Blowing the Dust Off of an IBM AS/400 Server
hackaday.com
hackaday.com
She knew exactly the right combo of Function keys followed by how many TAB she needed to start typing. She could do it without looking at the screen or the keyboard, just the sheet she had to enter/modify in the system.
It was incredibly fast, 99% of the time.
She hated the move to Windows. All the sudden, everything was slow, error prone, fiddly. Keyboard navigation was mostly gone and although employees could start working with less training, she knew the move was generally a big step backwards.
She despised computers with passion from that day on. Today she's retired and uses an iPad only.
I'm a UX designer/developer. There is something to be said for these old terminal systems, when it comes to repetitive daily tasks.
She was right.
Autohotkey can solve a lot of these problems.
Though it can require some politics playing to accept it as a legitimate tool.
Another thing is that the WIMP (Windows, Icons, Menus, Pointer) paradigm was, at least in part, born because of the insight that the block oriented approach of the green screen terminals wouldn't scale beyond a certain point. I'm more interested in the ideas that made IBM 5250 like UX successful than in a modern interpretation of it. Using keyboard shortcuts is certainly one of these ideas and so is tiling windows instead of stacking them.
Compared to that, these old systems were modal (edit, browse, etc modes) and keyboard-based with it's ~100 action buttons, where pixel-by-pixel navigating was not needed.
From an efficiency standpoint, petting a computer with mouse-based navigation is ridiculous. It might help a newbie until they explore a software's features but later on it's a waste of time. And your mother knew this very well.
There are obviously a plethora of use cases (mostly graphic design software, games, etc) where a pointing device would benefit far much better.
OTOH I would love to see a vim-like excel/spreadsheet software.
Out of curiosity, are you then employing the experiences of your mother when designing user interfaces?
I'm a programmer, and I rarely design anything more complex than a static website or a CLI. I have no illusions about UI design being easy. However, I am disappointed that it seems most UI designers, whom I do assume to be competent, are working on UIs that are optimised for ease-of-use by the novice, rather than the expert.
I use and enjoy several programs that have "expert-optimised" UIs. Emacs is one, and the Unix shell is another. However, these are grown in an ad-hoc way, usually by people who did not think too hard about UI. I wonder what an optimised expert-oriented UI would look like, if it were developed by true UI experts.
> imagine hitting Ctrl-Alt-Del, entering your login and password, hitting Enter, entering FFOX to run Firefox, hitting Enter, hitting F3 to open a tab, entering news.ycombinator.com, hitting Enter, hitting F3 for another tab, entering another URL, hitting Enter and it just works. Your machine can be waiting to show you the login screen, or applying some global policies, or launching Firefox, but it will never lose your input.
You'd have to wait after starting firefox for it to appear to get your input queued correctly.
Within a browser you can mostly hit Ctrl-T and start typing, although I've seen this break down. Edge for some reason was bad at this.
Within websites there is often anti-user javascript which causes problems: the absolute worst offender is Facebook, after you've put focus in a text box you need to wait for it to settle or your text will come out in the wrong order.
To give you an example, imagine hitting Ctrl-Alt-Del, entering your login and password, hitting Enter, entering FFOX to run Firefox, hitting Enter, hitting F3 to open a tab, entering news.ycombinator.com, hitting Enter, hitting F3 for another tab, entering another URL, hitting Enter and it just works. Your machine can be waiting to show you the login screen, or applying some global policies, or launching Firefox, but it will never lose your input.
However, I wish there was a way to ignore (or cache?) this buffered input if the process exits with an error or something else of the sort. Does this exist?
It can quickly get dangerous otherwise!
Keyboard > Mouse > Touch > Voice
Hates it every day and still gives me examples of how the old system was superior and faster.
Working in professional services a number of our clients still use AS/400 mainframe environments, with data migrations to other environments becoming more and more common.
I worked on a data migration from an AS/400 environment to an SQL DB. I kid you not, the database schema for the AS/400 environment was printed on old A3 scrolls that looked tea stained.
For simply that reason, they work and rarely have any issues. The problem is that those with the expertise to manage them and fix them in the event that things do go wrong are diminishing. And that's a big risk for businesses.
I wouldn't be given the people that use ours, day and night shifts, have never seen it crash or freeze for the decade or so they worked here. They say it only goes down when a cheap sysadmin accidentally unplugs it trying to swap out a different server or if they have to image stuff after suspected hacks. Otherwise, they're the Energizer bunny of servers.
Oh boy, I've seen multiple businesses that utilize AS/400 system(s). I even know a few people who manage them, building out applications, etc... They are still very much used.
You'd also be surprised how many records about you still exist in said system(s). Hell, I was using an AS/400 system only 3 years ago for business purposes (I'm not a COBOL developer, but something had to be fixed).
I appreciate the data point on that one. Sounds like it aged well. Which CPU('s) did it have?
The current one is an IBM 9406-800 which I do believe is the SStar which is the RS64 before the POWER 4.
Interestingly, the old machine has a tape drive that is not supported on the new Power 9 boxes (its a non-LTO QIC drive). I guess the good thing is now I can keep a single stock of LTO-7 tapes. I am a bit odd and have an LTO-7 drive connected to our FreeBSD server for a "last chance" backup.
When I started my first proper "IT" job, back in the late 90s for a big multinational, they were running their payroll system off it. The main node was about a decade old at the time.
The input system for it was a 16 bit VB4 desktop app that wrote text files to a netware share that this picked up.
I am always impressed at how these bits of old tech, sticky tape and string were so unbelievably reliable and scalable across thousands of users. I actually actively miss working on this sort of stuff.
I was actually trying to learn RPG mentioned in the article last week, a blast from the past!
Pretty solid box. never ever ever crashed or got slow or anything. just a mind fuck to do anything with.
Currently working on another mainframe os but wish that I can some day find a place with an i-series or z-box (IBM mainframes).
Only in my 20s, but these systems will be around until I retire. The hardware might change, but the cost of educating developers in RPG, COBOL, control languages etc must be far lower than translating the code.
To be fair, one day a few years ago I did get access to my clients' mainframe last year. A Z-something big-iron running on of the frames on AIX which I had shell access to. It was a "whoa man, I'm in. I got access to the mainframe!"-moment. My favourite tools were installed by someone else, so it was quite enjoyable; midnight commander, zshell and bourne again shell were present :).