------------ Forwarded Message ------------ Date: Tuesday, August 31, 2004 3:27 PM -1000 From: "(Cedric) Qin Zhang" <[EMAIL PROTECTED]> To: [EMAIL PROTECTED] Subject: [CSDL-LOCAL-L:2] telemetry defs
Just let you know that I have updated the telemetry defs on the public server with 358 stream definitions and 111 chart (report) definitions.
It's maintainence nightmare with so many definitions. Currently, I am using excel editing those definitions, and then export them in xml format, and then copy the xml file to Hackystat admin's directory. If anybody wants to define any new GLOBAL level difinitions, please coordinate with me. Otherwise, your change might get lost.
Please don't be alarmed that some of the scenes in telemetry wall (or control center) might not work, since I have changed chart names. Hopefully, I can write the some telemetry wall config by the end of tomorrow.
Cheers,
Cedric
---------- End Forwarded Message ----------
Hmm. Maybe we need to do some redesign. It seems to me that this combinatorial explosion of streams and charts is occuring because, for each stream type, we have the possibility of NumModules * NumDevelopers possible variants.
Here's a couple of ideas for how to improve the management:
(1) Allow for telemetry report definitions to support 'parameterized' streams. This would seem to reduce the number of required stream definitions (though not report definitions).
(2) Allow the TelemetryViewer to invoke the server with an anonymous report definition as an argument (i.e. instead of sending the report name, which was predefined, send an 'anonymous' report definition). To the extent that the TelemetryViewer wants to define streams/reports that regular browser-based users don't, this could cut down on both the number of stream and report definitions. (of course, the current named report based approach should continue to work).
This is actually really good news that we're having this problem. It indicates that we are implementing a fairly rich space of telemetry data.
Comments?
Cheers, Philip
