Netscape NSAPI/Oracle Pro*C on Enterprise 2.0

Netscape NSAPI/Oracle Pro*C on Enterprise 2.0

Post by Tim Devi » Sat, 01 Feb 1997 04:00:00

I have an existing application written in Pro*C which interfaces to the
Netscape server through their API.  The code works flawlessly with the
Netscape Commerce Server version 1.12.  We just installed The Enterprise 2.0
Commerce Server (not 2.01) and have noticed an odd problem with part of the
application.  There is a function that is called that does not consistently
work.  I put a print in right before the function is called and it almost
always works the first execution.  On subsequent exections there a 50/50
chance that it actually makes it into the body of the function.  I put a print
in as the first line and sometimes it gets into the function and sometimes it

If I code a print and then a return as the first and second line of the
function to bypass the execution of the rest of the code it still exhibits the
problem.  If I turn the optimizer on (drops the dead code), it accesses the
function properly each time.  (But I have to turn the optimizer on.)

What makes me think it's a thread issue is that if I set the min/max server
threads to 1, the code works flawlessly every time.  Which it also does using
the 1.12 version.

When I compile the app using the thread safe librairies it doesn't even get
into my shared object, the server just hangs.

Has anyone ever run into this, or is there something I could investigate that
might point out the problem.

Any help would be greatly appreciated.

Tim Devine


1. NSAPI, Netscape Enterprise server question...


I've been doing a project for NS ES 3.x.  This basically entails creating a
service that works like Apache's mod_usertrack.  This is a fairly simple thing
to do, or so I thought.  Basically, I think I'm having a caching problem.  I can
(as of this moment) get the system to set a cookie even when the browser has
cached the page as long as it is a new session, ie the browser was just started
up.  Then, one goes clicking along and as long as the request hasn't been cached
everything logs fine.  But if you type in the URL of a page after the cookie has
been set, and after you have already visited the page, no logging occurs at all,
even in the main access log.  I thought that under such circumstances, the
server was supposed to send a "304".  It does that if you hit the reload button
(but it logs that just fine).  It's when you click on, or type in the URL of, an
already loaded/cached page that this behavior occurs.  Basically I've got two
basic functions within the source:
the first looks for/sets the cookie (and attempts to play with the server
headers) the second is an addlog function that logs the result when there is a
cookie.  It seems that under certain conditions, even setting pragma to nocache
or cache-control to nocache fails to do what I'd like it to do... behave like

I load the routines with and Init directive.

the "spot_cookie" routine is a Service directive that is in the "default object"
(the first service directive in fact).

the logger is an AddLog directive, also in the default object.

Does anyone out there have any suggestions?

J. Petrakis

2. Strange IRQs in /proc/interrupts

3. Netscape Enterprise Server, NSAPI and NIS

4. A simple question about hardware

5. Netscape Enterprise Server -- NSAPI problems

6. RedHat can't see my nic!?!

7. Netscape Enterprise Server - Oracle

8. can't print, shared libraries eror

9. Netscape Enterprise Server, Livewire and Oracle

10. Oracle WRB + Netscape Enterprise server with chroot

11. Integrate Solaris ver. of Oracle and Netscape Enterprise Server on 2 Sun Machines

12. User-Account on Netscape Enterprise Server 2.0

13. Netscape Enterprise 2.0 and content negotiation