navi

@nprofile…8m52 for the kill one though
If the first argument is a negative integer, it shall be interpreted as a -signal_number option, not as a negative pid operand specifying a process group
to which i would read kill -1 is specified as kill -HUP (and thus should error out w/ “not enough arguments”)

2026-04-19 14:53:56 UTC

i find gcc's whole canadian cross thing for cross compiling, extremely funny, and also such a bad design
why do i have to rebuild all of gcc to get a custom backend =w=
it's even worse if i need to do it across 3 arches, say my build host is x86_64, my working laptop is aarch64, and i want to compile things for a risc-v board i'm working on, in the laptop
i first need to, from the system x86_64 gcc, build a x86_64 > aarch64 gcc, and then use that to build an aarch64 > risc-v cross compiler
over just, having the system gcc just capable of emitting aarch64 code that is in the codebase anyway, thus being able to just build gcc for aarch64 once and have that be able to build risc-v code, or whatever

2026-04-07 11:21:11 UTC

- reply

@nprofile…8m52 i really don't like libwayland at all
but hell every time i have to interact with libX11 or libxcb i wish it was libwayland instead
insanity

2026-04-07 09:21:28 UTC

/**
* Generic XI2 event. All XI2 events have the same header.
*/
typedef struct {
int type; /* GenericEvent */
unsigned long serial; /* # of last request processed by server */
Bool send_event; /* true if this came from a SendEvent request */
Display *display; /* Display the event was read from */
int extension; /* XI extension offset */
int evtype;
Time time;
} XIEvent;
cool, makes sense, shared event header… okay let’s look at the events…
typedef struct {
int type; /* GenericEvent */
unsigned long serial; /* # of last request processed by server */
Bool send_event; /* true if this came from a SendEvent request */
Display *display; /* Display the event was read from */
int extension; /* XI extension offset */
int evtype; /* XI_DeviceChanged */
Time time;
int deviceid; /* id of the device that changed */
int sourceid; /* Source for the new classes. */
int reason; /* Reason for the change */
int num_classes;
XIAnyClassInfo **classes; /* same as in XIDeviceInfo */
} XIDeviceChangedEvent;
no no no no no not like that, why is XIEvent not embedded there?? the fuck
i bet the implementation of those event functions is a cast-fest and strict aliasing nightmare

2026-04-06 18:33:34 UTC

> GetPointerDeviceRects((void *)0xffffffffffffffff, &device, &display)
this works
windows specialcases an invalid pointer value (that isn't NULL because ??????) to get a special response
why is this not NULL, microsoft, why must you hurt me

2026-04-06 17:55:39 UTC

- reply

@nprofile…8m52 i mean can you really even check it in a way that isn’t brittle as fuck
“it’s ascii and syntax consistent” is the closest you can get, but really the only way to solve this is to introduce a SHPATH just like we have CDPATH

2026-04-06 17:45:31 UTC

$ . env
/bin/env:6: parse error near ')'
i always forget it does do that

2026-04-05 21:34:50 UTC

sometimes i wish people would use the c type system properly
instead of "everything is number"

2026-04-05 20:01:35 UTC

i still don't understand why systemd is pushing for /efi, when like either /boot, or /boot/efi, make so much more sense

2026-04-05 13:35:48 UTC

:lizzy:: "the vim community is very normal and well adjusted"

2026-04-04 21:38:30 UTC

TIL sudo has a plugin system
and one of the plugins it has, is a python wrapper
so not only does the auth-bearing setuid binary load dynamic code, it may end up executing python while at it

2026-04-03 20:08:32 UTC

- reply

@nprofile…8m52 i mean my current iron is "i can hold the tip of it to the solder for a solid couple minutes and nothing happens" kinda deal so

2026-04-03 20:07:21 UTC

what is a good soldering iron
is the pinecil any good, or what should i get

2026-04-02 14:05:12 UTC

- reply

@nprofile…8m52 so both our april fools posts were something real underneath huh

2026-04-02 02:59:41 UTC

- reply

@nprofile…k3ke @nprofile…dpl5 as of posix 2024 -e and -o pipefail are both part of posix sh
-u is the only bashism now

2026-04-01 23:05:31 UTC

openrc’s reexport vars are now also being reexported by pam for user services!
so on the next openrc version, doing rc-update --user add dbus boot will be all one needs to do to have dbus working and the variable properly set on your login environment!
only applies for services in the boot runlevel, as some sort of special semantic for that runlevel – having it mean “synchronization point for core services of the {system,user}”
not sure if i am a fan of the semantic or not, but it’s the cleanest i came up with

2026-04-01 14:38:16 UTC

- reply

@nprofile…4fmn (the hurd support is real btw, just the rest isn't)

2026-04-01 14:26:56 UTC

https://www.gentoo.org/news/2026/04/01/gentoo-hurd.html
lesgo :3

2026-03-31 19:03:16 UTC

LLMs as (very roughly) explained to me: machine that works by mapping tokens to concepts in a N-dimensional vector
LLM company: but what if regex tho
like really could they not use the model to know if a word is roughly in the “negative” area or smth?

2026-03-31 17:51:49 UTC

i think people take "the unix philosophy" or such way too serious
and by that i mean both those who like it and those who dismiss it
for me it's just vibes, try my best to make things modular and interchangeable, and then go "huh, neat! feels unix-y... anyway moving on"

2026-03-31 15:56:58 UTC

Dame da ne dame yo dame na no yo

2026-03-29 12:56:53 UTC

most painful bit of testing some openrc features is the whole install reboot loop
rebooting my machine is so painful...
(yes i could do a vm but consider, lazy)
(whenever i'm not-lazy i'll set up a gardenhouse florist image to use in a test harness)

2026-03-27 23:29:01 UTC

i think the fact that there's people that 'look up' to me will always be a bit of a wild concept
like what do u meeeeaan i'm just a gal that yaps too much, has waayy too many ideas and opinions, and about double as many unfinished projects
=w=

2026-03-27 14:31:47 UTC

- reply

@nprofile…8m52 yeah but `need` means the service fails to start, which shouldn't be the case for dbus here so, will test out changing that to a `use` dbus probs

2026-03-27 14:08:38 UTC

- reply

@nprofile…8m52 and it might have been my mistake even
iirc pipewire doesn't need dbus if you're not doing portal things, but i didn't know this back then (and am not sure of it yet now)

2026-03-26 13:57:16 UTC

- reply

@nprofile…8m52 @nprofile…0pqg
my gentoo package for akkoma was like "okay fuck it disable the network sandbox and call mix"
which i think i copied from your pleroma ebuild

2026-03-24 09:15:14 UTC

- reply

@nprofile…8m52 oh, yeah
the fighting is meant to be inspiratory yet hopeless in most good stories, to which i see as a point of no return has been reached and thus we must *not* let it be reached
(i never actually seen bubblegum crisis, i probably should one day)

2026-03-24 09:09:49 UTC

cyberpunk stories are cautionary tales, not how-to guides...

2026-03-24 00:13:36 UTC

oh, posix 2024's `read` had -d for a delimiter
that makes a lot of things so much easier to do, wtf
love it

2026-03-23 21:08:03 UTC

- reply

@nprofile…k3ke @nprofile…z3mr
yeah, unless you’re writing complex bootscripts (which is not ever the case for most people), your scripts will look like this, takend from pipewire-pulse’s user service as a practical example:
#!/sbin/openrc-run
depend() {
need pipewire dbus
}
supervisor=supervise-daemon
command=/usr/bin/pipewire
command_args="-c pipewire-pulse.conf"
and openrc-run does most of the heavy lifting

2026-03-23 16:24:03 UTC

- reply

@nprofile…z3mr
> a supervision system (ideally a dependency-based one) can be far nicer, more flexible, more readable, and actually meeting what most people expect out of a computer these days
i mean openrc is both, dependency based, provides supervision, and makes it so the most common init scripts consist of only about 3 variable definitions and nothing else
but it's also a init script under the hood, so if anyone is so inclined, they can go and modify things, replace the default start() with custom logic, or keep the default start and add extra things on start_pre(), stop_post(), whatever
this duality and flexibility is what made me like it so, enough to step up and take maintainership
imo openrc is very much "doing better" while still being shell, though i do appreciate the different directions other init systems went for a lot too

2026-03-22 13:23:46 UTC

saying "sideloading programs" in a linux desktop context is wild, please fucking don't
there's not supposed to be one and only "official" loading, where everything else is "side" to it
it's called installing programs

2026-03-20 20:21:55 UTC

- reply

@nprofile…k3ke i mean cgit is a cgi tool so it's executed anew for every request, so it couldn't really keep it in memory
though i guess one could put the cache in tmpfs and that would make it fully in-memory

2026-03-20 10:58:54 UTC

got to the point where i'm considering buying a discontinued tv panel from a chinese supplier, and pairing it with an aftermarket dumb driver board
just so i can have a tv without any "smart" shit into it

2026-03-19 12:56:33 UTC

systemd and xdg-desktop-portal/flatpak jumping the gun to implement age tracking fields is so on character it hurts, rushing it out, despite that
a) system76 folks are trying to give exceptions for open source software
b) afaik, california and colorado laws may only require a field, but alabama, utah, and NY, seem to require full verification, so, how will that be
i hate this timeline

2026-03-19 08:45:29 UTC

on "growing up tech literate, and parental controls"
i was allowed to use the computer we had at the house at around 6yo
my dad would install games for me, mom would watch videos with me
but crucially, it was a computer in the living room, anyone going by could see what i was doing there
i had to ask permission to use it, and would be allowed to use it for about 2~4h
--
by 13, i got my first personal laptop, a cheap hp 2-in-1, that lasted all the way until i was 17
by then, my dad had already taught me how to install windows, how to pirate games and movies, and i was basically the one fixing people's computer problems on the house
and i already knew how to use the internet, what to look at and what to not
--
i believe that a small child, preteen, shouldn't be left alone with an internet enabled device ever, nor be allowed to be glued to it all day
and that a teenager should already have been taught how to use the internet properly, and have at least some notion of viewer discretion
when a kid is already a teenager, there must be trust between parent and child
--
then, in my opinion, parental controls is a way to offload some of the parenting responsibility from the parent, to the device
(not much different to how age verification is, in orders of magnitude greater, offloading parental responsibility to the government and companies)
--
in an ideal world, parental controls would be just guardrails, for parents that want to be sure nothing bad goes through, that goes away around the time the kid becomes a teenager
but in our non-ideal world, it can and likely is used by abusive parents to control their (teenage) children, to prevent them from doing or seeing things the parents disagrees with (e.g. queer media)
so i cannot be in favor of parental controls on devices

2026-03-07 16:48:24 UTC

the only good AI policy, is a no-AI policy

2026-03-04 22:53:54 UTC

what is happening to fedi
i'm getting dozen of those mass-ping spam donation-begging bot posts a week, this didn't use to happen more than twice a year before
and now they're basically always coming from different accounts on different instances, but same format
the hell

2026-03-03 15:38:48 UTC

- reply

@nprofile…8m52 on my old debian vps i used to use ufw bc it came with it
so i think? whatever does the same, idk if ufw is still cool or not

2026-03-03 15:33:24 UTC

can someone recommend me what is the new preferable way to firewall a linux server

2026-03-03 00:00:35 UTC

damn guilty by... unknown association?
and who even are those instances, not once have anyone from their team reached out to ask if we're even aware of that
#nevent1q…dnyz

2026-03-02 20:33:59 UTC

i am not surprised, slightly disappointed but this was coming anyway
#nevent1q…2rea

2026-03-02 15:13:09 UTC

and here's why "language specific" unvetted and user-driven repositories are hell, and by extension why distro(-like) packager-driven package managers are good
#nevent1q…3d2r

2026-02-27 14:35:47 UTC

- reply

@nprofile…8m52 i'll stick to it because it's pretty much the only solution for multi-user synchronized calendaring, and supported everywhere
storing each VEVENT/VTODO/VJOURNAL as it's own file simplifies a lot, and ical being flat means i can stream it in and deconstruct it as i go

2026-02-27 14:30:36 UTC

- reply

@nprofile…8m52 i should actually read the wire protocol of caldav huh
i knew radicale keeps them in disk as separate files, but the endpoints were giving me a single ics per calendar
maybe it's the common pattern to store them split, merge them for transport, then re-split

2026-02-27 14:16:11 UTC

the most ???? design choice in ical for me, is, why is the whole calendar a single file ~w~
could've made the calendar a directory, and events be files, blep

2026-02-26 12:30:13 UTC

if i did everything right, wkd aware gpg clients should be able to transparently fetch and encrypt mail to navi@vlhl.dev now

2026-02-22 19:20:54 UTC

if proton mail were serious about improving e2ee email, they would've worked to have a standard extension for e2ee via smtp that other MTAs could implement, so that encrypted emails would not be limited to proton mail's non distributed, non self-hostable, proprietary smtp servers

2026-02-22 16:19:50 UTC

> The DRM (graphics) subsystem has been an early adopter of the Rust language. It was still perhaps surprising, though, when Airlie (the DRM maintainer) said that the subsystem is only ""about a year away"" from disallowing new drivers written in C and requiring the use of Rust.
"no rust in core features" and "no forcing people to write rust" didn't last long at all, did it

2026-02-17 23:15:59 UTC

how bad is it to create a new fedi account with the same username in the same instance as a previously created-then-deleted account?
asking because fedi is fucked up and prone to blowing up

2026-02-17 13:08:29 UTC

- reply

@nprofile…8m52 @nprofile…xj4x
going from srv records to "if you host multiple things, well time to set up a subdomain for everything and plop a proxy in the middle" where the proxy is required to handle tls and copy bytes as well
it's pain for me

2026-02-09 10:20:41 UTC

tried calling the hospital to cancel my endo appointment
15 minutes on wait, someone picks up
they say a word in spanish i don't know, so i say i don't understand
they repeat the same word, and i again say i don't understand it
they then hang the phone up
i have no problem reading spanish at all, and only a slight difficulty with person-to-person but it's manageable
over the phone however it's really hard -- but i can't cancel this appointment in any way except over the phone, the app and website have the cancel option blocked out
holy fuck
honestly having ways to do something that doesn't involve a phone call should be a matter of accessibility

2026-02-06 17:24:56 UTC

even arm dev boards are now riddled with "AI NPU" """features""" huh
don't want any of that

2026-02-05 09:27:29 UTC

- reply

@nprofile…4fmn @nprofile…jt0w
on windows it might not have an xinput driver that games expect?
on linux everything uses evdev afaik so it should just work (the steam deck controller works without steam on my gentoo install just fine)

2026-02-02 16:58:10 UTC

i think akkoma's forgejo is broken, the source tarballs just don't work

2026-01-28 20:37:16 UTC

- reply

@nprofile…8m52 one of the best things i feel like i did for openrc was the very simple thing of git log $last_tag..HEAD --oneline | git tag -se -F - $new_version when tagging – atomic commits means my change log is already pre made, then i just fill in more important to note things on top

2026-01-22 14:12:34 UTC

i really hate this idea that every non-trivial struct in c must be an opaque heap allocated pointer because people can't trust their users to read documentation, or what?
abi stability is another topic, and that one is more valid of a reason, but i'd still say it's not that hard, use sonames properly -- but if you *know* the struct definition will be changing a lot, then it's fair (but consider structuring your api differently, be modular and rely less on one single big struct type)

2026-01-21 20:40:28 UTC

distro package managers do allow any package to be installed from offline distfiles, but all the tools to do that kind of assume you have at least *one* system with said package manager installed for solving the dependency tree and fetching those files
would be neat if there could be a website that would do the solving and downloading, then give you an archive with all necessary dependency packages/sources inside, then that step could be done in a phone and copied over to the system where one can run just the install step
#nevent1q…tyrj

2026-01-15 17:06:52 UTC

- reply

@nprofile…8m52 glibc's parses it in c, and another libc which i forgot right now
though i'd be willing to eventually implement it for musl in c if that was wanted

2026-01-14 22:39:59 UTC

there's some places where using wordexp in openrc would be nice
though i'm conflicted since musl implements it by spawning a shell, and i'd like to not do that just to parse some configs

2025-11-28 11:19:49 UTC

- reply

@nprofile…3gt0 @nprofile…spd8 i've been using it for 4 and the only thing i'd switch to is my own distro
and oh I do *not* have time to maintain a distro

2025-11-27 21:23:29 UTC

i'm sorry to everyone that hates usb-c but at least it's something other than having a whole ass laptop be rendered unusable because the power supply using a non-standard barrel plug died

2025-11-19 19:42:48 UTC

- reply

@nprofile…9e6e i'd rather not

2025-11-19 19:22:11 UTC

wanna start working on the c compiler but really don't wanna work on a lexer and parser rn
might just end up using lex and yacc

2025-10-18 15:51:46 UTC

now that i have money, i really wanna get into diy fpv drones tbh

2025-10-12 18:15:48 UTC

- reply

@nprofile…m3qg me too tho i'm happy about it?
specially with the new standards actually improving the language

2025-10-09 16:01:45 UTC

mnt reform is cool but i can't work on it, sadly
my job requires me running x86_64 programs with at least somewhat decent performance

2025-10-05 23:21:23 UTC

- reply

@nprofile…m3qg i could do it but i'd have to be very careful to not put anything i don't want to show, in the streaming workspace
versus having obs stream specific windows, that risk doesn't really exist
(though i do plan on having a jaci feature similar to niri's where "sensitive" toplevels can be marked and would be blacked out while recording the screen)

2025-10-05 23:02:32 UTC

saddest part of being a tiling window manager user is that i can't make a streaming layout with a "desktop" windows-y look
my desktop windows don't have decorations so it would just look like an normal streaming layout
(and i can't get myself to put the decorations in anyway bc that'd feel like lying, lol)

2025-10-04 22:35:04 UTC

- reply

@nprofile…h4yc designing something so it's already small parts from the get go is easier to write and to maintain
also "smaller parts" can also mean libraries with a self-contained apis, doesn't *need* to be multiple processes doing ipc

2025-10-04 22:28:45 UTC

- reply

@nprofile…h4yc a project with 40 .c files is a project too big imo and should be split into smaller parts. i did specifically mention "small, self-contained utilities", which is most of my tools
and i did mean actual globals and tu-local "globals", though more of the latter than the former

2025-10-03 21:33:06 UTC

funny how my instance on global variables and such flipped the more i work with unix code
small, self-contained utilities gain a lot in simplicity by keeping global state as global variables

2025-10-03 15:32:48 UTC

okay so we can't use READY=1 for gnome-shell under openrc
because gnome-shell uses libsystemd's sd_notify for it, and while we do support NOTIFY_SOCKET, we don't provide libsystemd, obviously
i hate libsystemd so much. this could've been a standalone library and thus work with any init system implementing their notify-socket protocol
but NOOOOOOO it goes in the libsystemd-shaped hole

2025-10-03 01:29:26 UTC

doing the activation environment for openrc, and now i wonder if per-service sharing of variables is of any use
use cases we have are, dbus socket, wayland/x11 display, and maybe some other minor cases?
i made it so far with the notion of each service can export a set of variables, and their dependees would automatically import those
the advantage ends up being that if say, dbus exports a variable, only services that do `need dbus` or `use dbus` would get that variable in their environment
but that comes with UI questions
how does a service specify a variable that is exported to it's dependees, and how does a user query said environment, say, if they want those variables to be exported to their login shell
that isn't quite needed if we only have one global activation environment

2025-10-01 15:48:28 UTC

- reply

@nprofile…h4yc i don't really have anything loose in my bag to puncture it
and anything from the outside would need to go through a lot of fabric, and possibly my laptop

2025-10-01 14:49:54 UTC

the steam deck case kinda sucks bc i can't access the steam deck itself without taking the case out of my bag
but i'm also not comfortable with carrying the deck without a case bc i don't want the analog sticks to get damaged
i wonder if there's such a thing as a case without one of the sides, that the deck just slides in and out

2025-09-10 13:49:09 UTC

next openrc release will likely finally have a init.d path in /usr (or /lib for split-usr systems) for packages to install init scripts to
so /etc won't need to be a mix of system and user provided scripts

2025-09-10 10:43:46 UTC

5h of sleep is fine, right, right??

2025-09-10 01:02:55 UTC

- reply

@nprofile…m3qg @nprofile…9klx meson as a project does accept they can't fulfill every use case and is happy to coexist with other build systems
so basically if there's a problem that can't be done with meson, they'll attempt to find a good solution that helps everyone without worsening the build system as a whole, if that happens, good, and if not, they just say that meson is not the tool for said use case
and honestly meson is actually a lot more flexible than it initially seems (see my post of using meson as a meta-build-system for 15 subprojects, those being meson, cmake, and autotools, that get compiled for x86_64 and i686, some for unix, some as windows PEs)

2025-09-09 22:06:34 UTC

- reply

@nprofile…m3qg i do understand how symlinks work, but even then it's a pain, because if i want shell completion, i need to either cd into the target dir, or do the completion first and then mentally calculate how many `..` to prepend to the link target
and even without shell completion i still need to do the logical dance of adding ../
meanwhile -r i just, do it like cp and not worry about mapping it in my head

2025-09-09 21:55:02 UTC

`ln -r` is so comfy
not posix but really comfy
(tbh resolving one path relative to another should be on a unix libc, it's as useful as realpath)

2025-09-07 04:02:17 UTC

i'm deeply sad that i can't use argv[0] meaningfully with interpreted scripts

2025-09-06 22:20:55 UTC

would there be a problem for me to use the sticky bit on files for application specific purposes?
i have 2 cases where i could use it, both are plain text files

2025-09-04 14:15:12 UTC

- reply

@nprofile…ynjf you can achieve most of that without rendering on the gpu at 120fps by just being smart about it, by doing proper damage tracking
but also 120 fps for scrolling is kinda too much asw imo, specially since now your power usage goes way up as your doing gpu things just to get scrolling a little smoother

2025-09-03 22:51:29 UTC

- reply

@nprofile…m3qg even image and video editors, if it were me i'd make the main canvas a sub-surface, and only render that at a higher fps
leaving the gui itself at "only draws when there's an update"

2025-09-03 22:00:13 UTC

In 2023, the authors of the Zed text editor announced a GPUI framework that draws all UI elements using the GPU at 120 fps
i am sorry but why do they need to render a text editor at 120 frames per second?
any gui program for that matter, 120fps is just, so unnecessary for something that is mostly static and at most scrolls text

2025-09-03 15:01:54 UTC

- on one hand projects adding rust with people wanting to make the new rust code MIT
- on another the BSD alternative in c is obviously under BSD
this is pain

2025-08-21 20:53:17 UTC

a instantiable snooze init for openrc might be a good alternative to systemd timers, but without the drawback of always needing two files (one timer and one service)
having timers as services means one can depend on a timer, but since the command is not tied to a service, no duplication would be needed
tho i'd add some syntax sugar to trigger a service from a timer i think

2025-08-21 16:05:29 UTC

why do we have user, and a group named after the user? it's kinda odd that we create a group that (technically) should include only that one user
and honestly, with ACLs, user/UID is kinda of just a one-member special group anyway, so we end up with two one-member special groups... why?

2025-08-20 18:07:41 UTC

- reply

@nprofile…m3qg i kinda want to try but also kinda not
i just feel like ditching /usr makes way more sense than selectively moving somethings into /usr but some things not

2025-08-20 17:57:09 UTC

usr-merge but it’s /usr -> . instead

2025-08-20 01:23:00 UTC

single header (non-macro) libraries are cursed
can we stop doing them
a library that implements containers via macros can only be done as a single header, so i'm talking about the "lazy" `#define FOONYA_IMPLEMENTATION`-style ones

2025-08-16 19:43:31 UTC

./setup-$game.exe /VERYSILENT /DIR=Z://home/navi/games/$game
cli flags for windows programs are so silly, and scream-y

2025-08-16 15:35:54 UTC

i wish man-pages-posix would have not just the utilities, but also the chapters
the Shell Command Language (V3_chap02.html) is the only page i access from posix that isn't on the manpages package...

2025-08-13 22:24:49 UTC

i honestly can't with grapheneos devs saying that ios and stock android perform better privacy-wise than desktop linux

2025-08-11 15:32:36 UTC

keyboard -> keyboards_from_resource
pointer -> pointers_from_resource
touch -> ????
touches_from_resource sucks, touch_from_resource is singular, bleh

2025-08-10 22:20:51 UTC

- reply

@nprofile…m3qg the ioctl enums are not stable like syscalls...?
does that mean any kernel update might break not only clients using ioctl directly, but also libdrm or libevdev or any library wrapping them?
if so, how can anyone use ioctls for anything...

2025-08-10 21:04:47 UTC

another idea for my "DE"
shell hook that pings you when a long command finishes, but, only if the terminal window is either not in focus, or not visible
implemented via a check to ext-foreign-toplevel

2025-08-09 15:25:18 UTC

c's array-to-pointer decay in function declarations is one of the saddest things to me, because we can't fix it
passing in a function actually passes a pointer (or rather, "by reference"), but wrap that in a struct, and now it's passed by value
add to that how you can't "return" an array from a function, but wrap that array in a struct, now you can
it's inconsistent, but it's impossible to change this since it'd break code