Pertec interface

On 1 Apr 2012 at 0:29, Jeff Woolsey wrote: It was I who made the assertion--and the occasion is burned into my memory. "L" tapes--so-called "Long Stranger" tapes--as opposed to "S" or "Stranger" tapes. If you've got the source for, say, SCOPE 3.16, 3.2, 3.3 or 3.4, you'll find it in the PP overlay 1LT (IIRC). I don't know about other versions. So long as the user program could keep OUT chasing IN during a read, the tape would keep rolling. When I got saddled with this particular bit of perversion, it was during some rather touchy negotionas with the Air Force, who was seeking changes to their contract with CDC. So the brass instructed their horde of GSA types to go over the contract specifications with fine-toothed comb to see where we might be found lacking in order to gain an advantage. Well, when the original proposal was written, most of the Common Product sales boilerplate was simply attached to the "special" stuff in the proposal--and long stranger tapes were part of that boilerplate. One thing that we'd neglected to consider was that our "special stuff" made heavy use of ECS--and we had 4MW of the stuff shared by up to 4 machines. Normally, the only really visible sign of that use was that the image on the console display dimmed and flickered a bit when memory access got thwarted by an ECS transfer. But it was really, really bad for L tape access--the driver assembled a word and wrote it immedately to CM and updated the IN pointer. Get delayed more than a bit and you lose the tape data stream. It was horrible to watch. The fly boys would up getting their notch on the bedpost. I don't know if L tapes made it much beyond SCOPE 3.4; I can't conceive of how 6RM would have handled them. "How long can an L tape record be? How long is a string?" Those were the days when QSEs and QSSs fell like rain for SSD. I'm sure the ROVER people have some eually interesting stories--if they can talk about them, that is. --Chuck