Text 5390, 70 rader
Skriven 2010-02-27 06:11:56 av Bob Ackley (1:300/3)
   Kommentar till text 5321 av Peter Knapper (3:772/1.0)
Ärende: Windows Update Fix
==========================
Replying to a message of Peter Knapper to Bob Ackley:

 PK> Hi Bob,

 WD>>> You never worked your way through a 7,500 page compile
 WD>>> print-out from the days that CRT's didn't exist as a
 WD>>> computer peripheral and the best you had was a T200
 WD>>> teletype.

 PK> This reminds of my early computing days. In 1969 I started
 PK> work as an Operator on an IBM System 360 model 40 CPU and
 PK> remember 2 brilliant types visiting from 500 miles south to
 PK> spend a weekend "tweaking" their ASM code for a SPOOL
 PK> (Simultaneous Peripheral Operations onLine) system. This
 PK> was 2KB of ASM code that controlled 2 x IBM 1403 Printers
 PK> that REQUIRED a full 2314 track buffer (5KB of RAM per
 PK> printer), and all this ran in "F1" using just 16KB of RAM
 PK> (the model 40 had a total of 128KB RAM of CORE memory...).
 PK> It even had user selectable page re-starts (for printer
 PK> jams or alignment issues). The Spool code saved us HOURS of
 PK> waiting for the printing to finish...

 PK> That weekend they started at 08:00 on the Satuday, ended at
 PK> 1500 on the Sunday and had 1,000 of pages of printed ASM
 PK> output from the Modules they wrote. During that time they
 PK> added a 2KB Card reader Batch Job storing module, and a 2KB
 PK> BiSYNC transmission module using an IBM 2703 to move files
 PK> (using a 56Kb leased line to Auckland, 84 miles away)
 PK> between this machine and a 360/65 in Auckland. Each extra
 PK> module expanded the Memory required in F1 by 2KB.......;-)
 PK> The later was not completed that weekend, but was completed
 PK> the following weekend. We used that environment (running
 PK> IBM Mainframe DOS release 19, 20 & 21) for many ears
 PK> before. 

 PK> Ah yes, those were the days when code was efficient.....;-)

And programmers were competent.  I once watched one of our civilian contractor
programmers type in about 100 screens of assembler code (24 lines per screen)
and then have it assemble and *work* in one try.  And Regis (Martin, not
Philbin)
was the stereotypical computer nerd - a bit pudgy, coke bottle glasses and all
-
but he could make that computer do whatever he wanted it to.  This back in the
mid 1970s.

Before I left the business the quality of the software and the caliber of the
technicians had declined considerably.  In 1994 the hospital I worked in
installed
a 'thoroughly tested and debugged' upgrade to its applications package (from
Shared Medical Systems in Malvern, PA) overnight batch system, and the
following
Saturday night the production cycle went into a "wait until forever" loop with
a
series of half a dozen or so jobs each waiting on the one ahead of it, with the

first one waiting on the last one.  "Thoroughly tested and debugged."  It took
SMS
about eight hours - and having to call somebody in to work on a Sunday morning
-
to figure out which one of those jobs had to be forced to start to get things
running
again.

Back in the days of the 8080 chip software had to be efficient, too.  SuperCalc
ran quite happily under CP/M on that chip - OS, program and data all contained
in 64KB of RAM.  M$ can't write a mouse driver that's smaller than 100KB.\

---
 * Origin: Bob's Boneyard, Emerson, Iowa (1:300/3)