If You Type 1+2+3 Into Your iPhone's Calculator on iOS 11, You Probably Won't Get 6 (qz.com)

A reader shares a report: If you've upgraded your iPhone's operating system to iOS 11, try this: Go to the calculator app and quickly type 1+2+3. You likely won't get 6. You might get 23, or 24, or 16, or 32, or something else, depending on what buttons you tap and in what order, and, obviously, none of which is the right answer. It seems to be because of a new animation in the calculator app, where a button briefly fades to white when you press it. The result is that if you press an operator button (i.e., the plus sign) before the short animation finishes, the app ignores it. So, 1 + 2 + 3 accidentally gets read as 1 + 23.

Comments

First CS assignment.

[Anonymous Coward]

After a basic hello world intro, I had to write a calculator to add subtract multiply and divide in the first week of college. Had mine worked like apples Iâ(TM)d have likely failed. How does this happen?

Re:First CS assignment.

[Anonymous Coward]

Math is hard.
In Windows, open the calculator and do this:
Take the square root of 4
Subtract 2
You should get 0, but you don't.

The calc app does some sort of weird floating point math (despite the fact that the square root of 4 is exactly 2 and no floating point math is needed) and comes up with an answer of

-1.068281969439142e-19

Math is hard

Re:First CS assignment.

[H3lldr0p]

Someone skipped the day(s) they were teaching about how to do significant digits in Numerical Methods.

Like they had to skip a lot of days because that was the follow up conversation we had after each major section. As in how to tell what is and what isn't a good answer to the math we were just given to do.

Re:First CS assignment.

[bjohnso5]

The Win10 calculator just gave me this result: -8.1648465955514287168521180122928e-39

Re:First CS assignment.

[hawguy]

After a basic hello world intro, I had to write a calculator to add subtract multiply and divide in the first week of college. Had mine worked like apples Iâ(TM)d have likely failed. How does this happen?

You didn't have a GUI keyboard with animated keys.

This is why I still keep my trusty HP-15C calculator on my desk -- for when I need to do some quick arithmetic, it's still easier to use the calculator than my computer or worse (as this example shows), a phone.

Re:First CS assignment.

[nine-times]

How does this happen?

It explains how it happened in the summary:

It seems to be because of a new animation in the calculator app, where a button briefly fades to white when you press it. The result is that if you press an operator button (i.e., the plus sign) before the short animation finishes, the app ignores it. So, 1 + 2 + 3 accidentally gets read as 1 + 23.

It's not doing math wrong. It does a brief animation when you press a button, and it doesn't necessarily read the next button you press while the animation is happening. Therefore, if you press the calculator buttons too quickly, it won't register all the things you pressed. Based on the buttons you do press, it does the math correctly.

So if you press "1+2+3" it might miss the second "+" and register "1+23" and give you "24" as an answer.

Re:First CS assignment.

[cheesybagel]

Typical Apple. Form over function.

They already made smartphones which couldn't make phone calls, so I guess this is just one more instance in that vein.

Re:First CS assignment.

[sexconker]

It gets even weirder than that!

4 @ - 2 = -8.1648465955514287168521180122928e-39 (@ is the shortcut for clicking the square root key)
4 y .5 - 2 = 1.0605907030850721689734498566293e-38 (y is the shortcut for clicking the "x to the power of y" key)
4 y ( 1 / 2 ) - 2 = -8.1648465955514287168521180122928e-39 (y is the shortcut for clicking the "x to the power of y" key)

It even shows "0.5" after you type the closing parenthesis when doing 4 y ( 1 / 2 ). Yet the result differs compared to 4 y .5 - 2 (or 4 y 0.5 - 2).

WTFOOK MS?

Re:First CS assignment.

[omnichad]

There's nothing wrong with animation. It gives feedback the same way as the click of a key on a physical keyboard. It's only the implementation that's at fault here.

Re:First CS assignment.

[Aighearach]

I'm using Mathdroid, which you can get from the fdroid repo. It is easily my most used mobile app. Scrollback buffer, variable support, hitting a bare operator automatically inserts the last answer in front, etc.

Actually, it works better than any of the linux calculators I got from a distro, ever. HP-15C is nice, and I'd probably use something similar if I didn't have Mathdroid.

Re:First CS assignment.

[fahrbot-bot]

Typical Apple. Form over function.

They already made smartphones which couldn't make phone calls, so I guess this is just one more instance in that vein.

They also don't grow, or even sell, apples.

Re:First CS assignment.

[fahrbot-bot]

It isn't a problem with it doing the math. It is a problem of capturing the key press. Being that the button decided to do some fancy animations I expect it just prevented the next plus button from being registered.

It takes courage to provide an input buffer.

Re:First CS assignment.

[v1]

As a developer I can see exactly how something like this can happen, and I've had variations of it happen to me numerous times over the years. This is a case of the person writing the app making assumptions about how the user interface behaves, and these assumptions turning out to be wrong OR the interface not behaving as documented. (more likely the latter)

The person that wrote the app was probably "safe to assume" that taps on the screen will be buffered during animations and other gui actions. So the app waits for you to tap, and the user taps a coordinate on the screen and your app is notified of this. You look at the location on the screen, figure out the user pressed "2, remember it, and tell the gui to do the fade on the "2" to indicate the press was accepted. Then you go back to waiting for another tap.

If in the meanwhile, the gui is displaying the fading of the "2", the user taps "3", the interface should buffer the tap, and it can either notify the app immediately (during the animation) of the new tap, OR it can wait until the animation is done and then notify the app.

I don't know which approach the gui takes. If it notifies immediately, I would expect the app to tell the gui to stop the "2" animation and start the "3" animation, as this leads to better response time. SOME apps I've seen just linearly buffer and the app won't be notified of the "3" until the "2" is done animating. In those cases you can mash a bunch of keys and then look at the screen as they are animated one at a time until all keys are animated. This is generally considered poor design but I still see it from time to time.

So what went wrong here? There's several possibilities, depending on what the gui supports.
- the gui always disables tap detection while the specific sort of animation used on the "2" is being done
- the developer used an animation api that specifically does not buffer taps when there was a buffering api available he should have used
- the application told the gui to disable buffering during the animation (or neglected to indicate it wanted buffering, if it's not the default action)
- a bug in the app caused it to not buffer taps made while the gui was still rendering an animation

There are probably a few other possibilities but that's the 99% of it. Could be a bug in the API, could be a documentation problem with the API, could be the developer used the wrong API, or used the right API in the wrong way, or got all the information he needed but just bumbled it and lost it in this specific circumstance. There's just no way to know who to blame, and I wouldn't care to take a bet either way.

But really... a user interface bug in a calculator app makes front page news on slashdot? Must be a slow news day?

Re: First CS assignment.

[BronsCon]

When AC says Honda, I'm assuming he means the in-dash navigation and radio. Imagine, for a moment, that you think you hear sirens, but you don't see flashing lights, so you decide to turn the radio down to hear them better. You can't really wait for a safe place to pull over to do that, since you may end up impeding an emergency vehicle before then, so you're stuck fumbling with the damn infotainment system and waiting for it to animate every... single... button... press... before... it... will... do... anything... else.

By the time you've turned the damn thing down enough to hear, the ambulance or fire truck is up your ass, because you had to take your eyes off the road (and your rear-view mirror) to look for the spot on the screen to touch for volume in the first place.

Can we just have fucking physical knobs and buttons back, please?

Re:First CS assignment.

[Anonymous Coward]

Typical Apple. Form over function.

It's actually the other way around - this happened for functional reasons. UIKit-based animations have an option to disable user interaction during an animation. This is switched on by default, and it has been since the very first version of the iPhoneOS SDK. It had to be switched on by default because otherwise, users who were used to double-clicking to activate things with mouse-based interfaces would trigger actions twice when double-tapping on buttons (and yes, if you watch non-technical people, absolutely loads of them double-click and double-tap even when they don't have to). Auto-disabling user interaction during animations is the smartest default.

All that's happened here is that somebody in Apple forgot to switch that option off, and it's the kind of thing that's easily missed during testing. After all, despite what people say, this isn't new to iOS 11 - the bug has been present for a couple of years now and it's only just been noticed.

Re:First CS assignment.

[dacut]

How does this happen?

It explains how it happened in the summary...

I think the "this" refers to "shipping an obviously untested product."

Part of the problem is that we (collectively) just don't get how complex software is. Sure, a good software engineer who sits down and thinks through the implications of a change will do so, but in the modern rush to market that's a rare happening. In this case, I'm guessing something like the following happened:

• "Hey, we're getting usability reports about widgets accidentally being double tapped while an app is being swapped in. What can we do to fix this?"
• "We would need to do this in the UIKit. Maybe some visual indication it's been tapped with a lockout period?"
• "Ok, great. We need this by next Friday to get it into the next iOS build."
• "That's tight, but we'll manage."
• Builds, tests... "Ok, let's see what apps use this. Calculator..." tapping deliberately "Right, looks good, results are correct. Ready to ship."

That said, much as we (software engineers) don't like it, there's something to be said for shipping quickly and, sometimes, before things are ready. Users reward this behavior, and shipping quickly can mean the difference between a product or company that succeeds vs. one that fails. The saying is "you can't shine shit," but I've seen countless examples otherwise (and the Mythbusters disproved this in, well, a literal sense). I hate it; despite it, in fact. Push back against it. But it's hard to argue when there's a throng of consumers ready to spend their money on it.

So, we're relegated to having to be judicious in what we push back against. Is it safety critical? Will someone get maimed or killed by this? If it's running on an iPhone, the answer is probably no. For Apple here, the main result is a bit of embarrassment -- a calculator that seems to give wacky results. A civil engineer using this for estimating should see this kind of issue immediately (and is unlikely to use the iOS calculator for final documents). Someone trying to split their restaurant bill, maybe not, it's an acceptable risk. The Excel 2007 multiplication bug was probably more serious because that's an application that is more likely to be used in civil engineering. If you're writing software that could lethally irradiate someone and encounter shady practices, immediately raise flags and alert everyone who will listen.

Ok, so we're not going to manually test everything on a smart phone before a new OS is released. What can we do? Push for more automation. Making aesthetic judgements automatically might still be a bit difficult, but we ought to be able to simulate key misregistrations. A quick check would be to do this in software to see the effect of a change while you're hacking away on your laptop, but a robot providing millions of taps and swipes on actual hardware would be even more insightful. This would be daunting for most startups trying to make the next Zynbookwitter on shoestring VC funding, but child's play for the likes of Apple or Google.

Re:First CS assignment.

[hey!]

This isn't a math problem. It's basic programming; you don't rely on tests of equality between the results of floating point calculations and integer quantities.

What's going on here is that the sqrt function yields a floating point number that's close to the integer 2, and that's rounded for display down to 2. The round for display routine clearly chooses not to round if the difference is large relative to the amount. For example take the number 1000, and add .00000000001. It will display 1000.00000000001. Now do the same with 1,000,000; it will display 1,000,000 even though internally the number is 1,000,000.00000000001.

It's not an entirely unreasonably way of doing it, although most people would tend to round to a fixed number of digits.

Re:First CS assignment.

[hvidstue]

Win10 calculator does the same thing

Re:First CS assignment.

[Paradise Pete]

It's a bug. A stupid bug, but I'm sure it wasn't intentional and that they'll fix it.

Of course it's a bug, but how do you not see it within the first five minutes of testing? When I saw this earlier I thought the claim must be ridiculous, but it's actually more difficult to avoid this problem than it is to demonstrate it. I like my Apple stuff and I get a new iPhone every year, but iOS 11 and Mac OS High Sierra are rife with small disappointments and annoyances like this.

Re:First CS assignment.

[fisted]

But does systemd-calcd compute the right result when you IPC it the math term via dbus?

Re:First CS assignment.

[Lanthanide]

Windows 10 with the calculator in Scientific mode gives me -8.1648465955514287168521180122928e-39

When I put it on Standard I get your number. Interesting that there's more to Scientific mode than simply extra buttons.

Re: First CS assignment.

[Anonymous Coward]

It may take a few lawsuits to get back to an interface which actually works while driving without requiring the user to take their eyes off the road.

Since we had a well proven system for decades and the replacement is defective by design because it's unsafe, hopefully it won't take a few juries long to award big bucks in a few high profile cases to get manufacturers to fix the error of their ways.

Normally, I'm not particularly plaintiff friendly in product liability lawsuits, but in this case the manufacturer better show me if I'm on the jury that their design decision SAVED more lives than it cost and that they could not have reasonably achieved the same life saving feature without costing (fewer) other lives. That's a pretty high bar to clear!

I actually drive an older car now and one reason is just this -- I looked at replacing it and found the UIs to basic functions in too many midrange cars to be unsafe.

Re:First CS assignment.

[techno-vampire]

And for that matter, in a perfect world, the testers wouldn't be told how the devs expected the software to be used. Instead, they'd be expected to mess around with it, use it in unexpected ways (not that typing too fast should be unexpected) and try to make it fail.

Re:First CS assignment.

[DontBeAMoran]

I have the Win10 Creators Edition and the calculator gave me this result: "I'm sorry Dave, I can't do that".

I knew Win10 was crap, but come on. My name's not Dave!

Re:First CS assignment.

[Obfuscant]

Wow, I could reproduce it easily in my updated 64-bit W7 HPE SP1 machine. Why hasn't MS fixed it?

The hatred for Microsoft runs deep in this group, to ignore the fact that the calculations are actually being done by an IEEE-754 compliant floating point processor in the Intel or AMD CPU, and not by Microsoft doing it in their code.

What's fun is having an 80 bit IEEE processing unit on a 64 bit CPU. You can get different answers depending on whether the calculations are done entirely in 80 bit FPU or 64 bit CPU.

So, it isn't Microsoft's failure to fix. It's a known limitation of using a digital computer to "do math".

You might pipe back that the computer could be written to work in arbitrary precision binary coded decimal, but that leaves the problem of errors at the edge of the precision, no matter what that precision is. Even "arbitrary" doesn't mean "unlimited", both for speed and hardware constraints.

Re:First CS assignment.

[Obfuscant]

So, how are we expect to get the true math answers with these calculations then? :(

If you want "true math answers", do math symbolically instead of numerically.

There are entire books and courses on numerical methods that talk about doing calculations on numbers instead of symbols, how to minimize errors, and why they appear.

Re:First CS assignment.

[Golden_Rider]

The issue is that the number buttons in the calculator behave differently from the operator buttons. If you look closely, you will see that if you hit a number button, it will fade out immediately after being clicked. But if you e.g. hit the "+" button, it will STAY highlighted, probably to remind you of what operation you selected (and that you actually HAVE selected something) while thinking about what number to put in next. Only if you then hit a number button, the "+" button will fade out, which takes about half a second, and while it does this, it does not register another click on that specific button. So you could put in 1+2-3, but you cannot put in 1+2+3 without first waiting for the "+" button to fade out completely.

Re:First CS assignment.

[Misagon]

You don't round to display in a calculator in the first place. What you get in the calculator display should reflect the true value in the accumulator.

Also, if sqrt(4.0) does not produce exactly 2.0 then there is something wrong with that code. 4.0 and 2.0 are powers of two, so there should not be any rounding when using IEEE floating point math.

You're typing it wrong.

[aix tom]

Although I wonder if the fix would be just to be brave enough to just remove the calculator. Maths is an outdated technology anyway...

Re:You're typing it wrong.

[Killall -9 Bash]

Math isn't outdated. The calculator interface is outdated. You should just ask Siri what 1+2+3 equals.

the rise of the eyecandy tards

[iggymanz]

UI and software quality is falling because of the emphasis on appearance rather than function, hence Unity and GNOME 3 and this article's stupidity.

Re:the rise of the eyecandy tards

[AmiMoJo]

Nah, it's just standard Apple quality control. Remember when the alarm clock stopped working? Multiple times? Over multiple years?

Apple Maps, the iPhone 4 antenna, Macbook thermal paste... Apple is just as prone to fuck ups as everyone else, it's just that theirs tend to be hilarious for some reason.

Re: the rise of the eyecandy tards

[Brockmire]

To me, nothing tops GotoFail. Though, I still believe that was intentional since their explanation of patching caused it doesn't fail spectacularly like they claim. But it does mean, their development and QA process is shitty.

Re:the rise of the eyecandy tards

[BronsCon]

Well, given that the feedback in this case is a button animation, I must be looking at the screen in order to see it... so the digits on the calculator display changing might give me some clue that my input was registered. Beyond that, the haptic feedback already provided by the phone does a fine enough job of letting me know my input was received.

Animation is cancer

[Anonymous Coward]

GUI animation is the absolute cancer. It's everywhere, on desktop, in applications, on websites.
Really feels like everything returned to the 90's web state full of animated GIFs, blinking text and endless pop-ups.
If you're a developer, please don't be a jerk and get rid of all the animated crap, please.

Re: Animation is cancer

[Megane]

Can we add smart quotes to that list?

Only apps can app apps!

[Anonymous Coward]

Only LUDDITES want LUDDITE accurate numbers! Modern app appers want appy numbers that let them app apps while apping other apps!

Apps!

Lesson: Blocking UI is really bad

[SuperKendall]

It's pretty funny really, because Apple makes a big deal about how app developers are not supposed to block UI, and about how to make animations interruptible. The fix will probably be pretty simple...

The calculator issue is really bad though. Even just moderately fast pressing of buttons yields input blocking depending on what you are doing.

Gimmicky animation is such a waste of CPU...

[acroyear]

CPU time and network time should be better spent on the really important stuff that computers are supposed to do: check for app and o/s updates.

Re:Gimmicky animation is such a waste of CPU...

[rudy_wayne]

CPU time and network time should be better spent on the really important stuff that computers are supposed to do: porn.

fixed that for you.

Unable to reproduce.

[Kenja]

Bug status : rejected.

Re:Unable to reproduce.

[93 Escort Wagon]

I have an iPhone 6S on iOS 11.1 beta - and this bug is damn easy to reproduce. Why on earth does a freaking animation get precedence over a button push? More importantly, why is it even blocking at all?

I'm old enough to remember typing on remote CRT terminals which were connected to a central computer over a 300 baud line (or maybe it was 110? This was back ~ 1980-1981). Back then, if you typed reasonably fast you could get ahead of the terminal's display by a few characters... but even way back then, this was a solved problem, those additional characters didn't get lost.

New Apple

[DarthVain]

Here is a 1200$ super computer that fits into your pocket! Unfortunately it can't do math.

Re:Windows Calculator

[sqorbit]

No, if you type 3+3 and then hit x you get the answer of 3+3 = 6, then you "x" and "3" you get 18. You are asking the calculator what is 3+3 and then asking it to multiply that answer by 3. This is working as designed because it's not a scientific calculator. If you simply switch to Scientific in the options, you can use "( )" and/or other functions to put in a formula. As far as I know the standard Apple calculator does not a scientific option. The user is in error if they want 3+3 * 3 in a non scientific calculator and start hitting buttons.

gmail

[Misagon]

Reminds me of gmail.com's login form which has a similar bug.

If you type username and then Return, the Return does not immediately switch focus to the password field - it only starts an animation and passes focus to the password field when the animation is done.
So, if you type your password too fast, the first few characters will not end up in the password field (or not at all, if your password is short).

Bugs the hell out of me. The older login form did not have this bug.

Re:I got 6

[clonehappy]

I just tried it too. You have to go out of your way to do things fast enough to beat the animation, and even then if you type 1+2+3 fast enough, it will just show "23" on the screen so you'll know there was an error.

It's not like the display shows 1, then 2, then 3 but still adds 1+23. It's a minor annoyance I suppose, and should be fixed in an update, but not at all worthy of an article on Slashdot.

Re:Windows Calculator

[MiliusXP]

On Win 10 standard calculator ==> 18 Scientific calculator ==> 12

Re:Windows Calculator

[mr.hawk]

As far as I know the standard Apple calculator does not a scientific option.

Sure it does. Just put your phone into landscape mode.

Reminds me of the TI debacle of the early 80's

[timholman]

Texas Instruments had a similar screwup in the early 80's. After capturing a huge chunk of the U.S. calculator market (TI and HP were the brands to buy), Texas Instruments released a series of lower-cost scientific calculators where the keys were not properly debounced . It was practically impossible to type in a long equation without having multiple double or triple press errors.

I tossed mine in disgust, tried out one of the new Sharp scientific calculators just hitting the market, and never looked back. Texas Instruments basically handed over their share of the scientific calculator market to Sharp and Casio in the space of two years.

At least Apple has the advantage of being able to fix this in a software update.

Re:Windows Calculator

[Obfuscant]

How strange.

Not strange at all. That is zero within "eps", and is because they are using the Intel or AMD math processor for the square root. WE look at "4" and know the square root is exactly two because we learned that. The CPU goes through a standard algorithm for determining the square root of a number, and because of the inherent imprecision of floating point math with a limited number of bits, the answer is not identical to zero because the square root of 4 is not identical to 2.

Interesting historical note:

[hey!]

This is similar in a way to the bug behind the famous Therac-25 incident. The Therac-25 as a medical radiation machine which had software which was supposed to prevent patients receiving dangerous doses of radiation. However it turned out the operators entered configuration command far faster than testers did, creating a race condition that could result in the machine delivering over 100x the safe dosage.

The bug never showed up in testing because the testers never got as fast at input as the operators, and in any case the specific keystroke combination that caused it was rare.

EXACTLY PRECISELY, UTTERLY

[AbRASiON]

Why nerds like us who are "sperglords" disable animation and flashy bullshit in applications and operating systems, ever since windows XP.

I've been abused countless times for it, but at the end of the day, the goddamn computing device should keep up with ME. Not the other way round. I'm extremely fast on the keyboard, I know endless shortcuts, I can get things done quickly, but I need responsive tools.

But nope, we're just "sperglords" and neck beards who should stop complaining....

Re:Windows Calculator

[pz]

No, sorry, that's just wrong by the basic rules of algebra. Multiplication gets precedence over addition. (That newfangled idea of a precedence tree in computer languages is predated by at least a hundred years in mathematics.)

3 + 3 * 3 is 12, period. If your calculator is giving you 18, it is wrong, and is the result of a lazy developer not understanding the appropriate operator precedence. Doing correct arithmetic is non-trivial (it isn't hard, but it isn't trivial: the trivial approach gives you 18 in this case, and that is wrong). This is not a question of "scientific" or "non-scientific" calculators. There is no such thing. Calculators have one kind of function, there is one correct answer,and if they don't get it right, then the designer or implementor is at fault.