interface. This stuff is going to dissappear pretty soon, once we
shift from the paperbag to the xmlrpcbag.

* Added a wrapper class so that you can invoke methods as
  experiment.swapexp or node.reboot. So instead of invoking as
  /XMLRPC/experiment can calling swapexp, you can call the server as
  /XMLRPC and call experiment.swapexp. This allows you to use a single
  connection to talk to different parts of the API. Note this is standard
  (or is it defacto) syntax in XMLRPC.

* Changed the demonstration client to talk the server this way.

* Changed paperbag to allow this as well; the xmlrpc server is invoked with
  no args, which tells it to export the wrapper interface instead of a
  specific module interface.

* A few more cleanups in the server, more permission checks, etc.

   This sometimes results in wonky, skewed viewing transforms.
   - Add hvkill function instead of depending on HyperViewer clean-out code.
   - Improve string handling, remove all fixed-size buffers.
   - Keep label truncation from segmentation violation with a wonky transform.

 . OpenDialog initialization: focus and load text from command-line args.

 . Fix a platform-dependent bug in LabelsMode, resulting from changing it from
   a dropdown to a choice type widget in wxGlade.  Failed to work on Windows.

zero, probably for debugging) so that silly ssh warnings do not appear
on console.

Add -T option to ssh to avoid tty allocation on server side; no need
for that.

Add a couple of destructors to make sure that ssh is torn down when
destroying a proxy wrapper.

Added three helper programs. These are utilities to make it easier to directly interface the output of ipassign to the input of distributed-dijkstra. Using these utilities, they can be tested as a unit without having to go through the rest of the infrastructure of emulab.

Make various combinations of picking nodes, typing their names, and clicking "Go to top" play together nicely.

interface (which will have its own frontend to emulate the original
script interface!)

* Refactor the experiment permission checks into a subroutine.

* Cleanup some parameter names that did not make sense.

* Add portstats, readycount, node_admin, eventsys_control, savelogs,

* A couple of little fixes.

 . Allow typing in a node name, go to it when Enter is pressed.
 . Get node name after the GoToTop button is pressed.

 . Improve control panel layout by removing spacers and using Border spacing instead.
 . Start to make resizing work by setting Expand options in the GUI.
 . Change the LabelsMode widget from a wxComboBox to a wxChoice.
 . Disallow non-numeric chars and handle Enter in the "Node/Link depth" spinboxes.
 . Handle Enter in the Open dialog text fields; complain politely about empty text.
 . Disable show/hide buttons when there are no links in or out, respectively.

partitioning scheme. They seem to perform about the same, which is not what was expected. Further tests and tweaks may uncover the cause.

* Change use of TBGetSiteVar to the non-fatal variant to prevent the
  batch daemon from exiting when mysql goes whacky.

store the result in. When called this new way, the value goes into
$rptr, and exit status is returned to caller instead. In addition,
when called this way, all errors are non-fatal; it is up to the caller
to decide what to do.

* The first involves swapmod. When a swapmod on an active experiment fails,
  tbswap will reswap the experiment back to the original configuration. The
  problem is that it is reswapping it with the *new* virtual state of the
  experiment in the DB. It is not until later when control returns to
  swapexp that the virtual state is restored. This is plainly wrong, and in
  fact was causing the event scheduler grief cause it was starting up,
  reading the the virtual topo, which was different, wrong, and about to be
  blown away.

  I reorganized the modify section of swapexp so that virtual state is
  restored only when its a swapmod on a swapped experiment. On an active
  experiment, I moved that code down into tbswap, which will now does all
  of the virtual and physical state retore before it does the reswap back
  to the original experiment. Just for kicks, its also done if tbswap
  decides to swap the experiment cause of a fatal error.

  Cleanups: I changed $NoRecover to $CanRecover. My feeble brain cannot
  deal with !$NoRecover. I know, two knots make a wright for most people.

  Renderer: I was annoyed by the fact that we rerun the renderer on a
  failed swapmod. The original reason is that the renderer runs in the
  background and so vis_nodes cannot be saved with the rest of the virtual
  state tables cause the renderer might still be running when the user
  fires off the swapmod. Well, the hell with that. We lock the vis_nodes
  table anyway in the renderer during update, so we are certain to get a
  consistent snapshot. We store the renderer pid in the experiments table,
  so if the renderer was running, just fire off another one; mostly this is
  not going to happen. In addition, tbprerun no longer starts a new
  renderer when doing the swapmod; I start the new renderer later after
  swapmod succeeds. I might end up tweaking this a bit depending on what
  people notice as being different.

* Termination changes to batchexp and swapexp: I've rearranged the
  termination code using an END block so that any uncontrolled exit from
  either batchexp or swapexp will go through the cleanup code, and
  hopefully insert a stats record, as well as not leave the experiment in
  some inbetween state. I've set the max DB retry count to zero in both
  cases, which means infinite retry. I've also added SIGTERM handlers to
  both so that again, we can kill a hung batch/swap and have it clean up
  things more or less. Note that END blocks are not caught when a signal
  causes the program to die; you have to catch it and then die() so that
  the END block is executed.

  Eventually, we need to clean up the various libraries so that we do not
  use DBQueryFatal(), but rather use DBQueryWarn(), and look for failure.
  Ditto for event system interface.

That will show the devil who means business. Right on.

of affairs.

This is apparently a no-no and was happening at DETER.

It isn't needed on clients and it requires the linuxthreads package
which is not otherwise needed.

I took it off when debugging popen streams on Windows; it shouldn't have been checked in.

to be reused if the DB is dropped and recreated, since when that
happens, auto_increment history is lost and it will go back to using
the latest highest index in the table. Usually not a problem, but
since we cross index three other tables using the experiment index,
this causes quite a bit of grief.

So, my solution is to do my own auto_increment using the
experiment_stats table (locked of course), which we never delete
entries from without deleting all entries from the other cross
referenced tables.

    DBQueryFatal("select MAX(exptidx) from experiment_stats");

I also added a sanity check to make sure the new index is not
currently in use in any of the tables. I also cleaned up the
error path when something goes wrong.

Add a cross-reference from the Windows installation section to the release notes section on setting up SSH or PuTTY/plink.

NFS, convert to using a proxy that runs on ops, which does the copying
locally.

Wrote new abstract and outline for WORLDS paper based on today's paper
meeting.

webfeedback to talk to the newer version of canaryd.  Add feedback
"estimate" stuff so that if we have no data (because of an overloaded
node) to work with, we can make some sort of "reasonable" guesstimate
on every iteration.

than jamming more stuff into the old one.  Most of the code came from
the previous version of canaryd, the cpu broker (process accounting),
and the janosvm (network interface accounting).  Its missing some
features of the old one, but those can be incorporated without too
much trouble.

Changes:

  Designed to permanently run on the pnodes:  it waits for START events
    before it begins recording.  However, I haven't done the work
    necessary to have it always startup on the pnodes.

  No more exec'ing: process stuff is taken from "/proc", and network
    interface stats are pulled from getifaddrs(3).

  Fixed some minor bugs: A typo caused the real-time priority to not
    be set, use setitimer instead of sleep to get more accurate
    spacing between samples.

return instead.

at the same time you clear the load address. admin people only.

instead of just their names.  Enhance experiment.statewait() to accept
multiple states to wait for, or, to wait for any state change.  Add
experiment.thumbnail() method so the topology can be displayed in
NetBuild.