Ok, I have a small dilemma, and I'm hoping someone here has had to do this before.
I have at a site (not PARI) a server running some mission critical software that was written in 1997 for libc5, under AOLserver 2.3. No, source code is not available for that version of AOLserver (it wasn't open-sourced until version 3.0, and the API changed rather dramatically at that point), not that it would help any, as gcc has undergone so many changes since RH5 that a recompile would be very difficult. The software is currently running on Redhat 5.2 (don't laugh; it's got a 2,716-day uptime right now! (yes, that's over 7 years!)) with the libc5 libs loaded. The server hardware is just about ready to fly apart, and I need to see about running this on more modern hardware. Of course, they already have bought the replacement server, a Dell PowerEdge 1850. No way RH 5.2 is going to load on a PE1850 with the megaraid PERC 4e/Si.
Now, I've thought about a couple of possibilities: 1.) QEMU or VMware or similar running a RH5.2 instance under CentOS 4; 2.) Running a libc5 setup on CentOS 4. I'd rather do this, as security fixes would at least be available.
Going back to a RH5 machine and working was a real trip; had to use telnet (no ssh on RH5); Kernel 2.0 is still there, no yum, no chkconfig/service subsystem, etc.
The custom code is written specifically for AOLserver 2.3 in its tcl dialect and using API functions that don't exist in AOLserver 3.0 or later (or it would be a snap to compile AOLserver 4.0.10 on CentOS4 and roll); there's a couple hundred thousand lines of tcl code here, and while they will be contracting a replacement webapp using Plone/Zope, they aren't ready to transition as yet. -- Lamar Owen Director of Information Technology Pisgah Astronomical Research Institute 1 PARI Drive Rosman, NC 28772 828-862-5554 www.pari.edu