[jbreckmckye]

About a year ago I was looking at Crash Bandicoot timer systems and I found that Crash 3 has a constantly incrementing int32. It only resets if you die.

Left for 2.26 years, it will overflow.

When it does finally overflow, we get "minus" time and the game breaks in funny ways. I did a video about it: https://youtu.be/f7ZzoyVLu58

[jsheard]

There's a weapon in Final Fantasy 9 which can only be obtained by reaching a lategame area in less than 12 hours of play time, or 10 hours on the PAL version due to an oversight. Alternatively you can just leave the game running for two years until the timer wraps around. Slow and steady wins the race.

https://finalfantasy.fandom.com/wiki/Excalibur_II_(Final_Fan...

[Gravityloss]

Am reminded by this quote from Ferdinand Porsche:

"The perfect racing car crosses the finish line first and subsequently falls into its component parts."

Games fit this philosophy, compared to many other pieces of software that are expected to be long-lived and receiving a lot of maintenance and changes and evolve.

[WJW]

The Porsche quote reflects a wider design philosophy that says "Ideally, all components of a system lasts as long as the design life of the entire system and there should be no component that lives significantly longer. If there is such a component, it has been overengineered and thus the system will be more expensive to the end consumer than it needs to be.". It kinda skips over maintenance, but overall most people find it unobjectionable when stated like this.

But plenty of people will find complaints when they try to drive their car beyond its design specs and more or less everything starts failing at once.

[creaturemachine]

Porsche was talking about racing, where the primary focus is reaching the finish line faster an anyone else, and over-engineering can easily get in the way of that goal. Back in the real world, no race team would agree that their cars should disintegrate after one race.

[AzN1337c0d3r]

> Back in the real world, no race team would agree that their cars should disintegrate after one race.

Wasn't F1 teams basically doing this by replacing their engines and transmissions until the rules introduced penalties for component swaps in 2014?

[jperras]

If you go back further than that, teams used to destroy entire engines for a single qualifying.

The BMW turbocharged M12/M13 that was used in the mid-eighties put out about 1,400 horsepower at 60 PSI of boost pressure, but it may have been even more than that because there was no dyno at the time capable of testing it.

They would literally weld the wastegate shut for qualifying, and it would last for about 2-3 laps: outlap, possibly warmup lap, qualifying time lap, inlap.

After which the engine was basically unusable, and so they'd put in a new one for the race.

[creaturemachine]

Yup, cigarette money enabled all kinds of shenanigans. Engine swaps for qualification, new engines every race, spare third cars, it goes on. 2004 was the first year that specified engines must last the entire race weekend and introduced penalties for swaps.

[kllrnohj]

Even today F1 teams are allowed 4 engine replacements before taking a grid place penalty, and those penalties still show up regularly enough. So nobody is making "reliable" F1 engines.

You can see this really on display with the AMG ONE. It's a "production" car using an F1 engine that requires a rebuild every 31,000 miles.

[pfdietz]

Don't highly optimized drag racers do this? I mean, a clutch that in normal operation gets heated until it glows can't be very durable.

[ortusdux]

Anyone can build a bridge, but it takes an engineer to barely build a bridge.

[mikepurvis]

Alan Weisman's lovely book World Without Us speculates a bit about this, basically saying that more recently built structures would be the first to collapse because they've all be engineered so close to the line. Meanwhile stuff that already been standing for 100+ years like the Brooklyn Bridge will probably still be there in another 100 years even without any maintenance just on account of how overbuilt it all had to be in an era before finite element analysis.

[ortusdux]

There was an aluminum extrusion company that falsified test records for years. They got away with it because what's a few % when your customer's safety factor is 2. Once they got into weight sensitive aerospace applications, where sometimes the factor is 1.2, rockets starting blowing up on the launch pad.

https://www.justice.gov/archives/opa/pr/aluminum-extrusion-m...

[The_Fox]

This is a great quote for the topic, but the quote is normally about a bridge that barely stands.

I'm chuckling at the thought of barely building something. (All in good fun, thank you.)

[woliveirajr]

In my county, a company asked the Mayor if it was possible to improve some bridge because they need to carry 40t and the bridge had a sign telling it would only allow up to 32t. Their proposal was to do the construction and get tax rebates.

After two weeks, the Infrastructure department changed the sign allowing up to 45t.

[signalToNose]

Consumer protection laws prevents businesses following this to it’s extreme. For many businesses the ideal would be to just sell stuff that immediately breaks down as soon as it’s sold. It has the fulfilled its purpose from their point of view

[delichon]

I run sous vide cookers 24*7, and they uniformly break within 90 days or less. But they don't like to admit their smaller duty cycle, so they don't, and keep sending me warranty replacements instead. I keep buying different brands looking for one with a longer life. I'll bet most people do that when their gadgets die, and purposely making products that die as soon as sold isn't often a successful business model.

[rlander]

That’s not a small cycle count for a normal household. 90 × 24 = 2,160 total hours.

I sous vide now and then, about twice a week for 6 hours each, so around 12 hours a week. That works out to roughly 15 years of usable machine time for the average person.

Not bad at all.

[cestith]

A friend of mine gets new headphones/headsets every six to eighteen months, and hasn’t bought a pair entirely out of pocket in years. For him it’s all down to buying the Microcenter protection plan every time they’re replaced. They fail, he takes them back, he gets store credit for the purchase price, and he buys a new set and a new plan. He doesn’t even care about the manufacturer’s warranty anymore.

Personally, most of my headphones I look for metal mechanical connections instead of plastic and I buy refurbished when I can. I think I pay about as much as he does or less, but we haven’t really hashed out the numbers together. I’m typing this while wearing a HyperX gaming headset I bought refurbished that’s old enough that I’ve replaced the earpads while everything else continues to work.

Computers and computer parts often have, in my experience, a better reliability record competently refurbished than when they first leave the factory too. I wonder if sous vide cookers would.

[hnuser123456]

Are there not industrial ones meant to last longer? Maybe you can buy a used but good condition one of those.

[account42]

Well from an evil business perspective their options are either

- the product doesn't break and you don't buy a replacement from them because you still have a working product

- the product breaks and there is a greater than 0% chance that you will buy a replacement product from them

Of course in practice it's more complicated but I wouldn't be so quick to declare that the math doesn't work out.

[FuriouslyAdrift]

Try a Breville PolyScience... https://www.breville.com/en-us/product/csv750

Or if you want something even beefier: https://sammic.com/en/smartvide-xl

[muzani]

What do you sous vide 24*7? It sounds like it would be party grounds for bacteria. Also curious if the bags and other components break as well.

[doubled112]

When the design spec seems to be a 3 year long lease I can see why people get bothered.

[aleks224]

There's a quote in the bible that says something similar:

"Verily, verily, I say unto you, Except a corn of wheat fall into the ground and die, it abideth alone: but if it die, it bringeth forth much fruit.”

(John 12:24)

[lelandfe]

So the invisible 12h timer runs during cutscenes. During Excalibur 2 runs, I used to open and close the PS1 disc tray to skip (normally unskippable) cutscenes. Never knew why that worked.

(I also never managed to get it)

[jonhohle]

I’m going to wager that the cutscenes are all XA audio/video DMA’d from the disc. Opening the disc kills the DMA and the error recovery is just to end the cutscene and continue. The program is in RAM, so a little interruption on reading doesn’t hurt unless you need to time it to avoid an error reading the file for the next section of gameplay.

[ad133]

This is a significantly better handling than the previous game (final fantasy viii). My disk 1 (it had four disks) got scratched over time (I was a child after all), and the failure mode was just to crash - thus the game was unplayable. The game had a lot of cutscenes.

[Insanity]

That’s a solid guess. And if that’s the case, that’s actually pretty good error handling!

[Jare]

I recall that handling disc eject was an explicit part of the Tech Requirements Doc (things the console manufacturer requires you to comply with). They'd typically check while playing, while loading and while streaming.

[p1necone]

> Never knew why that worked.

I'm guessing the game probably streams FMV cutscenes of the disc as they play, and the fallback behaviour if it can't find them is to skip rather than crash.

[jbreckmckye]

Oh yeah. The sword you pick up in Memoria. The problem there is that the PAL version runs slower; the way PSX games "translated" between the two video systems was just to have longer VSync pauses for PAL. So the game is actually slower, not interpolated

[reactordev]

Longer vsync pauses but larger frame time deltas so it’s basically the same speed of play. The only thing that was even noticeable was the UI lag.

[fredoralive]

Erm. No, like lots of games during the era quite a lot of stuff is tied to the frame rate, so the 50Hz region game just runs slower than the 60Hz one as next to nobody bothers to adjust for it. The clock for the hidden weapon does run at the same rate for both unfortunately, hence it being harder to get in 50Hz regions.

[reactordev]

Incorrect. I’m looking at the source code. It’s not perfect but it’s not just “slowed down to 50hz” like people claim.

[jbreckmckye]

When you say looking at the source code, what do you mean here?

AFAIK the source for FF9 PSX (and all the PSX ff games) has been lost as Square just used short term archives

Also, FF9 does not run at a constant framerate. Like all the PSX FF games it runs at various rates, sometimes multiple at a time (example: model animations are 15fps vs 30 for the UI)

In terms of timers, the bios does grant you access to root timers, but these are largely modulated by a hardware oscillator

(Incidentally, the hardware timing component is the reason a chipped PAL console cannot produce good NTSC video. Only a Yaroze can support full multiregion play)

[mungoman2]

Wouldn't a slower tick make it easier as you get more wall time to do the same challenge.

[fredoralive]

No? Wall time (that the challenge runs on) is unchanged, game time (Vsync) is running at 83% of full speed (50Hz vs 60Hz), so if something tied to frame rate (animation, walking speed etc.) takes 1 second to do on NTSC, it'll take 1.2 seconds to do on PAL etc.

[Forgeties79]

Lord have mercy fandom really has become unbearable with the ads and pop ups.

[coldpie]

Install an ad blocker.

[Forgeties79]

I opened this on an iPhone which has fewer adblock options. Desktop is better locked down.

Regardless I can still complain about how intrusive the ads are.

[coldpie]

There are many ad block options on iPhone. I currently use Wipr 2, but in the past I've used both 1Blocker and AdBlock Pro with success.

[JustExAWS]

I just opened this my iPhone with 1Blocker installed. I saw no ads. It’s been around since iOS 8

[Forgeties79]

Never heard of it, appreciate the recc!

Edit: ah only works on safari

[account42]

Don't accept devices that limit your ad blocker options.

[Forgeties79]

Does this discussion strike you as one where I’m deliberating whether or not to chuck my smartphone and buy into a new ecosystem to avoid ads on fandom?

These types of comments are always very unhelpful.

[elcritch]

We should rally together to force game companies to use 32 bit timers rather than 64bit ones so we can keep finding these fun little glitches. The time to protect overflows is now! ;)

[debo_]

So that's why it's called Excalibur 2!

[stevage]

You really managed to make the whole video without making a single "crash" pun? (Those freezes come close enough that you could call them crashes...)

[xhrpost]

Is it common to default to a signed integer for tracking a timer? I realize being unsigned it would still overflow but at least you'd get twice the time, no?

[jbreckmckye]

Some C programmers take the view that unsigneds have too many disadvantages: undefined behaviour for overflows, and weird type promotion rules. So, they try and avoid uints.

[tekne]

Umm, signed integers are UB on overflow; unsigned is always fine.

[jbreckmckye]

Sorry, you are correct. I don't think unsigned overflow behaviour was defined until C99 though.

Anyway, in answer to the question, I would guess the reason was because of signed / unsigned type promotion.

[aidenn0]

If you get to right before you need to be (taking as long as you want), then wait until overflow, then you still have 12h to do the last tiny part if it's unsigned.

[jonhohle]

I think many games were that way. SotN definitely has a global timer. On a native 32-bit system it makes sense, especially when the life of a game was a few months to a few years on the retail shelf. No player is going to leave their system running for 2.27 years so what’s the point of even tesing it?

Who knew at the time they were creating games that would be disassembled, deconstructed, reverse engineered. Do any of us think about that regarding any program we write?

[Gamemaster1379]

Can be more than timers too. There's a funny one in Paper Mario where a block technically can be hit so many times it'll reset and award items again. Hit enough times it'll eventually crash. Of course it'd take around 30 years for the first rollover and 400 or so for the crash. https://n64squid.com/paper-mario-reward-block-glitch/

[rybosome]

It’s a totally reasonable choice in that context.

I wonder if any sense this is criticism (or actual criticism) is based on implementers of SaaS who have it so deeply ingrained that “haha what if the users of this software did this really extreme thing” is more like “oh shit what if the users of this software did this really extreme thing”.

When I worked on Google cloud storage, I once shipped a feature that briefly broke single-shot uploads of more than 2gb. I didn’t consider this use case because it was so absurd - anything larger than 2mb is recommended to go through a resumable/retryable flow, not a one-shot that either sends it all correctly the first time or fails. Client libraries enforced this, but not the APIs! It was an easy fix with that knowledge, but the lesson remained to me that whatever extreme behaviors you allow in your API will be found, so you have to be very paranoid about what you allow if you don’t want to support it indefinitely (which we tried to do, it was hard).

Anyway in this case that level of paranoia would make no sense. The programmers of this age made amazing, highly coreographed programs that ran exactly as intended on the right hardware and timing.

[technion]

Let's say youre pedantic with code. Ive been trying to be lately - clippy has an ovefflow lint for rust i try to use.

Error: game running for two years, rebooting so you cant cheese a timer.

Does this make the bug any better handled? Bugs like this annoy me because they arent easily answered.

[account42]

There are always limits to what a program can do. The only fix is to choose large enough integers (and appropriate units) so that you can represent long enough times / large enough sizes / etc. that anyone could reasonably encounter. What sizes make sense also include how they impact performance and for a game from the 32-bit era, a crash (controlled abort or not) after over two years is probably a better choice than slowing everything down by using a 64-bit integer.

[jraph]

Isn't this common in the computer game scene? Shouldn't you asume your game will be disassembled, deconstructed, reverse engineered?

Although for old games released before internet was widespread in the general population, it might have not been this obvious.

[sim7c00]

aslong as it doesnt lead to online cheats having such code is fine. if someone wants to reverse the game find an obscure almost untriggerable bug and then trigger it or play with it. 2.6 year game session is crazy if its not a server, and if its a server, thats still really crazy even for some open-world open-ended game... its a long time to keep a server up w/o restarts or anything (updates?).

looking at the various comments, there might be even some kind of weird appeal to leave such things in your game :D for people to find and chuckle about. it doesnt really disrupt the game normally does it?

[lstodd]

> if its a server, thats still really crazy even for some open-world open-ended game... its a long time to keep a server up w/o restarts or anything (updates?).

Pretty much doable even without resorting to VM migrations or ksplice. My last one had uptime in 1700s (days). Basically I leased it, put a debian on it and that was that until I didn't need it anymore.

[lentil_soup]

they're still made like this. Just now I made a frame counter that just increments every frame on a int64. It would eventually wrap around but doubt anyone will still be around to see it happen :|

[account42]

For some games the timer is stored is save files so it doesn't even have to be continuous play time. 2 years is still longer than anyone is expected to spend on a game.

[ThrowawayTestr]

Great video, just subscribed

[teeray]

The true Time Twister unlocked