Caltrops - Tansin A. Darcos's Alter Ego





	[quote name="Commander Tansin A. Darcos"]When I first got started programming, call it around 1976, the mainframe we had at school was a Univac 90/60 running VS/9. This is roughly equivalent to an IBM 360/175 or possibly an S/370 running, oh, maybe TSO or something more advanced. That was one thing, I've used IBM-based mainframe computers, and IBM never really "got" the idea of interactive computing. Actually I don't think they ever understood anything about usability or accessibility as far as being able to use a computer when you don't have the equivalent of a couple of years of programming experience. Univac's support for programmers at terminals was much better than anything IBM put out. And even if you weren't a programmer you could reasonably understand how to log on to the system, execute a program, search the directory for a file, erase or print a file, and log off, and learn all of than in maybe 5 minutes to 1/2 an hour. Their batch language on VS/9 - which were almost the same commands as those used on a terminal - far exceeded the usability and capability of IBM's JCL, which was, to put it mildly, painful to use. To be able to reasonably learn how to use batch JCL on IBM systems for a non-programmer who just wanted, say, occasional reports, would probably take, not the 5 minutes to 1/2 an hour of Univac's VS/9, about a week, unless they relied on canned procedures written by the programming staff. Plus, as I pointed out, VS/9 used the same command language on batch as on a terminal. TSO on IBM is a wholly separate set of commands and options and is radically different from JCL Consider if, when using a word processor, you had to tell Windows how much disk space you were going to use to store your file. And you had to create the file first before you could use it, you couldn't just issue a "Save" command or activate a save option on a menu to do so. Everybody else did much better for interactive computing - or even batch processing - than anything IBM put out. RSTS/E on minicomputers from Digital was better. NOS on Control Data was at least marginally better. DCL on Digital's VAX/VMS was better. Xerox Data Systems Sigma/7 was better. Hell, even the original Bourne Shell on Unix/Linux is more usable than <i>anything</i> IBM ever put out. None of them had the advertising budget to compete and the manufacturers really weren't that good on figuring out what customers needed and selling them a solution. That's the real reason IBM's razor sharp and dangerous, painfully agonizing to use systems sold so well, because IBM knew how to market them and understood their customers. The people who actually had to work with this stuff? Not so much. I've used all of these and I've used TSO and IBM's batch processing language, JCL. They suck. Even the most recent update of ISPF (the shell that runs on TSO) is painful for trying to do serious work. Just trying to look for files can be a major agony. I've often felt that MSDOS' DIR command, even with the latest options in Windows 8/10 leave quite a bit to be desired. But compare that to the capability on today's mainframe computers: ISPF's directory lookup screens and any of TSO's commands for finding datasets (IBM mainframes do not use 'files' they use 'datasets') are like stone age axes and flint knives in comparison to a cheap $20 axe at Home Depot or a $2 knife purchased in a store. MSDOS has poor Directory lookup functionality but it's still light-years ahead of anything on IBM mainframes. The only thing that came close to giving IBM mainframes reasonable usability was a third-party operating system called MUSIC/SP that was developed by McGill University in Montreal and that IBM decided to sell as an option. Even that was much easier to work with than any IBM-developed terminal/batch system. You actually have a better chance of being able to get things done on a MUSIC system than you can working on TSO or even with ISPF, IBM's shell that runs on TSO. The one really good choice IBM made was when it developed the PC it decided to go to a third party for the operating system. They almost used a version of JCL or TSO, and as it's hard for even programmers to work with it, I can guarantee that it would have made the PC unusable for ordinary people who just wanted a way to get things done that was less bureaucratic and easier to accomplish than the priesthood that ran the mainframes they often had to connect to in order to accomplish anything. IBM is basically the only company left that still manufactures mainframe computers (at least for customers outside of Japan and Europe; I think ICL still does some sales in Europe and Fujitsu to some European and many Japanese customers sells its B8000 which is similar to Univac's 9000 series IBM clone). And chances are if you did anything for purchase or cash back some time during this week using a credit or debit card you at some point ran a transaction that passed through an IBM mainframe. They're very good for high transaction volumes and when you get into levels of processing where you'd need 1,000 or more servers to handle transaction volumes, a mainframe is much less expensive than a server farm of Intel/AMD PCs running Linux, even when you add in IBM's monthly software rental licensing charges. So transaction processing is one thing on which IBM can provide outstanding value. But as far as human interaction, IBM never got it and generally all of their tools to do anything either interactively or in batch processing were basically made difficult because they were designed by engineers/programmers for other programmers. Which gives a lot of power and capability for technically knowledgeable people but makes it a son-of-a-bitch for ordinary mortals to work with. [/quote]