professoretc

But the LLM won't interpret it formally; it doesn't know the semantics of any of the programming languages it can generate code for, it just "knows" that certain patterns occur more or less frequently. You can prompt it with invalid code and it will still sometimes give you a meaningful response.

But it could, perhaps, compute something like the degrees of freedom the prompt allowed. If, in choosing how to complete the prompt, it had to choose between a huge number of different responses, perhaps that's an indication that the prompt is not precise enough? (Of course, even the most precise prompt is still going to allow for a huge number of variations in response, so you'd have to scale the actual numbers down quite a bit for them to be useful.)

> This reminds me of Inform 7 (where you can use "natural language" to program text adventure games)

The problem (at least, the problem I had) with Inform 7 is that it's not really natural language; it still has a strict formal grammar defining its syntax, it's just that the grammar has lots of "synonyms"; multiple syntaxes for the same thing. But it's still following strict rules, and those rules can't cover everything. This leads to situations where you write some "plain English" and it seemingly understands it, but then you write tiny variation and it throws up a syntax error.

Yep, I'm pretty sure Borland C++ was what I used in college in my first "real" programming class. The nice thing about the older IDEs was how easy it was to get started programming: start it up and you're looking at a blank source file; often you can compile and run without even officially saving it! Made it really easy to try things out. The only "IDE" that comes close to that level of immediacy today would be something like Processing.

> Correction - if it's flowing water there is less risk than drinking from a stagnant pool of water.

That's what I've always thought, and it seems correct, but I'm not sure why that would be the case. Flowing water is more oxygenated which should make it easier for bacteria to survive, than in a still pool of water (getting baked by the sun's UV all day).

One of my colleagues uses multiple choice exams for his classes (online and in-person) but he gives them a ton of questions. Something like 90 for a 2hr test. His theory was that, sure, you could look up the answers to a few questions, but if you're doing that for every question you'll run out of time.

Some CCs actually pay better than 4-year schools. There's a lot of competition to get tenure at a university, which can drive down salaries.

I've noticed a similar phenomenon when asking it to solve problems; if you point out a "flaw" in its proposed solution, even something that isn't wrong at all, it will cheerfully agree with you and "fix" its proposed solution, often making it worse or wrong in the process.

> Even if it succeeded, it kept changing the output by doing ops which I never mentioned in the prompt (like reordering and merging duplicated values to a single key)

That's something I don't see mentioned enough; if you change the input to a LLM, that may potentially change the probabilities of all the output tokens. Most of us would be surprised if we told a junior developer to fix a bug in a specific module, and they submitted a PR which modified literally every file in the source tree, but that's entirely plausible with a LLM. Asking it to "fix" one thing may change/break completely unrelated things.

I asked Chat-GPT to prove that the set of all integers is uncountable (it isn't). What's interesting is that Chat-GPT not only spat out the classic diagonalization proof, but rephrased around integers where it doesn't work, it ended with "This may seem counterintuitive, because we know that the integers are countable, but the proof clearly shows that they are uncountable."

Not only will Chat-GPT mess up math on its own, you can ask it to mess up math and rather than refuse, it cheerfully does it.

I just want to make pretty pictures like the ones in the article.

> I have a sister of a similar age to me who doesn't have children. My 7 year-old asked me recently - and this is a direct quote - "what is she for?"

I once asked my niece, a bit after she started really communicating, if she remembered what it was like to not be able to talk. She thought for a moment and then said, "Before I was squishy so I couldn't talk, but then I got harder so I can talk now." Can't argue with that logic.

The borrow checker acts as a subtle low-pass filter, rolling off some of the high end and leading to less clarity.

You don't necessary have to have all your funds spread out enough that every account is under 250k; you just have to determine how much you are willing to risk. E.g., if you split your funds between two banks, if one of them goes under, you lose 50%. Your risk threshold determines how many splits you need; only if it's 0 do you need to keep every account under 250k.

> when directors us existing film scores as “dummy scores” to help establish mood.

Which then means that when the actual score for that film needs to be written, it pretty much has to sound like the dummy score, because all the scenes were filmed to match the dummy score.

> it is cross platform and constructs project files for the build system of your choice. if you have people using visual studio, they can continue to use it as it will generate project files for the ide, if you have people on unix using make, it generates makefiles, i think it even supports obscure things like jam.

But these days a lot of those systems support CMake natively. E.g., Visual Studio can open CMake-based projects directly, you don't have to generate a VS project file. I'm curious as to how many people still use this capability directly; i.e., how many people use CMake to generate a platform-specific build file and then actually work with that file. I think probably the most common use for CMake is to generate a build file and then immediately build it. To probably 90% of CMake users, it would make no difference if CMake ran all the build commands itself, directly.

> - Iosevka Term

Iosevka comes with about a million different variations (and you can build custom variations by building the font from source). From the screenshot, it looks like the author was using the light weight; I almost always use medium. Similarly, if you don't like serifs, you can turn them off; I use the "hooky" variant which adds loopy tails to some letters.

Because GPT doesn't "know" what you're asking it to do. If you want it to be able to call out to a chess engine when you ask it "Solve this chess problem" then we're back in the old world of grammars and trying to parse sentence structure.

My cat has no problem going inside the robot, but he gets scared when it rotates. The result was that he clawed all the buttons (and the faceplate) off the control panel.

> > I don't care if someone can answer FizzBuzz .... All that can be learned.

> Of course. But... sometimes it isn't.

Right, it goes both ways. If FizzBuzz is so easy to learn, it's reasonable to ask, why are you applying for this position without having already learned it?

I read the SO blog post and thought it was a pretty cool coincidence that Adams was talking about SplitMix64 when I had just recently given a talk about it. Then I scrolled down and saw the video of my talk.

There's also the `artificial` attribute which instructs the debugger to skip through marked functions.

Yep, the "nice" thing about AT&T is that, from a parsing perspective, you don't really need to know much about the target architecture to be able to determine what's an instruction vs. a register vs. a label, etc.

> The problem is figuring out which ones.

To be more precise: figuring out which ones are useless may cost more time/money/effort than you would save by removing them.

Someone will need to start an EU-based business where US developers can submit their rejections, and have the EU side get a proper explanation.

In ZSH `=executable` expands to the full path of `executable`.

Coming out of the pandemic, in math we're seeing a lot of #3. It's rather hard to blame the instructors when we have college calculus students who struggle to work with basic fractions.

That's true for any C++ compiler, really. Although C++ tries to retain some element of compatibility with C, there have always been differences (you can name a variable `class` in C but not in C++).

> Maybe it's possible to hire c++ dev and make them work on a rust project?

Maybe, but not after proclaiming to the internet "your job is obsolete".

> could stop their habit of splitting lanes of traffic into multiple lanes

Even worse are the roundabouts that are wide enough for multiple lanes, but with no lane markers. Just kind of a circular flood of cars, with people on the inside suddenly realizing they need to get off, etc.