This is very cool, but I'm having some trouble understanding the use cases.
Is this mostly just for codemode where the MCP calls instead go through a Monty function call? Is it to do some quick maths or pre/post-processing to answer queries? Or maybe to implement CaMeL?
It feels like the power of terminal agents is partly because they can access the network/filesystem, and so sandboxed containers are a natural extension?
> Monty avoids the cost, latency, complexity and general faff of using full container based sandbox for running LLM generated code.
> Instead, it let's you safely run Python code written by an LLM embedded in your agent, with startup times measured in single digit microseconds not hundreds of milliseconds.
Oh I did read the README, but still have the question: while it does save on cost, latency and complexity, the tradeoff is that the agents can't run whatever they want in a sandbox, which would make them less capable too.
This is very cool, but I'm having some trouble understanding the use cases.
Is this mostly just for codemode where the MCP calls instead go through a Monty function call? Is it to do some quick maths or pre/post-processing to answer queries? Or maybe to implement CaMeL?
It feels like the power of terminal agents is partly because they can access the network/filesystem, and so sandboxed containers are a natural extension?
This reminds me of a recent issue I had. I had just gotten a new laptop from IT. While picking it up from them, I had generated myself a password, put it in my password manager on my phone, and then entered it twice to set it on the laptop. Everything worked great. But when I got back to my desk, the password didn't work! I tried a bunch of times, watched myself hit each key to eliminate typos, etc.
I went back to IT and they asked me to demonstrate. But this time it worked! I walked back to my desk, thoroughly embarrassed. But a couple hours later I had to log in again and once again could not.
After thinking about it for awhile, I realized that I was typing at IT while standing over a sitting-height desk. Sure enough, typing in that position fixed my issue. I carefully watched what I was doing this time - something about the exact layout of the keyboard and the weird angle I was typing at ensured that I was making a particular typo I typed in that position - just a single letter switched to another, every time. Sure enough, making that one substitution to my intended password got me in.
It's worth noting that sometimes (incorrect) keyboard maps can get in the way.
If it's a key that you may not often type and one that is often transposed between regions, the fact that the entered char is not shown can lead to frustration.
e.g. " and @ are in different positions in UK vs. US keyboards. So user thinks they are typing @, but " goes into the box.
One of the more annoying things I've found moving country is the unavailability of keyboards / laptops with the layout I grew up with. I find it especially annoying as the country I'm from uses a US layout which I naively assumed would be easily available everywhere (and it is available but not without a long delivery and a premium price)
Side note: helping my French housemate with his uni assignments was an experience, none of the symbols were where I expected them to be
Meh, takes you like some days to get used to another layout being visible on the keys, while your OS (and brain) actually using another layout.
I've used US keyboard layout since I started programming (my first mentor essentially forced me to switch to it, he was right about it being easier), but throughout the years been using Swedish, Norwegian, British, Spanish and French physical keyboards, never cleanly mapped to the actual layout I've used on the OS, and never been an issue.
The last part though, is a real one, trying to pair program with Spanish programmers always have at least one moment of holding Shift and sliding the finger across all numbers to see where that specific symbol actually is.
On other layouts that isn't enough. For example French keyboards are AZERTY, not QWERTY. and here in Sweden we have å, ä and ö next to the (tall) enter key, instead of the symbols US and UK have.
(Side note: those are not a and o with diacretics, they are entirely separate letters in the alphabets of the Nordic countries, with entirely different sounds.)
The product I make deals with passwords. We’ve had several bugs over the years that came down to Unicode usernames and passwords containing unexpected characters. Solving them was simple, we just had to be sure to get the encoding and character sets right, but as an American it was eye opening to find so many people with the Euro symbol in password strings.
Related, a friend of mine uses a long list of heavy metal band names as one of his unit tests for strings. Says it catches a lot of weird encoding bugs.
I’ve done this before as well. It truly baffled me because of how much in undermined me sense of being totally aware of my body. I truly believed I was hitting the right keys (I know how to type after all) and I never noticed any issue when writing normally, but only when typing my password. But of course I couldn’t see my password as I typed, while in other cases I would subconsciously correct any resulting typos because I could see them. I had no reason to classify typos due to standing as any different than the regular errors I might make while typing.
Almost felt like a bug in error correction loop in my brain, or maybe more like an unconsidered edge case.
I somes subconsciously correct typos even when not looking at them. It drives me crazy when UI design breaks this, like fixed-length security code / PIN entry UIs that automatically submit when you enter the last character of the code.
I also tend to memorize long (8+ digit) PINs based on the physical layout of the keys, so if I need to enter a PIN set up on a phone-style keypad on a normal keyboard or numeric keypad, or vice versa, I need to visualize entering the PIN on the original input device to remember it.
I don't think the OP ever returned with a conclusive answer, but I'm somewhat convinced by the commenters that it was either a low-frequency engine sound rattling the neighbor's windows or something to do with the car's rear-cross sensor.
This has been happening to me and I had no idea it was this. Every time I sit down at my chair the monitor goes black for a second. Never would I have guessed this.
Obligatory mention of David J. Agans's "Debugging: The 9 Indispensable Rules for Finding Even the Most Elusive Software and Hardware Problems" where you can find dozens of such stories, including why their computer crashes when you wear a certain green T-shirt.
Ok I swear I had a printer that would do some kind of internal cleaning noise thing every time I plugged something else in to a 120v outlet anywhere in the same apartment. I never really tried to figure it out.
It's an urban legend that's floated around in various forms: in some it's an ice cream parlor rather than a store and they pack the vanilla faster, in some it's only the vanilla that gets hand-packed so it takes longer, it's pistachio that takes longer and triggers the problem, or butter pecan.
Snopes covered this one and they cite to an urban legends book from 1989: "Curses, Broiled Again" by Jan Harold Brunvand. Brunvand prints the "vanilla takes longer" version and reports also having a "pistachio takes longer" version printed in a magazine in 1978, which itself referred to another magazine as its source. In the book's version it's a Texas car dealer who's looking into the problem. The same author's later book from 1999 covers the story again and includes versions dating back further, plus a 1992 version which is this one where it's Pontiac and the problem is the vanilla being in a separate case at the front of the store.
In the Pontiac version you can go and pick at various implausible details stacked up together, that Pontiac's president cares enough to send an engineer out, that the engineer is there when the car won't start on the first night but still just comes back many more times rather than looking at the car or presumably noticing that it starts a couple minutes later, that this guy is buying a new container of ice cream every night and never stocking up, that he never takes any other trip where it's a short stop... You can go on each of those and say they do happen: like presidents and CEOs do sometimes go digging deep on random problems customers put in front of them. But if you look at the whole thing I think you need to recognize it as a piece of storytelling, not fact.
Maybe there's some kernel of a true story in there, but if so it's probably a pretty small kernel. Anyway it doesn't matter much: it's just a fun story that teaches a little lesson so people like to share it around.
In 2021, between jobs, I did some light data analysis and visualisation on the topics that Matt Levine writes about in Money Stuff and enjoyed seeing some trends, so I blogged about it!
I also emailed it to Matt, who included it in his "Things Happen" section!
I’ve seen some weird technical glitches in my career. One that i will always remember is that a customer was very happy with his new big computer, but could not work for multiple hours on it, because his office would get colder and colder when he kept using it. After some mailing and talking over the phone i suggested a visit to his office where i quickly found the cause of the problem: The big computer fan was aimed directly at the thermostat knob of the radiator, so it assumed the entire office was well heated and closed.
It's not quite in the same league as any of this, but when I was a child, we sent our Commodore 64 in for repairs several times because it started "writing" by itself. Gibberish would slowly appear as if someone was randomly hitting the keyboard.
Each time it took several days before they repair centre got to it, and they then contacted us to tell us there was nothing wrong with the computer at all.
After we picked it up, eventually, when it started happening again for the third or fourth time, we realised the problem:
The "large" (a whopping 26") CRT TV we'd recently started placing it under when not in use caused it... A few days away from the TV to dishcharge it, and it was fine - hence why the repair technician didn't find anything.
I know this story is a bit unrelated, but your story brought it to mind (probably the keywords child, repaired several times, etc): When I was a kid my father had a Nokia 3310. Any early Nokia owner would remember how you could configure your own 'welcome message', which would flash up on the phone's monochrome screen for a second while it boots. My Dad was an engineer, and would get totally hung up on the minutiae of anything technical. Picture him spending a totally inappropriate amount of time adjusting the TV's audio settings, for instance. One day I thought I'd play a prank on him, and changed the welcome message on his phone to say 'SIM CARD ERROR'. It sent him completely bananas, and triggered this huge quest to remedy the problem. He ended up taking it back to the local Vodafone store where he bought it multiple times, and they couldn't figure out what was wrong either. Maybe they got the joke and thought it was too funny, who knows. The store ended up escalating the case to Nokia and authorised an RMA. At which point I figured the joke had gone too far, and I reset the welcome message to 'Dad's Phone'. For some reason, he was happy just to leave it at that. I never confessed what I'd done! It caused way too much fuss, and I'd have been absolutely in for it if anyone figured it out!
I once had a desktop computer that had great uptime, but started to consistently crash when I got up and left the room to get a drink of water.
Turns out it was old building with loose floorboards. The vibrational force of standing up was enough to short out a failing power supply. As long as I sat my desk, it was fine.
But I had a co-worker who had a worse problem with getting up to get a drink of water. Once while she was kitchen, an eight foot steel lighting ballast came loose from the ceiling and felt right onto her chair.That what-if memory still haunts me.
(not disclaiming that it wasn't, but that "chair piston causes EM surge" had me driven crazy for the longest time til I was able to pinpoint the cause)
That reminds me of one. I had a PC that would fail to boot the first time every day. Second and subsequent times were fine, until the next day.
When it stopped happening in the spring, and started again in the fall, it became obvious -- my apartment was too cold. The heat from the first failed boot sufficiently heated up the system to boot the second time.
“WiFi doesn’t work in the summer” is one of the first anecdotes I learned about WiFi when it was still brand new. You set up WiFi between two buildings in the winter, spring comes and the water in the leaves blocks the signal.
I had one of these myself. WiFi wouldn’t work when my wife was using her laptop in bed. As soon as she gave it to me, it started working again. She thought it was the magic touch of the engineer, but it turned out that when she was in bed, she pulled her knees up and set the computer on her lap, while I would lay down completely and let the computer rest on my chest. Her knees blocked the WiFi signal enough to be quite noticeable.
I wonder if the feeling is excitement or horror when you encounter one of these weird problems that seems like it has to be the user.
Not computer related really, but I'm reminded of when my Mom was helping set up macs in the lab at my middle school. I, a 4th grader, tagged along and hung out in the other lab across the hall. I got very incredulous looks when i claimed that there was a lizard in there. It was the Midwest over summer break! I was obviously a kid seeing things. There's no lizards here.
Then I produced it, caught under a bin. It was a brown anole that had come back in a plant sent from Florida. I wasn't crazy that day.
Since you mentioned your mom, mine is not as tech savvy. At one point she needed a computer to type something and print it, a simple use case so I came up with this idea of setting up a computer that would give me no tech support trouble, since I was living in another state. I installed CentOS, libre office and made sure the printer was supported.
I told my mom to keep the system up to date and set up an ssh connection for remote access just in case.
A few months go by and one day I receive a phone call that she cannot find the system updater shortcut anymore. I started to think how I could get Gnome to load over ssh, I was sure she moved the icon accidentally or something but decided to google it just in case.
Lo and behold and there is a bug report that due to some bug in package management dependency resolution the graphic software updater GUI could remove… itself… if the user performed a routine system update. It seemed to even affect RHEL at the time if I’m not mistaken.
A yum install command away over ssh and it was solved but that was the day I realized that no matter how stable a distro is famed to be or how much support it has from a company, there was still lots of work to be done until Linux could be seen as friendly enough for the end user.
Most of these are good, but "can't log in while standing up" is just too implausible. I can't possibly be led to believe that every single one of a whole group of technically-literate touch typers failed to notice that keys were swapped.
There also is the (somewhat) famous caps lock gamble in XCOM 2 [1].
Quote: "Hitting the key, through a rube-goldberg-esque series of events, forces all outstanding load requests to be filled immediately in a single frame. This causes a massive hitch, and potentially could crash the game. If you don't care about those adverse effects the synchronous load is faster."
Is this mostly just for codemode where the MCP calls instead go through a Monty function call? Is it to do some quick maths or pre/post-processing to answer queries? Or maybe to implement CaMeL?
It feels like the power of terminal agents is partly because they can access the network/filesystem, and so sandboxed containers are a natural extension?