Do you have a small script that can reproduce this problem?

use Socket;
:
:
:  module stuff not shown for clarity

my $name=gethostbyaddr(inet_aton($self->{ip}), AF_INET); # Lookup host
+name
[download]

Let me rephrase. Can you supply a *complete* script (preferably one that does not rely on non-core modules), and show the output of it failing? Ie there are many things that could trigger this problem, and I'd like to see as many of them as possible ruled out.

Sigh.

I didn't ask for your complete existing script. I was asking for a cut-down version of it that is the minimum that reproduces the problem; then I might be able to reproduce the problem here. And if its a problem with Perl, I could get to fixed for 5.8.5. Just for the record, I think its highly unlikely to be a GCC bug.

My reference to non-core modules meant that it would be great if the sample script only relied on core modules (such as Socket), but didn't rely on non-core modules which I would have to separately downlaod and install to be able to reproduce the problem here.

Can you supply a *complete* script (preferably one that does not rely on non-core modules)

The script in question is over 450 lines long and also has stuff in it that is NDA. Actually, more accurately it is a module being called by my script. If I totalled the calling script, this module and the other modules the script calls I think we are more in the ball park of 1200 lines.

I think I've done a fairly good job of describing the problem and I have a couple of good replies. All this without posting all that code.

Non-core modules? AFAICR Socket.pm is part of the base distribution of Perl. I guess it could be argued it is "non-core" since you have to use it in order to invoke it.

But wait: my OP was about the Socket module... so why are you asking me to post code that doesn't include it when that was part of the problem?

When I post to this site I normally post the amount of information I need to describe the issue I'm trying to solve and not bombard people with stuff to wade through that has nothing to do with the problem.

The script in question does not give the warning I cited in my OP when the gethostbyname call and use Socket; are removed. Put them back and the problem reoccurs.

It is just that straightforward.

The moral (I think) to what I'm seeing is I now have the ammuntion I need to "encourage" my client to upgrade his gcc to something newer than what is on there now.