jbstack

"Please write me some documentation for this code. Don't just give me a list of bullet points. Make sure you include some context. Don't include any icons. Make sure the text flows well and that there's actual reasoning. Don't include comments about changes made during development that are irrelevant to the final work. Try to keep it concise while respecting these rules."

I think many of the criticisms of LLMs come from shallow use of it. People just say "write some documentation" and then aren't happy with the result. But in many cases, you can fix the things you don't like with more precise prompting. You can also iterate a few rounds to improve the output instead of just accepting the first answer. I'm not saying LLMs are flawless. Just that there's a middle ground between "the documentation it produced was terrible" and "the documentation it produced was exactly how I would have written it".

Where human performances are staged, the human still needs be capable of learning the choreography. There's a difference between a robot learning the routine vs having it hardwired.

Humans have a similar weight and can also fall, so would you be satisfied with a fall rate the same or better than an average human?

Why are they mutually exclusive? Can't you hang out with and chat with other programmers ... about food, hiking, travelling, culture, or relationships? In the same way that lawyers don't have to talk about law, and doctors don't have to talk about medicine when they hang out. When I think about my friendships, it's rare that we ever spend any significant amount of time talking about our work.

One way to solve this it to use a configuration management tool (Puppet / Chef / Salt / Ansible etc.). Alternatively, run NixOS. You apply the setting once and then it's applied to all your machines from that point onwards.

I don't understand your point. We're talking about ways to reduce dependence on big corp products. Some people object on the basis that it's not always feasible. I've responded that it doesn't have to be all or nothing. You've identified some products that you think you can't move away from. I've identified some that I think you probably can, and acknowledged that you might still by stuck with those others. I don't see how your latest comment fits in.

In the context of the present debate, Pixel phone is inherently bad because it's a Google product. You're putting money in Google's bank account when you buy one, and you're running your phone the way Google wants you to. The point of the debate is whether it's feasible to move away from such things. In the case of a Pixel phone, it is possible (to some extent, anyway).

I don't see any contradiction here with what I said. If you feel that using Google for email is unavoidable, that can be the part that you keep using. You can still easily ditch a lot of other things. E.g. Pixel phone, Google Docs, Google Drive, AWS. Each of those has plenty of, arguably better, alternatives.

It doesn't have to be a binary choice between "don't use it ever" and "continue using it as much as you are now". If people stopped using these services 50% of the time, it would have a huge impact.

There's nothing wrong at all with charging for your product. What I do take issue with, however, is convincing everyone that your product is FOSS, waiting until people undertake a lot of work to integrate your product into their infrastructure, and then doing a bait-and-switch.

Just be honest since the start that your product will eventually abandon its FOSS licence. Then people can make an informed decision. Or, if you haven't done that, do the right thing and continue to stand by what you originally promised.

I wasn't referring to you specifically. It's a general thing. Often the only thing I have for someone is their email address.

> If you work for a company that uses them, you ought to have them installed on your work machine. This is not unreasonable.

I covered that with "It might be fine in certain contexts (e.g. team members..." and "it breaks down when you're communicating with arbitrary members of the public". A lot of the comments in this thread seem to be assuming that the only communications that ever take place are intra-company. For many people that isn't true at all. For me the vast majority of work-related communications are to clients and third parties. I can't just say "screw them" or "they'll find a way to get hold of me if they care enough" or tell them to use slack to inform me that they've emailed me. I have to live in the real world.

Where's the boundary between "natural" and "artificial"? If we're allowing processing in the definition of "natural" (e.g. extracting a chemical from a plant using a solvent) then everything is natural: it's all ultimately derived from something that naturally occurred on Earth.

Even if they communicate visibly it doesn't always work. I don't use slack / pagerduty (not even sure what that is) and I'm not going to install or set up an account on some random proprietary service just to meet the demands of one email recipient. It might be fine in certain contexts (e.g. team members or friends/family who all use the same communication apps) but it breaks down when you're communicating with arbitrary members of the public.

Replace normal with common, and my original comment stands. In many cultures people often casually say "how is it going?" when they don't mean it literally and they are just expecting a token "good, you?" in response. Some people might view it as unnecessary and insincere, but they don't generally go around lecturing others and telling them not to do it. When you're faced with a harmless cultural behaviour like this, the polite thing to do is just accept it and move on with your life.

You're taking my comment out of context. This isn't about neurodivergents, it's about someone who thinks everyone should conform to his highly detailed rules (the article isn't only about not apologising - he has other demands on structure, content, and even plaintext vs html) when sending him emails.

(1) I don't have Jira, (2) I don't want to fill out a SLA ticket, (3) I don't use Teams, (4) I don't know your phone number and/or prefer to deal with things in writing.

Email works because: (a) it is ubiquitous, (b) you don't have to pay for some proprietary software to use it, (c) you remain in control of your data (no IM messages suddenly disappearing), (d) you have a permanent, local, copy of what was said in writing, (e) it's often the standard court-recognised form of communication, other than post, for things that matter legally (e.g. sending notices).

That's not to say that email isn't without many defects. But it's still the best we have for many work-related use cases.

For me the point is that if you feel uncomfortable over something that is widespread and considered normal social etiquette, it's on you to deal with feeling uncomfortable, and you can't really expect everyone else to change their behaviour just for you.

Are you really inspecting every app you install, including all its dependencies, and the dependencies of those dependencies, to a level of detail sufficient to identify sophisticated and obfuscated backdoors?

In the real world, nobody does this. Instead, you make a conscious choice to trust the apps that you install. Every decision of whether to install an app is a tradeoff between (a) the risk that that trust is misplaced, and (b) the benefits of the app.

Now there's a related problem in the premise: it pre-supposes that the company has an unlimited amount of valuable work to be done. If that were the case, all companies would simply expand their workforce as much as possible all the time, only constrained by money running out (which itself would be an exponential increase since "valuable" work presumably leads to more money in future). In reality, companies do not prioritise expansion above all else. In fact any time a company pays a dividend to its shareholders, or otherwise refrains from spending cash reserves on new hires, it's recognising that it cannot invest profits in an effective way into its labour force.

When framed correctly (there's effectively an unlimited labour supply for most companies, and effectively a limited demand for staff) then the question becomes "shall we hire an engineer to fix security bugs when we don't need an engineer for anything else?".

Ok, but this ultimately just comes down to a debate over the amount of the cost. The principle is the same. Even if we double or triple the cost, it's a drop in the ocean for a company like AMD.

> The calculation is "what could that engineer be doing instead and what is that worth vs fixing this issue?"

I don't agree with this, because it pre-supposes that there's a limited number of engineers available. The question isn't "shall I pull engineer X off project Y so that he can fix security bugs?", it's "shall I hire an additional engineer to fix security bugs?". The comment above mine suggests the answer to that question is "no, because it's too expensive to do that compared to just paying to clean up security breaches after they happen", which is what I was questioning in my first comment.

I didn't say I don't believe it happens. I'm saying I don't believe it's a based on a cost benefit analysis. I.e. that in a multi-billion dollar company someone consciously ran the numbers and decided "it's cheaper for us to pay to clean up the mess if there's a security breach than it is to hire someone to fix security bugs". The cost of the latter is too low for this kind of logic to make any sense.

I think it's more realistic that in any sufficiently large company the bureaucracy is so unwieldy that sensible decisions become difficult to make and implement.

I don't buy it. It makes sense for a small company where the cost of fixing it might be noticed. But AMD generates some ~$30bn in annual revenues. How much of a developer's time does it take to change the code to use HTTPS? $1000? $5000? Let's be extreme and call it $10,000. That's 0.00003% of AMD's annual revenue. It's barely even a rounding error on their accounts.

Also there's always the risk that the bolt cutter has a defect (perhaps deliberately introduced at some point when it was manufactured) which will cause you more damage than the thing you're trying to prevent by carrying it.

I'm personally a bit wary of introducing a relatively obscure security tool into my setup, to protect against a rare possible attack. The chance that I'll get caught copy-pasting a compromised URL into my terminal is fairly small, and there's also a small chance I'll compromise my system either now or at some later point via a supply chain attack if I use the tool. Which chance is bigger?

Maybe I've misunderstood, but isn't the whole point that it outputs to those other formats? So if you like the way Mermaid outputs look, you'd just output to that?

What you mean is that you don't find ASCII diagrams useful. Other people do (see other comments). I find them useful for Org Mode where I prefer to work with just text instead of having to keep separate image files together with my .org files. They're also easier to work with if you need to check them into a Git repo and you don't want to be bothered with something like Git LFS or git-annex. The fact that there are people who want them is enough to explain "why there's a push toward" them (assuming there is actually such a push).

I don't know how it works in France, but in common law systems opinions of other courts are always at least capable of being persuasive (not binding) precedent, so they are not irrelevant. Other courts can be, and often are, influenced by persuasive precedent when appropriate.

Not sure it's good as a starter distro, but other than that I agree. I was put off NixOS for a long time despite loving the principles behind it. Then a few weeks ago I had ChatGPT give me a short course on it, including flakes and the basics of the Nix language. I completed that in a few hours and achieved more than I ever had reading the Nix docs and blogs etc. Now I'm able to use an LLM to help me write flakes while also understanding what it is doing (I'm not a fan of blindly using AI generated code).

There are so many distros that it really depends on your use-case and it's hard to make a generic suggestion. Ubuntu is a common recommendation for first timers, mainly because as the most popular distro you'll easily be able to Google when you need help with something, and it also uses the most popular package format (.deb). There's also Linux Mint which is basically Ubuntu but with some of the latter's more questionable choices removed (e.g. snaps) and minus the big corp owner. By using one of these you'll also be learning skills relevant to Debian (which Ubuntu is derived from) which is a solid choice for servers.

Regardless of which distro you choose, your "desktop experience" will be mostly based on what desktop environment you pick, and you are free to switch between them regardless of distro. Ubuntu for example provides various installers that come with different DEs installed by default (they call them "flavours": https://ubuntu.com/desktop/flavors), but you can also just switch them after installation. I say "mostly" because some distros will also customise the DE a bit, so you might find some differences.

"Nicest desktop experience" is also too generic to really give a proper suggestion. There are DEs which aim to be modern and slick (e.g. GNOME, KDE Plasma, Cinnamon), lightweight (LXQt), or somewhere in between (Xfce). For power users there's a multitude of tiling window managers (where you control windows with a keyboard). Popular choices there are i3/sway or, lately, Niri. All of these are just examples, there are plenty more DEs / WMs to pick from.

Overall my suggestion would be to start with something straightforward (Mint would probably be my first choice here), try all the most popular DEs and pick the one you like, then eventually (months or years later) switch to a more advanced distro once you know more what your goals are and how you want to use the system. For example I'm in the middle of migrating to NixOS because I want a fully declarative system which gives the freedom to experiment without breaking your system because you can switch between different temporary environments or just rollback to previous generations. But I definitely wouldn't have been ready for that at the outset as it's way more complex than a more traditional distro.

You only have to run through a modern Windows installer to understand how screwed you are if you install it. Last time I did this for a disposable Windows VM (a couple of years ago) I remember having to click through a whole bunch of prompts asking about all the different types of data Microsoft wanted my computer to send them. Often the available answers weren't "yes" or "no" but more like "share all data" vs "share just some data". After that I recall being forced to sign up for an outlook account just to create a local login unless I unplugged my network cable during the install. I've heard they have closed that loophole in recent installers.

I'd already long since migrated away from Windows but if I'd been harbouring any lingering doubts, that was enough to remove them.