« Male Advantages III | Main | Male Advantages IV »
September 06, 2003
Unix Geek - History Continues
Well here was my first full time job as a pure Unix specialist - I had worked with Unix, programmed in Unix, slept with Unix (well, what else do you call an all-nighter in the lab?) but now someone actually wanted those Unix skills. The deal was pretty good, I had to learn operations tasks on the existing system (ICL System 25) plus provide Unix skills to a new migration project (moving a system from the ICL 25 to some as-yet-undecided-version of Unix). The migration project was too big for one lab so a number of labs had pooled resources - we had the Unix guy, someone else had a C programmer and so on. In the meantime, I wrote small apps "to keep things going until the new system arrives" on SCO (yes that SCO) Xenix and later SCO Unix. We knew the support people at BlueSky pretty well and in fact became SCO resellers.
This idea of cobbling things together was quite common but being naive, I actually fell for the "only be needed for a couple of weeks" story. Perl was not an option (didn't exist) so things were written in shell, awk, sed, grep - all the standard UNIX tools. Databases? What were they? You made do with files in directories with links if necessary. Lookup meant navigating to the right directory and a record (thanks to links) was a file that appeared to be in multiple places (like under Surname/Smith/record and Firstname/John/record and Diagnosis/pregnant/record). Front ends for "dumb users" were done in curses (and it is called that for a very good reason) because that worked on VT100 terminals. Note that the "dumb users" were in this case specialists with many years of study and expertise but they knew nothing about computers. One of my greatest achievements of this job was a cobbled mess for histo-pathology. It was used to not only process patient records but to generate statistics. Given the likely short nature of the project it was the normal curses/files setup. Some years later (long after I switched jobs) I had a panic call from the lab. My two-or-three month system had been running for almost five years and it had failed. Why? They ran out of disk space. I learned a really valuable lesson here: never write any code on the assumption that it will be short-lived. Assume that your code will still be in use in 20 years and you will have to go back and fix it. That ought to make you comment things better!
Another story from this period of my career is a classic - but true - story of the absolute dumb user. Back in that era, SCO was distributed on floppy disks - and I mean floppy disks, not these little 3.5" things we call floppy but the earlier 5.25" ones (I also used 8" ones on TRS systems). There was, from memory a B1 disk, and N1 through to N3 and S1 through to S8 or something similar. We had a customer who wanted to buy a copy of SCO without installation. In such cases we made almost no profit but it helped with the requirements of being a reseller (so many licences per year). If they bought installation, we would do the whole thing in about 4 hours, bill for a days work and everyone was happy. If the customer did not buy installation we just shipped the boxes and manuals, invariably the customer would then get stuck and call us up - good for anywhere up to a weeks consulting and on-site callout fees.
Well this particular customer called up and said "I can't get disk N2 into the drive". Now, I had heard all the usual ones - like typing the disk labels and putting them through fax machines but I had never had a customer fail to insert a floppy before. After lots of putzing around, I got an on-site visit. Unfortunately for good customer relations, I split my sides laughing when I got there. The customer advised that they had also had problems with N1 and lots of funny "disk error" messages as well. You see the instructions read something like this (my memory is vague about the exact wording):
1. Insert B1 and reboot you machine
2. Type boot-install when the messages stop scrolling up the screen
3. Insert N1 when asked
3. Insert N2 when asked
...
What the instructions failed to explain was that B1 (and N1 for that matter) should have been removed before inserting the next disk :-) It was possible to jamb 2 disks into the old full height 5.25" drives but not three (the gate would not close).
To be continued....
Posted by Ozguru at September 6, 2003 04:09 PM
Comments
Posted by: Jivha at September 6, 2003 04:09 PM
Posted by: ozguru at September 6, 2003 04:09 PM
