Which Lisp though?

Scheme, Common Lisp, Lisp2 or one of the countless others that are incompatible?

What is wrong with wayland.

systemd does a lot of things very right and a mayor reason that linux is in a way better place than 15 years ago. 27 unmaintained tools are not better than one that is professionally developed.

systemd has me in two places. Like, on one side, it's just SO much better and easier than the rest - at least to me, someone who doesn't work in IT and just wants his stuff to work. On the other, init freedom is essential, and we need the alternatives to stay alive.

In covering the use cases of X it became just as much a mess.

Me when I got my commercial NAS with 7 years of support, I was thinking: "who even keeps hardware for this long?"

Me, 7 years later: "What do you mean EOL?!"

I certainly learned to care for and appreciate the hardware these years.

I can concede that in games programming (which I'm not a game dev so I often forget about) x, y, and z coordinates are a valid exception.

I would disagree about w and h for width and height, typing them out feels like i for index

I worry about the phrase, "you'll immediately know what the variable is for at a glance even though it's a single character"

how often have you heard developers say that something is simple and it's just because they wrote it? maybe it is obvious, but if it were more verbose it would be just as obvious but without any possibility of misunderstanding

and even if it's obvious now there's no guarantee a future developer doesn't pull that variable out and mutate it or do something else with it, and maybe they should change it at that point but the effort to have it verbosely named from the start is miniscule

There's about 0.2% of people still using internet explorer, per one market share site I found.

Which works out to potentially millions of people.

Advertising is education with misinformation

I can't wait for the crash, so that we can get more sustainable AI and access to the hardware to run it.

I do my web development on a 2005 era windows XP machine without the unofficial SP4 because if it will run there (without silverlight) I know for damn sure it will run well on anything made today.

Nah, that would be like refusing to use a calculator, because they make them from unrecyclable plastic.

If I was king of the world, I would rule that all AI models were based on stealing from all the knowledge of humanity. Since it cannot be un-stolen, I would decree that all copyright is unenforceable and all current proprietary data must be public domain forever as restitution for the abuse.

Other languages took perl's regex engine

If you are getting paid for engineering software, your labour should be licenced in an appropriate jurisdiction, and you should be liable to losing said licence if you do shit job.

Systemd solves a very important problem that was defined extremely well.

It also solves it badly, and solved a whole lot of other badly defined problems that it shouldn't.

There's nothing in Functional Programming that stops you from writing in an Object Oriented idiom...

React is a plague. useState and all the other use hooks.

Mozilla has totally lost their way. But using any chromium-based browser is still bowing down to Google’s near monopolistic control over web ‘standards’.

I’d care if they had a Linux version

Not Op but..

OO languages have classes with inheritance based polymorphism. I don't see struts + methods as enough to qualify.

Functional languages have functions which can take functions as arguments and function composition as first class features.

Linq

AKA - We have Monads at home...

Rust is really missing something like do syntax. It's half-way there with the semicolon and question marks, but it's missing an idiom for .map_or_else and to generalize the thing to work with a trait.

APIs generally don’t need many alternatives, only implementations do.
There’s no reason devs should be trying to provide init files in multiple formats.
If the systemd team actually did something bad, then people could just make a different implementation of systemd without requiring a different format for init files.
Arguing for more init systems is like saying we need more APIs to compete with both Wayland and X11, instead of just talking about implementations of those protocols.

Zed sucks in many of the ways that modern editors suck.

Autocompleting on characters that appear in code, and autocompleting in a way that overwrites the code you have already written are two incredibly stupid decisions that put a mark over any tool that adopts any one of them.

But then, Zed almost working, but in a way that still sucks was enough to motivate me into setting my Emacs properly, in a way that doesn't suck. So Zed gave me a good editor.

Is this position against lookup or read-only db tables?

Largely, yes.

The kind of cases where I've seen people stick things in the database that I thought would better be elsewhere:

• If it's small and doesn't need to change between releases, you can save all the moving-part kind of overhead of pulling that data from the database just by throwing that data in the source code. Often times people do this just so they can have a "foreign key" to something in the data in question. But a nice, semantic "enum" sort of field in the database can refer to something static that's already in memory no problem.
• If it needs to change at an administrator's or developer's (or manager's) whim, it's probably configuration, not dynamic data. In the Java ecosystem, Spring Cloud Config makes it easy to change configuration without releasing the code. (I mean, in a perfect world, releasing should be as easy and straightforward as changing configuration, but not everyone lives in a company where all the managers have read The Phoenix Project and caught the DevOps bug.) Putting configuration in the database tends to create a big audit trail gap, unless you spend a bunch of effort building a custom audit trail kind of system.
• If it's big and not (very?) relational, it may be better suited for files in a directory. This can be true even if it does change based on user input. (Users uploading images, for instance.)

Mind you, certain kinds of behind-the-scenes users can still qualify as "users" for purposes of considering whether "it needs to change based on user input". For instance, if your marketing department (without software engineering expertise) needs to be able to mess around (without software engineering assistance) with marketing materials displayed on customer-facing websites, you probably need a content-management system of some sort, and I don't think the database is necessarily a bad place to store a lot of that information. (Things like images probably shouldn't go in the database, but metadata about those images are probably quite reasonable data for the database.)

Also, I'm rethinking the term "user input". I could certainly see it being reasonable for automated systems to put data in the database. Like, if you have a scraper doing stuff to be displayed to end users, that doesn't belong in the source code or configuration, and if it's very relatioal, it could well belong in the database. Maybe it'd be better to say "if it doesn't change based on client input, it doesn't belong in the database."

I do really disagree here. Name lengths should be proportional to the size of their scope.

If your function has 1 line, your local variables should almost all have 1 letter.

Here your crown sir 👑

If there are spelling mistakes in your codebase, that’s a huge smell

This is my argument. Do init. Don't pull in literally everything else because you think you know better, like system logging.

Bad comparison, it's like refusing to use a calculator because it uses 200 times the amount of energy of existing calculators, was trained on stolen data, and despite all that gets the answer wrong

OOP dogma isn't just bad, it's actively harmful. OOP leads to unnecessary boilerplate. Any boilerplate that is not absolutely necessary is noise and bad. Full stop.

Private (Getters/setters): Strictly not a necessity in any sense of the word.

Short functions everywhere.

Short files everywhere.

I prefer to see the steps of execution that are actually happening in the real world on hardware. Not an ivory tower of code that happens to execute.

It's funny how both anti-AI and pro-AI folks think they're the one in the middle. But the anti-AI crowd is proclaiming that any use is the devil!, and one drop means the whole project is forever worthless, and ten minutes will make your brain dribble out your ears... while the pro-AI crowd just says it's probably fine. Not miraculous. Flawed and suspect. Local would be better. Basically would you please just relax about the chatbot, and condemn companies doing mass layoffs for any reason?

When even Cory Doctorow is going 'LLMs have problems, but here's how I use them,' pack it in. This fight is over.

I saw a nice counter to the "a poor craftsman..." analogy the other day that basically boiled down to a master craftsman will have very high standards about the tools they use.

It sounds like you’re comparing the most reasonable views of the pro-AI crowd to the most reactionary views of the anti-AI crowd

Nobody hates Javascript like people who use Javascript.

@Humanius @danhab99 Formating is so 1990. Modern would be to store Program code as database. Formating and UX of program code would be separated and an IDE would show Program code in the user preferred way.But every Developer: "Hey modern way to store structured Data, you are wrong for my Work."#structureddata #developer #conservatives #notonmywatch #codeformatting

Perl is write-only code. Larry Wall raised a monkey's paw and asked for a language that works the way programmers want. So most scripts are the first thing someone tried - unfiltered by rigid syntax putting your thoughts in order, or compiler warnings bringing side effects to mind, or even names forcing you to decide what a variable is. An uncommented Perl script is raw brain patterns decoded as ASCII.

Thank god it runs like shit. Fast C isn't exactly gorgeous, but any project that compiles is proof the coder fought the machine and won.

Yeah, I wholeheartedly agree with this. And it's not only frustrating that PHP doesn't live up to the high bar, every ""major"" update to PHP (that you graciously only get once a year) is only a small step in the right direction. They try to modernize it, buuut also try to keep it as backwards compatible as possible.

And I think I've also seen a counter to the "a master craftsman will have very high standards" argument that boils down to "a master craftsman can work with any tool regardless of quality" but why can't we have higher standards in the programming world?

I have no idea how anyone who ever used it for anything actually useful can be defending it

Node.js TS is great

PHP is not only for the web/short lived scripts

Cloud computing is the biggest breach of privacy out there that we've adopted whole hearted.

Init needs to do system logging when you want to collect it from the daemons stdout, which is the logical place to collect it from. Havin apps handle their own log files is a bad time at best and dangerous at worst.

Right. You can still put that through a linter; the linter just wouldn't be concerned with human-readability.

In any case, my editor would not hardcode my display preferences, so if I open and save a document, nothing changes except what I actually changed.

It’s a bit of a kitchen sink language now. I haven’t used it professionally in quite a while but using it for Godot (anything but gdscript) has me scratching my head over some of the architectural decisions. It seems to be led by „if people do a certain form often enough, add a shorthand to the language“, which is one way of cutting boilerplate, but does lead to something very hard-to-parse if you’re not familiar with it.

Having worked at one for a for a few years, I can't say I agree, nor do I technically disagree either.

Theres nuance.

Because that's all the Fediverse contains.

In this thread, there's people reposting "never slop" doodles, performing ingroup loyalty over thought-terminating cliches. All my efforts to engage with that have made clear they mean never use any AI for anything ever, and then they repeat their chants as empty signalling, or use one of their made-up slurs for anyone who's not virulently against all forms of spicy autocomplete.

There's entire communities for that attitude. Entire instances. It's immediately the dominant tone in any discussion of companies using AI for literally anything. Even if the AI content was removed before a game launched, the fact it was there has indelibly stained the publisher's whole catalog, past and future.

Meanwhile the opposition to that is just... would you relax, please? There's no "boosters" on this service. Nobody's snorting a line and raving about basilisks. Vanishingly few dinguses have sneered 'the future is now, old man!' and they find zero support. Even measured responses to shrill absolutes begin 'now I hate AI as much as the next person, buuut.'

Pro-AI on Lemmy means anything short of kneejerk opposition, on questions like 'can generated images ever be art.' Everyone suddenly developed a crystal clear philosophy on the nature of beauty. Those philosophies don't agree, but the people do, which is... indicative. Especially when you discuss those claims seriously, and any illustrative counterexamples are treated as a personal attack.

Recently someone repeated claims that LLMs cause instant cognitive impairment, and I pointed to the researchers behind the study begging people to stop saying that. Do you want to guess how that went? It wasn't a rational exchange of comparably reasonable positions... and the little numbers showed the crowd was not on my side. Or the researchers' side. They were perfectly happy with their tabloid narrative, thank you very much.

Thanks!

If someone wanted to take your rights away, maybe even didn't believe you should be allowed to be alive, would you be unprofessional for not wanting to use software they make? Would you be unprofessional for making your own version of that if the project refused to remove them as the lead? No. You wouldn't. A way to solve that problem as a professional is to fork the project. What's not professional is calling people doing that unprofessional.