On Fri, 15 May 2015 15:48:15 -0400, John Eells wrote: > >> Why was this ever designed as a global repository rather than per-addressee? ><snip> > >We'd probably need a time machine to even ask why. > >But as Skip pointed out, disk and memory were absurdly expensive at the >time by today's standards. Also, when TSO (before the /E) was first >designed I'd think messages were few and far between. Certainly most >systems initially had mere handsful of users, and the number grew slowly >at first. ... > Likewise, the tracks were smaller. So allocating a 1-track message data set for each of a handful of users hardly seems profligate.
>(Anyone remember terminal rooms, and waiting for your turn?) > Indeed. And I regard that as an enabling factor in the prohibition, now onerous, of multiple concurrent sessions per user. (Hardly anyone could afford two terminals; why support it in the architecture?) It's as if roads once designed for horses had never been upgraded for automobiles. >That the first design lasted as long as it did before it had to be >extended is, in my view, quite a tribute to the original designer! It's >source of continuing amazement to me just how durable many of the early >design decisions were, and continue to be now. > But your recent remarks about the constraints imposed by PDS(E) architecture contribute to my perception that OS/360 was designed from a 16-bit point of view. It's now suffering growing pains. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
