What's the "scroll lock" key for -- better answer

In the article ‘What’s the “Scroll Lock” key on my computer for?’, http://www.straightdope.com/mailbag/mscrolllock.html , Una Person describes what the Scroll Lock key does, but fails to explain what it’s for. The answer is actually pretty straight forward. It’s there to provide compatibility with IBM mainframe terminals. IBM 3270 terminals were the deluxe terminal of choice for IBM mainframes at the time the first IBM PC was created. Both the Scroll Lock and the SysReq keys are present on IBM 3270 terminals. These keys were presumably provided so that IBM PCs could act as 3270 terminal emulators. In fact, IBM release IBM 3270 terminal emulator adapter cards for IBM PCs shortly after the first PCs shipped.

I don’t really know what these keys did on a 3270 terminal – I never used them much, and I remember really struggling with the user interface, which was a little strange, to say the least. I imagine Scroll-Lock acted as described in Una’s article. I don’t know what SysRq did. The premise of the 3270 terminal – which was radically different from other terminals in this period – was that the terminal should provide all local editing, and should – as much as possible – send information to the mainframe (or the front end IBM system 1 computers) in batches, instead of keystroke at a time. This strategy was neccessary to allow a multi-million dollar IBM 360 or 370 mainframe with less horsepower than a modern desktop computer to simultaneously host 100 users in a “time-sharing” environment.

To find out what these keys are really supposed to do, you would probably have to dig among the thousands of pages of documentation on IBM’s Systems Architecture strategy papers – IBM’s early grand vision of how mainframes and personal computers should co-exist in peace and harmony.

Personally, the one button I would really to see on modern PCs is the explosive power off button. IBM mainframes and minicomputers had a big round red “Stop” button on their front consoles that triggered an explosive charge on the outputs of the system’s power supply. The theory – as I understood it – was that if the system seriously malfunctioned, flicking a switch just wasn’t good enough to protect a multi-million dollar computer system; you had to blast those leads off the power supply in order to shut down the system as quickly as possible.

No. Either you’re wrong, or you’re living in an alternate universe. There is not and never was a Scroll Lock key on the 3270, nor is it even logical for there to be one, as a 3270 cannot scroll in the first place. And the SysRq key didn’t appear until the PC-AT, years after the original PC, but years before any serious attempt was made to converge PC and 3270 keyboards, and the implementation of the SysRq key in BIOS makes it perfectly clear that it was intended to communicate with the PC operating system, although it ultimately did end up being used to emulate the 3270 SysReq key on some emulators.

Will people who were not working in the computer field prior to 1981 please stop posting on this subject?

I don’t know too much about the history of the scroll lock or sysreq keys but I am intriqued by what rerdavies said about the “explosive power off button.” I do recall those big red emergency power off buttons (they usually had a handmade sign or piece of tape on them that said something like “Don’t touch under penalty of death!”) but I was never aware that they were wired to explosive charges. I know there were a number of urban legends floating around about them (I wish I could remember some…) so it’s possible that the explosive charge thing was one of these ULs. Does anyone have a cite confirming or denying this?

The emergency-off switch on mainframes is (or was) a true yank-out-the-cord instant power disconnect, which had a high probability of damaging the system with power surges, and of causing head crashes, and tape breakage. For this reason, it mechanically locked in the off position, and you had to call in an IBM repairman to check out the system before resetting the switch and bringing the system up again. Essentially, it was there only in case of fire. The normal power-off switch would send a “please shut down now” signal to all the components, and wait for a positive reply before finally shutting down the CPU.

(Also, some components had to be powered on long before – hours before – they could be used. The regular power-off switch would leave these alone.)

A lot of these questions bring up questions that boil down to “where did 7-bit US ASCII come from, anyway?” I dug up an old treasure, namely:

USA Standard Code for Information Interchange, X3.4-1968

It is reproduced as RFC 20 by a none other than Vint Cerf. While RFC 20 is informative, it leaves off the fascinating appendix:

Appendix A: Design Considerations for the Coded Character Set

which describes the choices that influenced the selection and code assignment of all of the characters. Such as…
[ul]
[li] Which characters should have one-bit differences to facilitate keyboard layout[/li][li] Which characters should have multi-bit differences to reduce the probability that they might be mistaken from each other due to transmission errors[/li][li] How the code layout facilitates subsetting[/li][li] Code layout to facilitate sorting[/li][li] …and on and on.[/li][/ul]
It’s a truly fascinating read.

Unfortunately, I can’t find it online anywhere. I copied mine at the Rice University library in 1978.

The 3270, although stupid by today’s standards, was quite amazing at the time. I first saw one in 1975. The “store a whole screen and send it at once” is quite correct. What that required, though, was a lot of local intellegence about how to edit the stuff on the screen. It had a lot of forms capability - it could do most of what basic text-based HTML forms can do. Probably not pop-up menus, but certainly field validation.

Remember, this was at about the same time that the ultimate dumb terminal - the ADM 3 “glass tty” came out. You had to consider yourself fortunate that you had “backspace” on that thing. :slight_smile: No lower case, no cursor controls. Yuuch!

And, no, I don’t miss either of them.