TahoeJimbo

Doing a faithful VT100 implementation wasn't too taxing, though emulating all of the bugs and bizarre edge cases was. That's where DCOM ruled the roost. The real issue was getting the serial performance up without dropping characters. (Apple never really embraced hardware interrupts until the //gs.) DCOM worked well on the Apple //e and //c because it could poll the serial port ~1000 times a second, even while drawing on the internal 80 column screen.

Wow. Imagine finding this 34 years later. The Videx VideoTerm driver caused me no end of grief. I suspect there were a few different versions of the card's hardware. It had an Hitachi video chip that required programming all the video timing values directly to the chip. The chip also needed all read/write accesses to be synchronized to the vertical blanking period. Without synchronization, you'd get timing artifacts on the screen where the scan would be interrupted by a few clock cycles. The architecture of DCOM was designed to support a mix of serial cards, many without interrupt support. This precluded me from using the I/O routines provided by the Videx ROM, which would cause input characters to be dropped from the serial card. ? Of course, the 55 year-old me would have designed it differently. ? -Jim Hayes