Software for IBM 36030 was DOS360: Forty years 517

OK; if it's after New World the rules change for newly-introduced products. Do you recall the product number? If it was something looking like

360S-CB123

it's pre-New World (yes, the number above is bogus. I don't recall what the ID was for the COBOL compilers shipped with OS-360). If the number was something like

5740-CB1

then it's a program product, and not free.

Another clue was the color of the manual covers: at least in that period as a general rule purple covers were (usually? always?) a flag that the product was billable.

I have no idea what discussions went on about that at Galactic Headquarters, but some possible points:

* Pre-New World, customers were sold machines with explicit statements that the currently-available software (and updates) came at no charge

* The existing software was shipped with no copyright notices, which under the copyright law in effect at that time had the effect of placing the material in the public domain

It's amazing how much function we could pack into what today would be an impossibly small memory space. Kids today casually talk about their systems with over a gig of memory; one of my lecture stories describes getting a box with 256 KB of memory at the bargain price of $330,000 *after* educational discount...we had a REAL incentive to write memory-efficient programs.

That was a common buttignment on a 3-channel machine. Channel zero was always a byte multiplexor (slow, but supported multiple simultaneous operations); channel 1 had the disks (by convention, 19x for 2311s ... which is where the standard addresses in VM originated), and channel 2 was used for tapes. Are you sure your tapes were 281 and 282, and not 280-281?

You *could* have done anything with enough money and time, but the result would probably have been a rather messy hack. The channel architecture of the S-360 certainly had its limitations and was hideously expensive to implement (look at the signal timing relationships and tolerances vs. the available technology, especially in the early days), but it provided a clean, standardized design for both the physical and programming interfaces. Not only that, but the control unit would have had to be redesigned to hold an arbitrarily-long block of data, and would need internal paths to support the simultaneous data flows.

This *was* implemented in some cases with the "read-while-writing" option, but it was predicated on having two control units on the tape string, attached to different selector channels. If you've got only three channels you're out of luck since 0 is a byte mux, 1 is used for DASD, and you're left with only the second selector for tapes. (And while I never researched the issue, I doubt that the 2415 ever support read-while-writing.)

Joe Morris