Re: Memory faults while freeing OCCI objects

No Such Luck wrote:
> Dave wrote:
>

>>No Such Luck wrote:

>>Could be a programming error, for instance not having stuff between
>>createEnvironment and terminateEnvironment between braces.

OCCI 10.1 docs, Chapter 2, under Creating and Terminating an Environment.

The braces look out of place, but the reason for them is that anything created within the environment MUST be destroyed BEFORE the environment, and the order of deletion within the same scope is not guaranteed by the compiler.

The problem is that not using braces works for a while, then goes on to produce weird errors after the application has been growing for a while, thus suggesting this isn't the source of the problem.

> Thanks for the help. I've narrowed fown the offensive call to
> conn->terminateStatement (stmt); where the stmt was specifically a PK
> trigger auto ID query from dual. When I comment out the calls to free
> these stmts, the app runs fine.
>
> It may yet be a programmer error, but this specific condition leads me
> to believe it's still within OCCI somewhere.

It's possible the problem is within OCCI. From the problem description it sounds like you should be able to knock together a testcase of less than 20 lines based on the chapter 2 code which could be suitable for creating a TAR if you have access to Oracle Support (if not, bung it up here anyway and I'll have a look if I have time). Also a short SQL script that creates any necessary database objects would be useful to avoid any ambiguity in what is needed (iow: I haven't got a clue what you mean by "PK trigger auto ID query", and the SQL would be shorter and more useful than a detailed explanation :-)

Both testcase and SQL should be absolutely the minimal necessary to reproduce the problem, with EVERYTHING non-essential to the problem removed. Received on Tue Oct 18 2005 - 07:05:30 CDT