Notes from the Meeting of May 28. B. Clark noted that the Nucleus toolset does provide an overlaying linker, so that it is possible to have the software for the device in a separate module from the system and services. This will require careful management if we decide to do it, since the links can only go from the overlay to the system, and not vice versa; also great care will be needed to make sure that the overlay is linked with the correct version of the system. B. Sahr reports that the OS port procurement is proceeding with all deliberate speed. He has separated the port procurement from the tools and simulator purchases to make sure that long lead item gets underway as soon as possible. W. Koski says he is proceeding with all deliberate speed to make the mezzanine card to interface the on chip MAC to a network PHY chip. B. Clark made a list of the various communications one expects with the MIB, or with the Object residing therein. Reproduced below, with some afterthoughts added. What does a MIB say Monitor point to be logged Monitor data to verify command Checker message to the operator Error message for bad command Flag "My device makes the data no good." Here I am; I have just booted Monitor point to be snooped by screen What do you say to a MIB Execute a command to your device You are "in use"; reject some screen commands At the tone, the time will be .... Reboot Write to your flash Change your logging/verification/chatter time intervals Report your currently active checker messages Q and A What is the value of this monitor point? What are your monitor points? Do you have Device objects below you in the hierarchy? What are your commands? R. Moeser reported on some speed measurements of various communication protocols. Fastest was just sending a binary buffer, of course. This tends to be about a factor of ten faster than the next faster, which is Java RMI, which serializes the object, sends it over a socket, and then reconstructs. Slowest is SOAP, which involves converting all values to ascii, encasing them in XML, and sending them serially. SOAP is inefficient in both bandwidth (it takes a lot of characters to represent floating point values) and CPU time (mostly spent in sprintf). ----- End Included Message -----