Peter Ford

In 2020 I've gone from never having recorded myself to being annoyed by continuity errors in my pretend-cello-filled piano quartet youtu.be/9c_2-fMP2yw But viola+piano is a bit limited, so I'd love to hear from anyone who'd like to join me for some video-assisted music making.

@MCCCANM "enabled" or "disabled". But without quite a lot of Googling (or a wideband spectrum analyzer) I wouldn't be confident to say which frequencies someone else's device was using. So I would expect any modern plane needs to cope with a +23 dBm device held against the cabin wall?

@MCCCANM One thing I've never understood here: if consumer electronics which transmits in cellular bands is dangerous in the cabin, then consumer electronics shouldn't be allowed in the cabin. As an RF engineer, I'm 90% sure I know which bands my phone/watch/tablet/etc use when /1

An honest, fair question. Truthfully, in the vast majority of situations it probably wouldn’t have an effect on the airplane to have your phone on, but it *could* in some scenarios. The WiFi is on a different part of the spectrum than cell signals, and doesn’t interfere w/ avionics (though the rollout of Starlink on regional jets was paused briefly when it was discovered there was a bleed over to a wire bundle that wasn’t shielded, causing problems. They turned off systems already installed until they could figure out a fix, which they did). The cell signals are much closer to some of the frequencies our avionics use, though, and there is sometimes a possibility of bleed over. For example, the rollout of 5G in 2019 to 2022 caused several instances of radio altimeters to malfunction, in some cases dangerously. Radio altimeters are a key part of our landing procedures; as such, the FAA & wireless companies each developed mitigations until frequency filters could be installed on the avionics (which was completed pretty quickly by most airlines). A single cellphone searching for signal can be heard in our headsets, if it’s close. I’ve left mine on by accident before & hearing it in the headset is what reminded me it was on. The effect can be amplified if a bunch of phones are sending signals. (If you owned those old computer speakers from the 90’s & early 2000’s & sat a cell next to them, you have probably heard what this sounds like) Realistically, the chances of a cell phone interfering with the airplane are next to zero, but I hesitate to say it’s absolutely zero. The EU got rid of this requirement in 2022, and the FAA has studied it, but they are a very cautious agency. You have to prove to them that something is safe on your own dime, and even then they’ll be skeptical…they move glacially slow when it comes to overturning old rules. There are other concerns, as well: in one hearing, the FAA solicited public comment on whether the ban should be lifted. More than half of the comments received asked for the ban to stay in place, expressing concern that people would have conversations that annoy their neighbors & could lead to increased “air rage”. Honestly, I can see this argument. Imagine being trapped next to someone on their phone… Beyond that, your phone will just wear out its battery constantly searching for a signal, which it won’t get in cruise. It also has the potential to overload cell towers at lower altitudes, as a plane full of phones hops quickly between towers, connecting to multiple at a time…the networks were not designed for that. In any event, it’s an FAA requirement & the airlines are required to enforce it. If they don’t, the FAA can impose very stiff penalties. It’s also an early indicator that you are not going to follow the commands of a crewmember…there is no upside to arguing with a Flight Attendant about this. If you’re already argumentative, we’re not going to take you in the air. Hope that helps!

@SoVeryBritish The Portuguese think this is the expected accompaniment to a steak. A fried egg is also obligatory.

Only a British person would look at a portion of rice and think “I know what would go with that… a portion of chips”

@levelsio Huh, I (in Lisbon, Portuguese and UK SIM cards active) didn't get anything, although I have had SMS warnings from ANEPC about Chuva forte in the past. Maybe they sent it when the cellular networks were only intermittently functioning...

We did get this SMS from the ANEPC "ANEPC is the national authority on emergency and civil protection of Portugal"

@crypt2d2ot @michaelmiraflor I'm very happy that the standardised port exists. I'm very unhappy that the version numbering for the transport is a complete dumpster fire, and mildly unhappy that the USB-IF couldn't be bothered to come up with connector labelling until it was far too late.

@michaelmiraflor @peter_x_ford if your rebuttal to a standard cable is some forced contrived scenario that is supposed to defeat the entire overarching purpose of the cable existing, then yeah, i will simply tell you to get over yourself.

it's extremely obvious who is tech illiterate and doesn't understand how USB-C works. we've "made something better than C" 4 separate times now. it's called USB-C. the entire point of the standard is that you can make new connectors that use the same standard port.

@crypt2d2ot @michaelmiraflor I have a 65W (?) charging cable, an alt-DisplayPort (?) cable, and many cables without these features. All of them are visually identical C-C cables. When I travel these all get chucked together. Yes this is solvable, but it's annoying there's absolutely no marking on the cables.

@michaelmiraflor this is a nonexistent problem.

@hikari_no_yume @Love2Code @blelbach I don't understand this argument. Nothing forces you to avoid dividing by zero, but... we manage not to? If I'm given a value/reference, I can use it unconditionally. If I'm given an optional or pointer, then my code immediately has two logical paths: do I have a value or not?

@Love2Code @blelbach in my limited experience, option types in c++ are less useful than they are in something like rust because nothing forces you to actually check that they have a valid value in them, and it's more ergonomic to simply not do so. you're just reinventing null pointers at that point

Adding a borrow checker or similar for C++ or switching to Rust protects code written in the future. But the issue with C++ isn't future code, it's the billions of lines of existing code. Solutions like sanitizers that can be applied to existing code are critical.

@fasterthanlime I think this is the key point that I had been missing in these discussions before. In most tasks in our codebase, I spend more time thinking about what to implement than I do implementing (with the help of some codegen), and I hadn't realised that might not be that common.

@imobulus @UCEconomist @the_coproduct @yacineMTB github.com/serge-sans-pai…

@peter_x_ford @UCEconomist @the_coproduct @yacineMTB A map which is (algorithmically) constructed during compile-time but key lookups to which are not known during compilation? Highly doubt that. The best I can do is such map constructed before the start of execution of main() but during runtime

holy fucking shit

@imobulus @UCEconomist @the_coproduct @yacineMTB We use lots of constexpr maps and sets. It just requires going to github to download a nice header someone wrote.

@UCEconomist @the_coproduct @yacineMTB Also comptime maps are impossible while being perfectly sensible. If you can build and run code, why not do it as part of building the other code?

@cratilo2 @theo_nash ... although I'm quite lost about the purpose of the partial order on the Groups?

The regulations for Mods in 1920.

@brainiac256 @caesararum pdftk?

@caesararum Isn't there a pdf library floating around that's like that? Or Ghostscript maybe is what I'm thinking of?

there are a couple libraries that are _so good_ at what they do that they prevent other libraries from existing almost all of them have horrifying ergonomics ffmpeg, imagemagick, openssl what are others?

@cybergibbons We use these a lot at conferences where there's usually people from 3 or 4 different countries sitting at a table and plugging in their equipment. Is it just cheap ones that are a concern, or are there unlikely to be any good ones out there?

UK peeps: please don't buy or use these multiway sockets that can take UK/EU/US plugs all in one. It's really hard to design the contacts to work with the wide array of pins that can be used. The end result is poor contact. They also often have very thin wire.

@RudRecciah @uxorious_cosine @keysmashbandit @samadamsthedog At my previous job trying to use a float would have got me laughed out of the code review. But perhaps targeting an ARM M0 isn't so different from driving a car which has no steering wheel.

asked a kid who just finished programming community college how he would go about finding the largest float in an array and he could not answer. he did not know what i was talking about. he got out his phone to ask chatGPT. he asked me to repeat the question into speech to text

@gorunmane @lefticus We compile our code with all three in order to support the platforms we want to; what are the problems you have with them? The only thing that trips me up is frequently forgetting how far Apple clang is behind the others.

@lefticus I’m really curious as to how this is even possible. Do they use clang for everything? Because imo the only way to do it would be using gcc , clang , AND msvc - which is absolute insanity when you compare the alternatives and benefits offered

PSA: It's 2024 It's been 13 years since C++11 was released. It was 13 years from C++98 to C++11. If you still feel like you cannot "trust these modern features of C++" that's 100% on you.

@robertgraham I'm surprised by how much difference the comment makes; or is that common? That without comments it can (partially) explain the effect of a block of code, but doesn't manage to say anything about the purpose of it?

For example, consider this code where I implement the classic Unix `wc` word-count program using a state-machine. github.com/robertdavidgra… This code is meant to be unreadable, as a teaching aid to show exactly what it means "using state-machines". The student is meant to have to think deeply what it means to have state transitions. On this link, I pass the code to ChatGPT, minus the first comment. chat.openai.com/share/08eca42f… In this link, I pass the code along with that comment: chat.openai.com/share/1dfb24b1… The point is that ChatGPT can understand and explain this code is that is expressly designed not to be understandable without me explaining it. Programmers spend a huge chunk of their time trying to understand things. That's their job, to understand a thing, then express it in code. ChatGPT does most of that work, does the "understanding" for you, and does part of the translation of that concept to code. Thus, software engineering has completely changed. There was a time when humans did it without AI, and now all future programmers will do it with AI.

I disagree. It's completely changed the field of software engineering. It's now pointless teaching students how to write code without including AI in the process. There is no going back to the days when we write software code without AI. It's difficult to appreciate because people can't precisely describe what the AI is going to help them code. AI doesn't write the code. AI doesn't debug the code. What it does do is a bunch of other tasks. For example, it can be used to explain difficult to understand code. Or, it can teach you how to use an unfamiliar API. These things take a lot of a programmers time. I hate Python and am thus struggle to modify Python tools. With AI, it's now trivial for me to work with Python. While AI doesn't write the code for you, it can get you started. It can create an example that almost works, doing most of the grunt work to get you started, after which you the programmer finish it up.

@RaghavMalik15 @joshuagrochow @linguanumerate Does that vary between languages? For C++ it has good debugger integration, good code lookup/autocomplete via clangd; I have full VS set up for the same project and I've opened it ~3x for some profiling that was awkward to do via my normal profiling tool.

@joshuagrochow @linguanumerate I also dunno that I’d call vs code an IDE at the same level as pycharm or eclipse, I feel like it’s more an editor with an attitude.

@linguanumerate What do you like to play, and where is "become" starting from?

@jen_canary @JayHulmePoet What I find odd is that it seems restaurants are actually taking the payment 1st and then taking more later. I see: 1) card taken away 2) bank app tells me $30 has been taken 3) write $6 on receipt 4) sometime later payment changes to $36. I assume they only pay one fee, though?

@KirkegaardEmil Sadly I rarely do any interesting maths anymore. But I do need to collaborate on technical documents with colleagues; I can't check Word documents into git, and markdown is awkward. What's the sensible alternative to LaTeX?

"We show that LaTeX users were slower than Word users, wrote less text in the same amount of time, and produced more typesetting, orthographical, grammatical, and formatting errors. On most measures, expert LaTeX users performed even worse than novice Word users."