Remove "delated" sql statement modifiers from all queries for now as delayed
insertion/replacement is apparently broken in our version of mysqld (5.0.20).
The problems are supposedly fixed in v 5.0.23, so we may be able to revert this
change.

Technically this binary should be installed on the FS server and we should
have an "fsdir" link in case fs/ops are different.  But for now...

where it can be used by rc.route, ixpboot, and, eventually, rc.simulator.

I can afford to give it.  It is in "if (0)".

most of what is in the node_types table (and node_types_auxtypes).
The goal is to make most of the type information free form key/value
pairs so that type information is more extensible.

This commit simply creates the table and populates it with initial
info using init_nodeattrs.pl. I expect this script to change, but it
can be rerun later so no big deal.

Also updated the precompiled jar.

maps, changing building floors, etc.  Also revamped the applet to work
with all the new stuff.

Updated plinabox template to use new PLC image (PLAB-PLC).

Note: I can't figure out how to get python to let me put linebreaks
into if conditionals. So, there's one nast long line right now - if
someone who knows python better could fix this, great!

Many minor fixes. Still not quite functional. TODO: Move the rest of the networking code from stubd. Move measurement code from stubd.

template show page ...

where no failed nodes for that pc type, it will now display "2 xxx's
... successfully ...".

Add "no warnings 'uninitialized'" inside loop since I want undefined
to mean zero.  Revert previous (incomplete) fix by checking if it
is defined first.

- when a site has no available nodes or the bestnode. changes,
  measurements to this node are stopped from all others.

- tests are started to a new node chosen for a site.

- before a site's bestnode is selected, automanage verifies bgmon is
  running correctly on the node

- pseudo-balanced latency initialization is done using randomly
  selected path directions between a node pair.

- other misc. stuff.

Added distributed start times during an INIT to spread out the scheduled test times over half the desired period.

Added error codes for determining "down paths" using a bw value of -1 or -3 to indicate a timeout or "unknown" (such as bad host name, no socket, etc..), respectively.

Some other miscellaneous changes.

Quick fix for a race condition in the auto-framework. This makes the race condition happen less, but it still happens. The real solution is to have the stub daemon-ize itself just before it is ready to accept connections.

Various bug fixes. Especially with the system for determining whether a connection is sending as fast as it can. Now it correctly determines when it is in slow start, and when the receivers buffer is empty so that these do not cause erroneously small bandwidth/max_delay figures.

Basic skeleton. Not functional yet. Getting this working and then adding some modules is what comes next.

between multiple link characteristics which are defined in the data file.
As a result, any type of link characteristic for the mote, wifi, or SDR
testbeds can be shown as long as it's in the file.  Also partial support
for metacharacteristics (things like the min/max/stddev of a single link's
primary characteristic).

Basically, a nice refactor into what I hope is complete flexibility
for display of any kind of link characteristic on any of our wireless
testbeds.

Fix HTML "character entity" references like &amp missing the trailing semicolon.  This is actually a bug in the Cocoon online doc generator, affecting anchors for C++ functions with a reference ampersand in the type signature.

Apparently that port is long (long) dead, and somewhere along the line
the stuff that sits below it changed, which caused our stuff to break.

The modern and maintained version of the same API is now
/usr/ports/databases/p5-DBD-mysql (/usr/ports/databases/p5-DBD-mysql50
when using mysql 5.0).

Fixed perl warning about Use of uninitialized value in numeric gt.

Load up the madwifi driver after the wired ones and after dhcp is finished
to avoid the latter succeeding on the wifi interface (this behavior has been
seen several times).

Fix a couple of typos.