@retoor Ah yes, the GNU Make architecture. Orchid started out that way, but I rely on the preprocessor being very very fast so from the day I had 200 token functions I needed at the very least a tokenizer that interns names.

spring weather means I'm either drowsy because of rapid changes in air pressure, or drowsy because of the heat, or drowsy because my pollen allergy is acting up. I get like an hour of productivity each day.

cc-by-nc-sa does pretty much that

it took me a week the first time, and the result was kinda crap but functional.

@AlgoRythm intercal?

@Liebranca the behaviour of realloc always felt common sense to me. This word doesn't really come up in other languages, except with vectors, but there it explicitly involves moving the elements.

There are two related main reasons to find pointers difficult:

1. the language errors, even compiler errors don't give beginners enough information and usually mention rules that don't suggest the behaviour that necessitates them

2. the code doesn't give experts enough information and usually mentions low level details that don't suggest the high level structure that uses them.

A pointer could be a borrow, an owned object that may or may not have live borrows and thus isn't generally safe to move or destroy, an owned or borrowed buffer, a nullable value the author felt too large to zero, an out parameter that may or may not be pre-initialized.

** has legitimate meanings for every combination of the above. If the exact parameters of both pointers aren't specified (and I've never seen them specified with sufficient rigor outside the standard itself) there's no point guessing because your chances are slim.

Even still, I feel like defining a struct that's entirely local to a function that returns it and another that consumes it is better than any alternative. I'm not sure what makes a type expensive, I guess duties like representing a different type but with constraints?

A software developer is a person who develops software.

@jestdotty i3 like the other Xorg WMs will suffer from an even worse lack of manpower and perspective than they already did, until the end of time. Any particular reason you aren't starting with Wayland?

The only desktop I ever really found aesthetically pleasing was Windows 10. Flat, functional buttons, simple clear icons, no extra borders, shading, curves, and the first thing I always did was to disable transparency effects like blur backgrounds. At least KDE allows me to replicate that.

In title: ... retrofit[1] ...

In body:

[1]: [retrofit](https://en.wiktionary.org/wiki/...)

Append "your mom can tell you what that means" to every message

@jiraTicket "if it pleases you" or "if you like" is a poor translation of "s'il vous plait", it specifically means "please" in French. This is a fine example of mirror translation losing parts of the original meaning. I flinch because 3/4 of "rsvp" already stands for "please", and the remaining 1/4 is the same root in English. If you want to spell "please" in english, then the translation is "please respond".

@Lensflare Elixir also handles graphemes, and Rust has a very popular crate for this purpose. The problem is that unlike codepoints, with graphemes you can't take the first grapheme from a byte stream without lookahead, because any valid grapheme can also be the first half of a combination using an infix ZWJ, and indeed the most common way to denote some Arabic letters uses an infix ZWJ.

I find React fairly testable as far as frontend goes. Flushing the task queue is clumsy but other than that it's fine IMO.

Rust doesn't represent strings as character arrays either.

I don't know Go, but

- arrays are evenly spaced in low level languages for O(1) indexing
- strings are Unicode in modern languages
- strings are more compact than an array of their code points, because they're understood to consist overwhelmingly of low values, and also represent most of a program's memory usage.

that's cool, OCR is pretty old tech though so in a different world this could be standard OS functionality.

In our reality though, even just copying text that's actually presented to the OS as text from applications is a premium feature.

@awesomeest Törzsidő, literally trunk-time or tribe-time

@jiraTicket yes, I don't know what it's called in English

very frustratingly, it looks like this becomes very easy via any single one of the famous pending trait features; negative statements, overrides, const generic conditions...

I have to check if it works via GATs by some miracle, although I bet they have a mighty convenient limitation to avoid it.

ER denormalization

@atheist Hungarian labour law is a joke and I'm on probation anyway

@kiki Does common lisp support any kind of type checking or static verification?

@kiki I didn't observe that kind of behaviour yet, but I guess it's enough for one in 50 people to be condescending for newcomers to feel unwelcome.

@tosensei Superior to all general purpose languages that I know of, and a couple DSLs.

I tend to agree that the community is annoying though, they made a language with good qualities and maxima, therefore every quality and maxim of the language must be good and anything that doesn't encode well in it isn't worth writing.

but it IS superior

If you want to take a crack at this, take a look at the pattern below, the solution would almost definitely have to employ a modified version of it. The article is a step-by-step explainer written by me, but the original forum question with two variants of the pattern is linked in the first paragraph.

https://lbfalvy.com/blog/...