What was the weirdest bug you ever encountered?

As developers we find and troubleshoot bugs everyday. What was the strangest bug you've ever stumbled upon?

Comments

[Dave Cridland]

Personally, the weirdest and hardest to solve bugs for me are when the C++ optimizer takes a shortcut that ends up ignoring my code. These are usually cases where, for example, a pointer cannot be NULL according to the standard and therefore my NULL check is ignored.

I know of two other bugs which are gloriously Byzantine. I don't know Trey Harris, but his bug is legendary: The 500 Mile Email

I do know (and used to work for) Dave Baggett, and have heard this story first-hand. My Hardest Bug Ever

[Michel Renaud]

Those optimizers can be sneaky. I don't know how long you've been doing this, but I remember when the first C and C++ optimizing compilers for PC came out (late '80s, roughly) that they did a lot of nasty things and you had to be very careful with what optimizations you picked. Let's just say it was a rocky start.

I was on a project with some C++ just three years ago and it's a good thing my supervisor was borderline genius because he caught VC++ doing some really strange things. The ways he figured it out was beyond the skills of pretty much everyone else I know.

[Dave Cridland]

I've been using C/C++ since around 1990 or so. The old problems were usually weird compiler bugs when optimizations went wrong dramatically, from what I can recall. But I was thinking more of cases like this, which tripped me up recently - I'll simplify, of course:

int * p = nullptr; // Start with a NULL pointer.
int & r = *p; // Later on, convert to a reference, probably in a function call.
int * i = &r; // Convert the reference back to a pointer.
if (i != nullptr) { // If it's not NULL
  std::cout << "i is now " << *i << std::endl; // Print it.
}

That code crashes with a segfault trying to deference a NULL pointer. This is because the standard says that if the address of a reference is the NULL pointer, it's Undefined Behaviour. So most modern compilers now choose to optimize away the test, even without optimization turned on...

[Michel Renaud]

"Undefined behaviour" is hell.

[Andreas]

Thank you so much for sharing! The 500 mile email is indeed legendary! 😂

[Darren Burns]

That time I snoozed my iPhone alarm and instead of ringing again when it hit 0, it continued to count down past zero into the negatives. It's happened about 4 times now.

[Michel Renaud]

Snoozing for 44 minutes? Get up! lol

[Andreas]

Awesome 😂 I'm wondering how long it would go on this way 😅

[Mark Freedman]

That's the new time travel feature. Apple is ahead of their time... um... literally :)

[Andreas]

Hearing a crackling noise in my headphones when touching the fingerprint sensor on my Android phone.

[Javier ⚛️]

Lol that was weird

[Andreas]

Indeed it is! And it is reproducible. The sensor must somehow produce an electrical charge that's affecting the jack connection 😅

[Peyton McGinnis]

I get this in my PC a lot. I'm pretty sure the reason is that unless you have a dedicated sound card a lot of motherboard noise is likely to pass through the audio jack.

[Paceaux]

The Stacking Context in CSS. More specifically, that there are a few CSS properties that can trigger a stacking context. And even more specific, that particular properties, in particular browsers, will give you very different results within that stacking context.

It'd started with a simple request to remove a double-scroll bar that was displaying in certain browsers. Over the course of a month I reviewed every CSS property on every element until I discovered that we'd unintentionally created a new stacking context, and that CSS we'd written accidentally handled that very well... until it didn't.

I ended up editing the mozilla developer network page on stacking contexts, and producing the most boring codepen ever that illustrates exactly how certain CSS properties totally bork stacking contexts.

took a month to discover that no one on the internet had, until then, documented that perspective would change the behavior of position:absolute. That double scroll bar was the hardest bug I ever tackled, but is also one of my proudest moments.

[GjeloshajAntoine]

I've never heard of stacking context in css!
What is it?

[Paceaux]

The stacking context is the ... well "context" in which the browser determines how to layer things. We hardly ever think about stacking context and it hardly ever matters.

But stacking context is about paint order ; when the browser has elements that overlap, the browser has to know which element to draw first, and which to draw next.

contrary to popular belief, z-index hasn't anything to do with positioning but with painting. When we have overlapping elements, and the browser is painting them in an order that we don't want, we use z-index to change the order in which the browser draws them.

So, that means that "the order in which the browser draws them" has to be determined some how. So what the browser does is find "root contexts" to determine which elements need to have their drawing order sorted out. Think of these "root contexts" as buckets with sheets of paper.

The bucket is any element with position:fixed, absolute, perspective, transform, etc. When the browser encounters those, it says, "aha! I need to make a bucket with this, and then figure out the order in which to paint things". That's your stacking context.

You can live a happy and fruitful career never having to think about this.

But occasionally you will discover that the browser sometimes has some strange rules about how it creates stacking contexts. Sometimes the browser has to change the rules of the bucket just to figure out how to paint those sheets of paper.

A great example of this might be an element that's position:fixed, with a parent container that has opacity: .2.

That position:fixed child element is _supposed to be` wherever you intend it to be, relative to the browser window.

But you see, it's inside of an opaque parent.

The browser is supposed to show just a little bit of the element underneath that parent, because the parent is opaque. That means the grandparent is supposed to influence the look of the grandchild.
But how can it do that if the grandchild is position:fixed and nowhere underneath this opaque parent? So the browser makes a decision:

I will create a "Positioning context" that matches the stacking context.

So what happens is position:fixed breaks . It's no longer relative to the window, it's relative to that parent that has opacity: .2.

Again, you can live a fruitful live never knowing about the Stacking Context because it mostly never matters. But there will be rare times where the browser can't reconcile these CSS properties where layering matters with other CSS properties where positioning matters, and it creates a new "positioning context" that matches the stacking context.

[Andreas]

In short: positioning and z-index. Paceaux provided a link with a detailed explanation. Here is an example:

[Glenn Stovall]

I used to work on an application that colleges used to process certain types of applications. College could create their own forms. When the user was done, it would compile the answers into a PDF and mailed out to relevant people.

For one client, the PDF would come in empty, but only 10-20% of the time. We couldn't figure out why.

I spent three days tracking down a bug in legacy code that ended up being a bug in PHP itself. . Our system did a lot of parsing of the data to move in between different formats. When you did a Regex find/replace, the method should return the updated string. If there were no matches, it should have returned the original string.

BUT

If you passed in a string over a certain length, and there were no matches, it would return null instead. This one client ask a lot of essay questions. The script would only fail for the most verbose students.

We couldn't patch PHP, so we wrote some checks to strip useless markup and keep strings under the limit.

[Peyton McGinnis]

Good gracious. I've had to deal with similar bugs in APIs that have string limits as well, always very annoying to deal with. But never anything that strange.

[Zachary Hamm]

My favorite bug has to do with not software, not coding, but my last desktop build. Four hours later after two trips back to MicroCenter that day already, I had it "done" and booted it up, and got to install Windows. Another few minutes after getting things installed on my Windows profile, it blue screens. So, I do start it up again, and lo and behold, another blue screen. And I think even some memory dump, too.

The reason?

It overheated.

Because I forgot to buy a heatsink. And put it on a supercharged 4.2Ghz i7.

I literally nearly burned $2000 from heat in the first day I had created that monster.

[Louis]

Everytime I inserted headphones into my Huawei's headphone jack, Google Assistant was activated. I guess that there are some accessibility devices for disabled people that work through the headphone jack's connectors. Because there was some dust in my headphone jack, the connectors got either short-circuited or triggered in a wrong way, idk. After I cleaned the port and removed the dust, everything worked well again.

[Clive Da]

only posted it last week ! the windows 10 add new user popup in 20x10 pixels

[Andreas]

Well that's... small 😂 Screenshot?

[Clive Da]

just found it but can you read the 1x1 pixel font

windows 7 to windows 10 upgrade problems

Clive Da ・ Jan 14 ・ 1 min read

#window #os #upgrade

[Clive Da]

yes it wasnt funny until i found the workaround

[Clive Da]

on the bus now you can search for my post with screenshots or wait until im on the wifi in the pub

[NLxDoDge]

I had a lot of bugs in my 6 years of programming, but the one I will always remember is a Database Connector file being called PG for Postgress, bu actualy being a MySQL driver renamed to PG.

Why? No clue but I reinstalled everything but the driver was in my project folder, hens it not working correctly. Took me a week to find it.

Madness.

[Jonathan Apodaca]

Encounter strange bug: git reset --hard, retype the exact same code, and it works. :shrug: