"In system programmable" means that the micro controller code memory can be programmed to a new code image while the microcontroller stays in the system (meaning that it remains in its socket or soldered down to the board) and getting its power from the system power source. The means by which the memory is actually programmed can be performed in any of various schemes from special on board circuitry, access connectors, JTAG port, serial port access and/or forcing certain processor pins to special states.

"In application programmable" means that code image programming can be initiated while the device is running its actual end application firmware. This is most often done via the UART port (although this certainly could use some other scheme such as a 8-bit port interface or even USB on microcontrollers equipped with such). Usually there is a special operational procedure programmed into the device application firmware that initiates the code image download process. This could be a keystroke, special command from serial port, or a signal from a port pin. Once the code re-programming is restarted a portion of the old application code image is erased and replaced with a new one obtained from outside the part. A well designed "in application programming" system will be capable of recovering automatically, even through a power recycling" if the new code programming sequence does not complete normally. A common attribute of an "in application programmable" microcontroller is its ability to erase/program the code memory of the part via a user programmable procedure that runs resident on the part. This user procedure can be either user firmware that runs out of another block of the same memory space that is being programmed OR it can a special ROM programming image that runs from a user accessable entry point.

It is even possible to design an "in application" programmable system where a generic 8032 type part is attached to external FLASH and RAM. Such system would need to use FLASH that can be erased in blocks that are smaller than the whole memory space size. The portion not erased would retain the part of the code that interfaces with the external equipment to receive the new code image and this part would be cabable of restarting the programming process in case of error or loss of power in the current effort to re-program.

Hope that helps...
Michael Karas