barrier mechanism.

have made any progress yet.

before doing the loghole, so that currently running programs get a chance
to finish up and close their logfiles.

 * Define vfork to fork. Why? Well starting at 6.X FreeBSD switched
 * its underlying pthread impl, and popen is broken in threaded apps,
 * since it uses vfork. I have no understanding of any of this, only
 * that avoiding vfork solves the problem.  We can back this change
 * out once we figure out a real solution.
 */

ssh-mime.pl to indirect telnet to the IXP board: return "ixpN-gw.emulab.net"
name as the gateway.  You can then convert that to an IP and subtract one
to get the IP to telnet to from the IXP host.

putting back the leading quote rather than removing the trailing quote.

too.

than vnconfig).

Fix up emulab-boss and emulab-ops port Makefiles to work with FreeBSD 6
and also to look for it instead of 5 when deciding what to do.

It is amazing the amount of stuff you have to do, even if you are not going
to do anything!

node.tcl: Fix the quoting of the -dir argument for the program agent
(yeah, I know what you are thinking; but I am just mimicing what is
already done in sim.tcl::make_event)

names, ends up with things not being ordered as I want them to be.
So, reorder things.

It shouldn't matter in this file (only tb_compat.tcl), but just in case...

but now perl is unhappy.

users in the emulab-ops project (elabman, elabckup, operator) all with
the same email address. elabman is treated special (since it was the
only special user way back), but now there are lots of places that
either need to include tests for all special users, or we need to flag
these users in the users table.

For now I made a change to genelists.in to get past the problem. I
will address the user table at a later date.

Switch to using gtar for building the plab rootball;  apparently the
--numeric-owner flag disappeared from the FreeBSD version of tar.