The Magical Excel 97 Far East Language Build Screwdriver | Hacker News Reader

The Magical Excel 97 Far East Language Build Screwdriver

devblogs.microsoft.com

124 pointsr4um79 comments

Last week, a PC in my wife's medical office wouldn't boot, with the dreaded "no drive detected" error message. Figuring "what do we have to lose", I turned it off cracked the case, started whacking the C: drive with a screwdriver handle, and punched the power.

It booted.

The office manager who was watching me had to pick her jaw up off the floor. We proceeded to copy everything possibly important off it onto a USB drive, knowing that may well have been its last spin-up ever and it's next power off may be the final one.

While in college I worked tech support for a pc manufacturer. Western Digital had a 1G drive that regularly would just seize up due to some sort of issue.

The short term fix was exactly as you describe. Give the drive a few strong hits and it would resume spinning.

I had to do this with a Quantum drive in an old 90s Mac - my approach was to hold it horizontally and rotate the drive aggressively to try to unstick the platters. Worked well but we also copied stuff off of it asap after that.

OMG that reminded me! Someone sold me a cheap 50MB HD in the late 90s (it was probably full of dubiously obtained files or such). It would teeeechnically spin up, but my Amiga was pulling about 50KB per second off of it, and it made a grinding bearing sound. Figuring I had nothing to lose, I flipped it over and squirted a little 3-in-1 oil into the bearing.

RRRRNGNGNGGNGGGGGRRRRRRRRMMMMMMmmmmmmmmm.....

...as the grinding slowly morphed into the high pitched whine of a happy SCSI drive, and the drive transfer meter crept up to a reasonable 1MB per second. I got the files off of it with a quickness, then threw it away.

Yes, I have officially, in front of witnesses, sped up a computer by oiling it.

Back in early 90's, a friend had a 20Mb hard drive with stiction problems that wouldn't spin up. We wound up putting it on a radiator, and that warmed it up enough that it would spin up and be a working hard drive. Didn't store anything important on it after that, but it did stay running for a couple of months in that state.

> whacking the C: drive with a screwdriver handle, and punched the power.

Disk heads have been "sticking" like that forever[1], I guess. And it's a good bit that it will stick back down soon and plow right into a set of cylinders.

And the internals of disks have gotten more finicky (helium filled), faster spinning aerodynamically lifted heads etc.

[1] - http://ars.userfriendly.org/cartoons/?id=20010326

whacking the side of a TV also was a common "solution" for TV's back in the CRT days.

https://www.reddit.com/r/AskElectronics/comments/2b3cpi/why_...

https://www.wikidata.org/wiki/Q7167193

Percussive Maintenance

(uncountable) (humorous) The use of physical concussion, such as a knock or a tap, in an attempt to make a malfunctioning device or person work.

Device or person? I've been tempted on the "or person" part, but never actually tried it...

"Beatings will continue until morale improves"

Pete Conrad famously had to smack around some surface experiments during Apollo 12.

Growing up, we had a CRT that my father had built from a kit. Occasionally, the image would tint green-ish. When this happened, my father would slide the cabinet 45 degrees out (so he could get behind it, and then turn a blow dryer on the back of it. Given a little bit of time (5 minutes? it was a long time ago) the picture would go back to normal.

Also, he helped me fix a problem with my car using a piece from the fish tank. I hope one day to be as amazing as my father was.

Reminds me of the good old ST506. Lost quite a few files to the gummy lube on that drive.

Haha yeah. We used to call that “percussive maintenance”.

Could also have been a loose cable. Completely unplug (both ends) all the cables, and then plug them back in. I've had cables come loose over the years, especially back in the IDE days. It's easy to overlook, because the cables can look like they're properly seated.

I checked that, and used canned air to blow out the sockets before reconnecting everything. It was honestly just a case of internal stiction that needed to be freed before the drive could spin up.

I've had it happen to me a few times with SATA cables that don't have a clip (or the socket can't use it, like before SATA 300). IDE sticks in there so well that I think I'm going to damage something when pulling a cable out.

I remember my mother calling me with a machine that was making horrible noises and shutting down randomly. I told her to kick it. It took a few minutes for her to believe me. And she did. And the noises went away and it worked fine. I don't think I ever explained to her why I prescribed that remedy...

The old Apple IIe’s used suffered from an occasional manufacturing defect that would cause the chips to be slightly disconnected from the main board. The recommended fix was to pick it up a few inches off the desk and drop it to re-seat the chips.

Apple ][+ too. In addition those 72x chips would fail with time. We figured out that a can of compressed air was handy.

Power on. Wait for machine to freeze. Spray a chip with compressed air to cool it. Try again. If it stayed working it was probably that chip.

Thank goodness they were socketed.

That was the Apple III.

That reminds me of the story about the magic switch: http://catb.org/jargon/html/magic-story.html

Great story though you could have just pulled out the switch and soldered the wires together :-)

I look forward to the practical soldering test for developers at FANG's - every one on here can solder right.

That was my reaction too. Maybe I just work much more in hardware, but if I came across that issue, I would have just pulled out the wires for the switch and soldered a new one on there.

Is there something I am missing?

I know a guy who used a machine with case off, and turned it on by touching the power button legs with a screwdriver...

I did that for a couple of days when the power button cable was just too short to reach the header.

Just that it's funny

Fair enough. I should probably just enjoy a story for a story.

No solder needed, the front power switch plugs in to the motherboard using a typical DuPont 0.1" connector.

The odds of knackering something else on an irreplaceable machine while you're in there. Yes, it's small. No, it's probably not worth the risk.

Well the part you're missing is that if these folks were anything like me and just pure software people they probably pronounce it "soul-der" and would look at a soldering iron with deep concern and confusion.

Seriously, I have no idea how to solder so it wouldn't even come to mind.

This is why I think CS degrees should require at least EE 101 or Computer Engineering 101. Soldering is such a basic skill that is super valuable. And it's not hard:

1. Plug in soldering iron

2. Wait

3. Touch tip of iron to metal you want to fuse with 1 hand

4. Touch solder to tip of iron with other hand

5. Remove iron once desired amount of solder has melted into place

It only takes an hour or two to go from solder noob to pro.

It wouldn't help me very much lol, I have a philosophy degree.

Let your degree define what you can do. Never let it define what you cannot do.

I'm a mathematician, doctor of philosophy, and my mom taught me to solder at around age 10. I've done some embedded systems work that involved etching my own PCBs and soldering circuits...

I agree with the other comment! The best time to learn is now. You're never too old.to learn about how all of this works.

Oh I agree, but I've never really been a hardware tinkerer though... that's not what I get enjoyment from so I don't really think about that stuff too much.

I'll probably pick it up eventually... maybe.

Side note, Soul-der is the correct pronunciation in the UK (and I believe also AU/NZ).

No soldering required, the motherboard wants a momentary contact to tell it to boot. Soldering the wires together would be more like pressing and holding it.

That’s true for ATX cases, but remember that this is the build machine for Excel 97 East Asian Languages. It’s probably an AT case.

Ah, TIL. So AT cases had a maintained toggle button on the front that had to stay depressed, and the computer shuts off when it's released? I suppose that makes more sense for the screwdriver keeping it powered on. I was figuring it was something like a momentary button sticking and the long hold causing a force shut down.

Not only that but the switch toggled line voltage that fed the power supply. Off meant off in those days.

Ah, the first computer I remember had the hard power switch on the back and a button on the front to turn them on, but that would've been early/mid-90's.

The ATX standard was released in 1995 and it took a few years for them to become common. Are you sure it wasn't the late 90s. Time-frames get hazy this far back, its actually not improbable these Office '97 machines are ATX.

It ran System 7 which puts it somewhere between 1991 and 1997. Not super helpful.

But Apple might have been off doing their own thing with respect to motherboard form factors and power buttons. And I could also be mixing it up with later computers.

Apple did their own thing and didn’t follow ATX. Most Macs had soft power switches (often on the keyboard) except for the very old ones with the integrated CRT.

> I look forward to the practical soldering test for developers at FANG's

This seems equivalent to whiteboarding algorithms that you'll never use.

FWIW, I'm 42, have built and repaired many, many computers over the years, dating to 286's, and I've never picked up a solder iron.

I'm also 42, and just recently starting picking up the soldering iron for Arduino projects. Never needed it in two decades of software development or general computer maintenence, including quite a few builds.

As they say, the only way to know you'll never need something is to learn it.

I'm not older than you and I've had to solder several times at work.

Heh, thanks for making me feel old. I was repairing Apple ][s when you were in the womb.

I wish more people would make me feel young here :-)

Office 97 actually gives me a little bit of nostalgia. It's the last version for me that actually felt pleasant to use (assuming you turn off Clippy, which isn't much of a barrier).

>assuming you turn off Clippy, which isn't much of a barrier

It was a high barrier for my young self.

When I was a little kid and started using computers for the first time, Office 97 was one of the only programs installed on my family's PC. I swear the only reason I ever opened Office was for Clippy and its fellow animated characters :)

Peak Office imo is whichever Office version introduced that blue underline that let you immediately undo whatever frustrating autoformatting catastrophe was just inflicted by hitting the tab key or resizing an image.

I think it was 2002?

Everything afterwards feels like change for change's sake. But I haven't used any version after 2016, thankfully, so maybe they finally got collaborative online editing or less obtuse revision tracking.

The fun part is that there is def a disconnect between the dev team and hardware itself (or help desk). To resolve the issue, you only needed to short two jumpers on the motherboard to get it to turn on.

If it was a NON-ACPI hardware, then the hardware switch was wired into the power supply itself, which, then again you could've undone the cable and connect the circuit to boot the computer.

Yeah I used a computer for years with no case, to turn it on all you have to do is briefly short 2 pins on the front panel connector. Although once inside a case it can be pretty awkward to access.

This is how you start a pc today when you are diy building it.

The first stage is testing that the MB will boot to bios outside the case

hmm not a bad idea, I've never done it, I always assemble inside the case. And then after the loud beeping sound, check around to find which connectors I've missed.

Back when I had some Bitcoin GPU rigs (2011), my setup used 2 wires connecting those jumpers to a cheap push button switch.

I would imagine that a build machine for Office 97 would be of the pre-ATX days and have the switch wired directly to the power supply. Like a clicker-style pen, those power buttons were often push-on-push-off, and occasionally the mechanism would break. Jamming something, like a screwdriver, into the power switch hole would be a completely acceptable thing to do for a one-off.

It probably took a little time for the initial panic to wear off and allow the folks doing the build to realize it was a simple matter. The thought "we may not be able to" is a bit of a reach. At worst they could swap power supplies or whatnot.

Doesn't that mean the switch was working with mains electricity though? In general, I'd recommend against sticking a screwdriver into anything 120v AC

Extremely unlikely. It was probably 12V. Just like the switches today are 12V and you can 'power on' an ATX power supply by jumpering 2 wires on the main 20/24-pin ATX cable.

No, pre-ATX power supplies did have their main and only power switch on the 120V AC side. And when the power switch was on the front of the case, the wires going to that switch carried the full 120V AC voltage and current.

Turning on the power supply by shorting a pin to the ground (it's not 12V, it's 5V, pulled up to the always-on 5V stand-by rail by the power supply) is an ATX novelty.

Can confirm. I shocked myself a few times by bumping into the terminal of the 120V AC wire at the power switch on the front of a case. As long as it's plugged in to the wall, one side of that switch is hot even when the PC is off.

You can manually make it always on by just cutting the switch and splicing the wires together. That's probably what fixed the button - jamming a screwdriver pushed the switch/contacts together.

In high school, circa '97, I was working in a computer lab. As I entered the room, I was rapidly admonished "don't touch the table!" So naturally, I reached out, tenderly as possible, brushed the table with my fingertip. Sure enough, the computer immediately reboots. He cusses a blue streak, as he's been trying to install Windows but folks kept touching the table! I regret not talking him through the issue, because obviously nobody would use such a flaky machine... but I'd (quite reasonbly) lost his goodwill in my initial mischeviousness. To this day I don't know how the machine was so sensitive to, presumably, a miniscule change in capacitance. There's a possibly that it wasn't grounded and I was carrying a nontrivial electrostatic charge, but the table surface was insulating!

Turning off the computer or putting it in suspend mode is somewhat expected nowadays unless it is a server in a rack that runs untold virtual machines that get 'spun up' as needed.

Back in 1997 the idea of turning of an specialist idle PC that consumed company electricity was not the done thing to do. If you were a lowly office worker using Office 97 or its ancestors then you would turn off your machine at the end of the day and start the day with a ten minute boot time. It took a long time to get these efficiencies right and a silly amount of time was wasted. CRT monitors ruled the roost back then too, after 6 you would see lots of silly Microsoft OpenGL screensavers running.

I wonder what the actual power consumption is these days compared to then? It could be a factor of ten.

I'm surprised there weren't backups they could've restored? If the difficulty was setting up the build environment correctly, then even a duplicate hard drive for each build server would've saved them from a pinch.

Office 97 is older than my undergraduate students. That's so weird.

The amusing but strange sub-story here is the build engineer claiming that the product may never be built. That is a type of endemic first class whining the type of which I am sick, sick, sick of.

To me, that appeared to simply be effective communication. Raising a potentially serious issue early rather than waiting until all options have been exhausted and giving the security lead no time to react. Had they not been able to succeed, the advance notice could help in getting other resources moving in advance of the internal or external deadline.

Inability to power on a build machine is not a serious issue. It is a temporary problem with many solutions. To suggest that the build may never happen would imply a much more serious problem. That does not seem like effective communication, it seems like reckless panic to me. Perhaps they thought the motherboard was fried. Even then it would be irresponsible to suggest that the build could not happen. Certainly communicating that there is a problem is a good thing, but offering up the possibility of no resolution when the diagnostic is only just beginning is not.

And if the issue was that the machine ate a power surge at some point and the hard drive (+other components) was fried? It may have no longer been practical to recreate and QA that build machine...

Ok, but please don't post unsubstantive comments here, and especially not ragey ones.

You could make a good comment out of this by replacing the second sentence with some more specific observations.

Amazed nobody thought of virtualizing the old hardware.

In 1997?

This wasn't in 1997. Article is talking about patching Excel 97 years after it came out, which is why all the build machines were dormant and turned off. Based on when the team moved (follow link in story), this would have been after 2007.

That predates Hyper-V, and I suspect qemu or virtualbox would have been a challenge.

VMWare, though...

I used VirtualBox heavily in 2007, worked flawlessly

Reply on news.ycombinator.com