Re: stability of pro/c embedded sql - developers experiences?

mrwizard_at_attcanada.ca (David Perinbam) wrote in message news:<b1b020ed.0107311754.57810319_at_posting.google.com>...
> I having been working some pro/c programs in which I have been finding
> problems that seem to be in Oracle's generated code for embedded sql
> using pro c (8.1.6).
> The programs are mid size (4000 lines + with numerous sources) and I
> have been finding that,intermittently I will get a weird segmentation
> faults for a particular input file (but not a data problem) . The
> cases have all occurred fairly deep within Oracle's generated code for
> embedded sql (in code that I can't see). For example, in one case for
> a particular file the RELEASE part of an EXEC SQL COMMIT RELEASE
> generated a seg. fault. It was the very last thing I did and every
> thing else worked ok. I showed it to Oracle and the saw no problem wih
> the code.
>
> I'm just wondering if anyone else has had stability problems with mid
> to large size programs using pro c. The examples I have are a bit to
> large to post here so I'm just curious as to others experiences. Most
> of our code using pro c is much smaller than this and works ok. I'm
> wandering if there's some scalability problems with pro c , over
> reliance on global variables in the generated code or is this all just
> me. My background is really ansi c systems programming - I'm not an
> Oracle expert. Also, all my sql's were all very, very simple but the
> errors would occur in the sqlcxt or in the generated OCI calls.
>
> Anybody had similar experiences with pro/c ?

I've used Pro*C pretty extensively for 15 years. I've also had to help lots of developers with their programs. Yes, it sometimes crashes in the Oracle code, but I only remember 2 or 3 instances in 15 years. It should be treated as an Oracle bug. That is, the DBA should submit it to Oracle along with the alert.log. If you crashed, there really should be something in the alert.log. At a mininum, you can submit the stack dump, and Oracle should be able to find/fix it. Received on Wed Aug 01 2001 - 09:51:32 CEST