g. DBD::ODBC (you may not need to build that as its often available pre-built on some platforms). You can build it with Ingres CLI if you want to skip the UnixODBC dependency. When I need to use Perl and Ingres, I use ODBC.

This is a shot in the dark; you could look into trying II_APP_NO_SIG_HANDLE - its possible that's kicking in https://docs.actian.com/actianx/11.1/index.html#page/SysAdmin/II_APP_NO_SIG_HANDLE_(Linux).htm