So that's basically it from the NDT perspective.  But it really helps to
understand my changes in the context of the overall site architecture.
So I'll spend a few moments summarizing that.  The map we use on the
site is a Google Map, and as you probably know, all Google maps are
driven by Javascript.  I've used Javascript to create the map, populate
the markers, return the current location, get info about an existing
marker, and create a new marker.  All my Javascript that does this can
be found in the file http://www.ecorridors.vt.edu/maps/loadmap.js


When a user visits the site, we've tried to encourage the following
order of events:
(assume steps 10-11 have already happened)  
1)   User clicks the "Run Internet Speed Test" button.  This opens
speedtast.htm, which contains our custom NDT client, in a new window.
At this point the window.opener assocation is made in speedtest.htm that
allows it to pass its results back to the main form.
2)  User elects to trust the NDT applet, because it's signed by Virginia
Tech and hey, everybody can trust Virginia Tech.
3)  User runs the NDT test.  Gets some results.
4)  User clicks the "Return to Map" button.  The results are pasted back
on the main form.  We also have hidden fields that store the server name
from which web100srv was run (harvested from Public String host in NDT),
and the IP address of the user's machine, which comes from PHP.  This
gives us both endpoints.
5)  User clicks on the map somewhere to mark their location.  The
lat/long of said location is transferred to the main form.
6)  User adds the various optional information in the "Service Details"
area.
7)  The PHP validates the submission.  Hopefully everything goes OK.
8)  PHP inserts the data, along with the hidden fields mentioned above
and a timestamp, into a MySQL database.
9) Page refreshes.
10) Google Map loads its markers from dynamicxml.php, which just takes
as input the "envelope" of minx, miny, maxx, maxy that represents the
user's current viewing extent, queries MySQL, and then builds a
well-formed XML document in the format that Google Maps expects
containing the points, and their attributes, that fall into that
bounding rectangle.  
11)  We then do a little kludgy logic to create the map symbology.
Whereas a GIS would do this for us out of the box, since Google Maps is
not a GIS by any means, we have to do it manually.  So I have 45
different marker symbols for every permutation of connection speed class
and access tachnology that follow a naming convention.  The big IF
statements in the beginning of loadmap.js take care of assigning marker
symbols by building the filename in two separate steps.



So that's the context in which the custom client is used.  I would say
that the changes I have made are very dependent on this context, and
don't make much sense outside of it.  Now, what might be interesting
would be if you and I (and/or other members of the NDT community) could
take what I've created as a starting point and then build something more
generic that would essentially "spatially enable" NDT.  Or maybe build
some kind of web template for users to customize that would already have
the linkages to NDT built-in, so that they only had to customize the
testingserver parameter and the look and feel of their site.  I'm open
to any suggestions you might have with regards to the future of this
project.  Of course, the entire source code I'm sending you is free (as
in speech AND beer) to the community.


One additional thing I was thinking of trying was to add a
load-balancing element to the NDT client.  We have two servers here that
we plan to install NDT on (we have it running only on 128.173.240.137 at
the moment).  I was thinking that if a user accessed server #1, but it
was currently serving another client, the java client could be modified
to check server #2 before entering the queue.  If I get this running
here at our shop, and I can polish it enough to be worthy of
contribution, I might send that your way also.


So... I see I've been at this for a while and now most of the morning
has gone by.  I hope that the information I've provided will be useful
to you as you examine my changes to the client source and try to
determine if/how my work would fit into the NDT project.  Please let me
know if you have any questions, and I look forward to continuing this
discussion!


______________________________________
Seth Peery
Environmental Design and Planning
Research Assistant, eCorridors Program
Virginia Tech 
sspeery@vt.edu