navi

♫♪ never fade away ♪♫

openrc maintainer, works on wine

<3 lizzy
<3 @[email protected]

Public Key

npub1f67qrzm95xe497g0jda70x6uxwr9ru2j39sg38axfj4ff05c979qmps5qm

NIP-05 Address

Profile Code

nprofile1qqsya0qp3dj6rv6jly8ex7l8ndwr8pj379fgjcygn7nye255h6vzlzspz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dszd6acp

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub1f67qrzm95xe497g0jda70x6uxwr9ru2j39sg38axfj4ff05c979qmps5qm

Show more details

Last Notes

2026-08-03 22:06:05 GMT

- reply

@nprofile…8m52 considering the -1 on the text box, i think it's something on wine erroring out and the game taking that as a correct value

2026-08-03 22:01:59 GMT

i love old games
https://social.vlhl.dev/media/b5d4b54b4798ad4d3684edcdbb299a06d75d56b7d8515e7ff9661c866274d734.png

2026-05-24 17:31:56 GMT

there’s something so satisfactory about managing bitflags w/ &= ~FLAG and |= FLAG

2026-05-05 15:06:58 GMT

i'm happy to see folks working on better openrc debian support
having that available as a recommendation over devuan would be so great

2026-05-03 13:11:30 GMT

the point where that's needed is the point i'd say the tool shouldn't be written in shell, or should be simplified over multiple tools and then those go in a shell script
#nevent1q…ma6p

2026-04-23 13:56:32 GMT

- reply

@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 GMT

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 09:21:28 GMT

/**
* 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 GMT

> 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 17:45:31 GMT

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

2026-04-05 20:01:35 GMT

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

2026-04-03 20:07:21 GMT

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

2026-04-01 14:38:16 GMT

- reply

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

2026-04-01 14:26:56 GMT

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

2026-03-27 23:29:01 GMT

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 GMT

- 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 GMT

- 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 GMT

- 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 GMT

- 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 GMT

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

2026-03-23 21:08:03 GMT

- 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 GMT

- 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 GMT

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 10:58:54 GMT

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

2026-03-04 22:53:54 GMT

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 GMT

- 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 GMT

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

2026-03-03 00:00:35 GMT

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 GMT

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

2026-02-27 14:35:47 GMT

- 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 GMT

- 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 GMT

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 GMT

if i did everything right, wkd aware gpg clients should be able to transparently fetch and encrypt mail to [email protected] now

2026-02-22 19:20:54 GMT

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 GMT

> 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

2026-02-05 09:27:29 GMT

- 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)

2026-01-15 17:06:52 GMT

- 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 GMT

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

2025-12-30 21:51:29 GMT

i wonder if i can do x11 over waypipe
since ssh -X is unusably slow and waypipe is actually functional for things that don't use the x toolkit

2025-12-22 15:06:01 GMT

"restart to keep using firefox" -- statements dreamed up by the utterly deranged

2025-12-16 16:03:34 GMT

- reply

@nprofile…3gt0
> And I don't think write perms is enough, like imagine overriding/overlaying /tmp.
i was thinking of write perms and ownership (but forgot to mention), which would work here (or a number of heuristics, more later)
i'm not a fan of fuse, at least not for traditional filesystems-- vfat, ext4, nfs, etc, all of those seems like something a user could mount in their own home frequently, and having them implemented twice as fuse and non-fuse is very bleh imo
plus, fusermount is a suid helper, if users could just mount things, that wouldn't need to be the case, you could continue having fuse for out-of-kernel file systems, but now be able to mount them without suid and user-space "ran as root" heuristics -- plus, the same heuristics that work for fuse could be used on "user mounts", since, if it's enough for a setuid program, it should be enough for the generic mechanism
adding extra binaries is, imo, irrelevant, if the user can write to a dir, it can copy a binary with +x there... and if the mountpoint is already under noexec, then the user-mounted device would also be forced to have noexec, just like nosuid/nodev but conditional, as imo, mounting executables is valid -- e.g. i could have games installed to a removable hard drive

2025-12-16 15:28:31 GMT

another question, why is mount effectively root only (unless user is set on fstab)?
i’ve seen arguments such as:
the device could contain a setuid binary
force user mounts to be nosuid
the device could contain device nodes with insecure permissions
force user mounts to be nodev
the device could contain files that could trick already-installed suid binaries
only allow users to mount over directories they have write perms, since if they already do, then nothing stops them from doing that anyway
is there more reasons to not allow users to mount devices? (without a suid or daemon helper, i mean)

2025-12-16 11:01:14 GMT

what are the use cases for dd now a days?
the only thing i used it for was flashing isos in usbs, but cp iso /dev/xxx works just fine by now

2025-12-14 03:14:01 GMT

> receives an "invoice" email from "[email protected]"
huh, didn't know i had a noreply and invoice system setup for my own personal domain /j
---
honestly those scam attempts are more creative than i thought, maybe their bot notices custom domains and assumes my address is part of some company

2025-12-11 12:40:59 GMT

> In C23 this initializes all members to zero, and all pointers to NULL. This is a death-trap. If anyone compiles this with a compiler that does not support this feature you will get entirely uninitialized memory, without any warnings or error! Stay away from this feature, and if possible add tooling to detect accidental use of this.
????????????????????????????????????????????
"if the compiler is broken you will get broken results, therefore, it's on you to not use this feature" excuse me what
a compiler that supports c23 must compile that properly, and one that doesn't must error -- any other case is a bug in the compiler

2025-12-09 20:02:53 GMT

i never once had an enjoyable experience with x11 forwarding via ssh
it's always so sloooooooow

2025-12-02 17:58:30 GMT

> Now that there is a Steam Deck Native build, is Baldur’s Gate 3 supported on Linux?
> Larian does not provide support for the Linux platform. The Steam Deck Native build is only supported on Steam Deck.
oh fuck off *why* -- i have the game on gog, but i play it on the steam deck
just release the damn binary ffs, everything else is the same, if i had the game on steam i likely could've just force download the depot -- but i don't

2025-11-29 16:26:07 GMT

i think my left monitor is kinda too big
i have two 27" monitors, and looking at anything on the left side of the left monitor, is so uncomfortable, so only the right side is usable
but if i flip the monitor vertically, then it's suddenly too tall... hmmm

2025-11-28 19:12:09 GMT

okay a question, a lot of people say to not have more than 6 arguments to a function for performance reasons, because anything past 6 will be passed on the stack instead of registers
and then say that if you need so many arguments, to pack them in a struct instead -- rust even has that as a lint by default
and, while i do agree that a function with 6+ arguments is unwieldy and hard to read, i don't think i believe the performance thing?
because if your struct is larger than a register, it will be passed via the stack anyway, with a pointer to the stack struct "passed" in as the argument proper
so unless you do something like methodically split it into many structs each less than the size of a register, going with a big struct versus many arguments does nothing, or did i miss something?

2025-11-28 11:19:49 GMT

- 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 21:39:54 GMT

- reply

@nprofile…3gt0 i mean sure but i meant more my brother would need to look at the label of the current brick, figure out the laptop needs 19.5V at 9.23A, set that up with a properly shaped plug
and i can probably get one of those for him, and set it up properly -- i guess my point is more like, this shouldn't need to be a thing, and if i weren't here, he shouldn't need be out of luck and think his only option is to send the whole laptop to a (rather expensive) technician

2025-11-27 21:29:42 GMT

- reply

@nprofile…3gt0 yeah but then i'd need to find one with a gigantic plug like this one *and* know how to set it up properly as to not fry the laptop
which i, personally, would know how to do -- but my bother who's the actual owner of this laptop, wouldn't

2025-11-27 21:23:29 GMT

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

2025-11-27 21:00:29 GMT

TIL a fully qualified atom is valid in portage’s command line
so, it is valid to do $ emerge cat/pkg[foo,-nya]

2025-11-22 01:54:17 GMT

- reply

@nprofile…3gt0 i had a manual build for a while but realized that i don't really care too much? dist kernels automate the install and such
one day i'll go back to a custom config just to (drastically) speed up compile times, but i'll keep using a dist kernel ebuild, just, with my own savedconfig
(now i wonder if the /etc/kernel/config.d snippets overlay savedconfig or not)

2025-11-22 01:03:23 GMT

"adds" kernel config snippet to /etc/kernel/config.d
spends half an hour rebuilding the kernel
reboots
config not set
i put it in /etc/kernel instead...
ughhh

2025-11-19 19:42:48 GMT

- reply

@nprofile…9e6e i'd rather not

2025-11-19 19:22:11 GMT

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

2025-11-19 11:48:24 GMT

rust on cpython, aka, "who needs to be able to bootstrap a system"

2025-11-18 16:15:22 GMT

if i'm reading the posix issue correctly, it might be that now `n>&n`, e.g., `3>&3`, can be used to keep the fd `n` open in the executed utility
if that "self" redirection isn't done, posix shells are free to close or not close open file descriptors not directly passed in as redirections
this makes some things quite easier, and seems to be what most shells are already doing (by mapping redirection to dup2)

2025-11-18 12:13:02 GMT

- reply

@nprofile…3gt0 @nprofile…4zgr @nprofile…0ga3 a modern init without supervision shouldn't be called a modern init
even openrc had supervision for it's init script daemons for a while, not great supervision, i'm working on that, but it's something

2025-11-06 22:51:30 GMT

behold, the result of me and wifey discussing what counts as bread versus what counts as cake
https://social.vlhl.dev/media/b5a99f0c524347379d181496742ad07833471095a4729b0bcf89748258f41546.png

2025-11-05 23:28:18 GMT

by all means, ig, we should go with /etc/openrc over /etc/rc
just, guh, i don't like /etc/openrc at all, and we use rc-* basically everywhere, binary name prefixes, rc.conf, variable names (rc_* in rc.conf, RC_* in the environment)
internal conflict
my brain sucks, i hate it sometimes

2025-11-04 17:44:48 GMT

so many "isekai" / "transmigration" stories would be so much better if they didn't have this stupid plot point

2025-11-03 20:53:54 GMT

"can we please leave politics out of *punk metal*?"
??????????????????????????????????
what
punk is a political movement, since forever
just, what
same energy as "can we leave politics out of FOSS"

2025-10-20 00:34:46 GMT

- reply

@nprofile…3gt0 yeah, though more specifically, debian's https://wiki.debian.org/LSBInitScripts/
which has dependencies and helpers, but instead of building the infrastructure around the init script, it does dependencies by scanning the script for specifically-formatted comments, and doesn't guarantee any commands, but say "compliant scripts must do them!" -- so script writers are still expected to always implement `start` and such
plus `log_*_msg` is awful compared to einfo :/

2025-10-20 00:28:27 GMT

openrc is like lsb if it were good
it's kinda funny how both projects had somewhat the same idea, but one managed to do it so much better

2025-10-19 19:17:24 GMT

- reply

@nprofile…9e6e
static inline bool streq(...) {
return strcmp(...) == 0;
}
any compiler worth their salt will optimize this as well

2025-10-19 17:50:42 GMT

- reply

@nprofile…9e6e for strcmp, i like to define a streq that returns bool, because that is more clearly what i want to do
i don’t think != 0 is that much better tho, (error = foo()) and foo() != OK are clear, “if error” and “if not okay”, == 0 is… “not zero?” (same with -1)
i use != 0 for libc things because well, libc, but still, better than !foo(), worst than error = foo() or foo() != OK

2025-10-19 17:03:49 GMT

it’s kinda funny how
if (get_foo(&foo))
print("failed to get foo");
looks wrong, it feels like if (get_$thing) should be true on success
but, if i do
if ((error = get_foo(&foo)))
printf("failed to get foo: %s" str(error));
now it looks correct, now it reads as “if there’s an error getting foo”, the same apply to
if (get_foo(&foo) != FOO_OK)
all of those check the same thing, not-zero is error, but code readability changes drastically, imo at least

2025-10-19 15:30:25 GMT

i got an "openrc appreciation" mail thanking me for working on it
i kinda melted a bit tbh ~w~

2025-10-18 15:51:46 GMT

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

2025-10-17 12:00:32 GMT

this is imo one of the few actually good "new" features youtube added
videos can be uploaded by two channels at once, and it indicates contributions in videos a lot better
https://social.vlhl.dev/media/4a79cdc28169080ba29c3e758c92d46c7845742c42f3246e26682c21dd1ce19e.png

2025-10-12 18:15:48 GMT

- 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 GMT

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 GMT

- 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 GMT

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 GMT

- 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 GMT

- 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 GMT

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 GMT

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 GMT

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 GMT

- 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 GMT

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 GMT

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 GMT

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

2025-09-10 01:02:55 GMT

- 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 GMT

- 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 GMT

`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 GMT

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

2025-09-06 22:20:55 GMT

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 GMT

- 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 GMT

- 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 GMT

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 GMT

- 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 GMT

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 GMT

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 GMT

- 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 GMT

usr-merge but it’s /usr -> . instead

2025-08-20 01:23:00 GMT

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