<bslsk05>
'AT&T Archives: The UNIX Operating System' by AT&T Tech Channel
Lucretia has quit [Read error: Connection reset by peer]
alexander has quit [Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in]
karenw has joined #osdev
k4m1 has quit [Quit: %bye%]
alexander has joined #osdev
HD36079 has quit [Quit: Leaving]
<Ermine>
3d graphics in unix?
<heat>
desktop UNIX is closer to being a reality than desktop Linux
<heat>
don't @ me
<zid>
osx is desktop unix
<zid>
fite me
<heat>
worse. desktop freebsd.
<n00by>
openbasedBSD
<nikolar>
certified posix
n00by has quit [Ping timeout: 252 seconds]
<pog>
does the workstation era count as desktop unix
<nikolar>
yes
<pog>
or does it have to be home users
<pog>
i don't think many contemporary home users had NeXT or Sun boxes
<zid>
sgi
<pog>
certainly not sgi
<zid>
ironically, onyx
<zid>
when onyx/onyx btw heat
<pog>
gnu/onyx
<zid>
gnu-onyx-onyx-gcc
<Ermine>
those wayland people took the Glorious X11 UNIX Desktop from us
<Ermine>
or rather, X11 people took away X10
<Ermine>
and X* people took away NeWS
n00by has joined #osdev
n00by has quit [Client Quit]
k4m1 has joined #osdev
k4m1 has quit [Client Quit]
<heat>
bad news guys the WOKE MOB just cancelled our linux X11 Xorg desktop
<heat>
i guess the wait was for nothing
k4m1 has joined #osdev
<GeDaMo>
I'm still waiting for X12 :|
<zid>
I should make my own x12
<zid>
is valve going to fork wayland yet
<GeDaMo>
zidWindows :P
<zid>
Z12
<heat>
and the Z stands for zombocom
* kof673
<inserts beos as quasi-unix-ish next to next/sun/sgi boxen>
<zid>
honestly all we need for linux desktop is for.. someone to make one
<zid>
rather than 800 people making pieces of 800 different ones
<zid>
like, steam big picture on stim dick
<zid>
until there's a user32.so.6 linux sucks
<pog>
i prefer the fractious anarchy
netbsduser`` has quit [Read error: Connection reset by peer]
<pog>
it's either that or fucking windows
<nikolar>
and we don't want windows
<nikolar>
we already have one and it sucks
n00by has joined #osdev
<kof673>
i'm not being negative and don't think these things are that big of a deal but...people were saying all this 20 years ago :D maybe even justifications for gnome and kde
<nikolar>
i mean gnome sucks too
<nikolar>
imagine if all of linux desktop was just gnome
<nikolar>
ugh
<kof673>
i mean i get by fine with twm...c89 i could make do on dos lol
<zid>
at least C++ being un-linkable means kde tends to glob together slightly I guess :P
<n00by>
(the glib haters club includes me too :P)
<kof673>
but that's mainly because i am only coding perhaps, and use another machine for web when needed, etc.
<heat>
glib is based
<n00by>
how so
<heat>
"GLib provides advanced data structures, such as memory chunks, doubly and singly linked lists, hash tables, dynamic strings and string utilities, such as a lexical scanner, string chunks (groups of strings), dynamic arrays, balanced binary trees, N-ary trees, quarks (a two-way association of a string and a unique integer identifier), keyed data lists, relations, and tuples"
<heat>
"GLib implements functions that provide threads, thread programming and related facilities such as primitive variable access, mutexes, asynchronous queues, secure memory pools, message passing and logging, hook functions (callback registering) and timers. GLib also includes message passing facilities such as byte order conversion and I/O channels."
<heat>
the standard library C never had and never will have
<n00by>
aight
<n00by>
kek
k4m1 has quit [Read error: Connection reset by peer]
k4m1 has joined #osdev
<n00by>
kof673: i still run a fork of twm
<kof673>
if (tabbed_wm) { ... } else { if (tabbed_xterm) { ... } else { screen_or_tmux_or_vim_tabs(); } i'm not using twm right now, but it wouldn't make a huge difference lol
<kof673>
i just used it on a laptop a long time ago didn't bother to set anything else up lol
<n00by>
i assume most of you experts run tiling window managers :P
<kof673>
i never tried, didn't think it would be a huge gain
<heat>
i'm on GNOME
<kof673>
same reason i never learned dvorak, then when i go to another machine, what happens?
<GeDaMo>
I used ratpoison a long time ago when my machine was a little underpowered
<pog>
i've nver used a tiling window manager and i don't think it has anything to do with expertise
<nikolar>
kof673: using tiling wms isn't exactly comparable to using dvorak lol
<heat>
totes ricing them UNIXes takes away a lot of time for... useful things
<kof673>
nikolar, i just meant in some situations i probably have to use non-tiling
<nikolar>
you can still move windows around if you're used to tiling
<nikolar>
you can't type if you're used to dvorak lol
<pog>
yeh, like i'm not opposed to learning new things or changing my workflow; i do a lot of that, but a floating window manager has never been a hindrance to me
<zid>
isn't a tiling wm just a subset of a normal wm
<nikolar>
zid: it just automatically tiles windows when you spawn them
<zid>
like, windows has fast hotkeys for positioning windows at the edges split with other windows
<kof673>
i'm not sure, i think tiling was even old old stuff and then sort of revisited
<nikolar>
instead of you having to tell it to snap somewhere or whatever
<GeDaMo>
Windows 1.0 was tiled if I remember correctly :P
<pog>
most floating window managers have some kind of feature to sort of do tiling
<zid>
I could write something in 10 mins to just resize other windows
<zid>
forcibly
<zid>
and I'd still be able to move shit
<zid>
so then isn't a normal wm just a superset
<pog>
like i can press a few hotkeys and windows will tile up what i pick
<nikolar>
zid: you can move shit in tiling wms too
<nikolar>
you just have to ask for it, because it gets snapped automatically
<zid>
anyway, I never understood the appeal of tiling wms
<zid>
cus.. if I want to see multiple windows, I almost certainly want to try to preserve as much screen real-estate as possible, by carefully overlapping the windows
<zid>
so that the bit I wanna see is visible
<n00by>
pog: i just meant in a sense that it has somewhat greater efficiency; dwm family of window managers :)
<zid>
nikolar's seen me wrestle with editor + compiler + docs + discord etc :P
<nikolar>
except that when you click on a window zid, all your careful overlapping goes to waste
<nikolar>
because it gets brought up front
<zid>
some WMs let you pin properly
<zid>
to glue windows together
<zid>
windows just doesn't
<zid>
eve online, for example :P
<n00by>
GeDaMo: ratpoison is such a based name lol
<nikolar>
til eve online is better windows than windows
<GeDaMo>
n00by: ratpoison == no mouse :P
<n00by>
LOLLL
<nikolar>
ooooh
<zid>
oh it's a PUNE
jcea has joined #osdev
<pog>
n00by: efficiency is subjective when it's different users with different habits imo
<n00by>
certainly, yup
Lucretia has joined #osdev
azonenberg has joined #osdev
<azonenberg>
Hey so i've been sitting on some crazy ideas for a while and was wondering if anybody had tried this
<GeDaMo>
Tried sitting on things? Yes :P
<azonenberg>
(sorry got pulled away for a sec)
<azonenberg>
so on ARM (specifically Cortex-A9), is there anything that would stop you from setting the interrupt vector table to a bogus/unmapped address
<azonenberg>
disabling interrupts
<azonenberg>
then exiting kernel mode to user mode without re-enabling them?
<azonenberg>
if so, would you effectively be locked in userspace permanently until the next reboot?
<azonenberg>
(which could only happen by external power cycle or JTAG or something as you have no access to the SFRs to initiate a reset)
<zid>
doesn't arm triplefault or anything?
<zid>
that's both a blessing and a curse I suppose
<azonenberg>
zid: well i mean, if your fault handler is a nonexistent address
<zid>
yesand? that's how you triplefault x86
<azonenberg>
what happens if you segfault?
<zid>
fault, page fault handler faults
<azonenberg>
does it just lock up the core?
<zid>
because there's no good address there
<clever>
i have had something similar happen on the rpi, but it was mapped
<azonenberg>
So in this case i am interested in doing it on purpose to create a sort of sandboxc
<clever>
during bootup in linux, it sets the address of the vector table, but doesnt populate it immediately
<azonenberg>
i want to have the core given access to a very specific access of resources
<zid>
I mean, even if it doesn't work
<clever>
it then tries to use load-exclusive to manage a mutex while doing printk
<zid>
you can just set them all to iret
<zid>
and now irqs and syscalls etc do nothing
<azonenberg>
and be very sure that no matter what happens, it can't get out
<zid>
you could even unmap 99% of the kernel before you left
<clever>
and because i didnt set a special flag, load-exclusive is an illegal opcode
<clever>
so it jumps to undefined code, and crashes, then jumps to undefined code, and crashes, forever
<azonenberg>
so, for background
<zid>
leave a single bounce page that only has enough code left in it to wipe some memory to 0 then iret to userspace after
<azonenberg>
any of you familiar with my antikernel project?
<azonenberg>
it was my phd thesis work back in 2015 (and has mostly gone stagnant since), basically an OS without a kernel
<azonenberg>
you add some minimal features to hardware for threading and IPC etc then run userspace on bare metal
<azonenberg>
the problem is, it needed custom hardware
<azonenberg>
so one of my long standing interests has been, how close can you get to this on existing silicon
<azonenberg>
i had a cursed idea involving the xilinx zynq fpga soc
<azonenberg>
basically you do what i described to run one usermode thread on each core of the A9 in a sandbox that can't access any peripherals or anything but a mailbox interface to the FPGA over axi
<nikolar>
you can just map a shared page for ipc
<nikolar>
no special hardware neede
<azonenberg>
nikolar: there's more than that
<nikolar>
and you can have trivial interrupts that basically just iret
<azonenberg>
you need to have things like hardware memory allocation and threading without ever re-entering kernel mode
<nikolar>
instead of trying the bogus interrupt table trick
<nikolar>
well if you assume you have other cores present, you can have them take care of that
<nikolar>
you only need to reload the memory map on the interrupt or something
<azonenberg>
except the entire premise of the platform is that you dont trust ring0, its a violation of least required privilege
sortiecat has joined #osdev
<azonenberg>
nothing should have access to that level of power
<azonenberg>
anyway, so my concept was to build a coresight bus master in RTL on the FPGA
<nikolar>
well yeah, then you need custom hardware
<azonenberg>
nikolar: yes but the unanswered question is, with silicon that's already on the shelf
<nikolar>
because you're going completely against the last 50 years or so of hardware design
<azonenberg>
how close to this ideal can you get by abusing it in clever ways
<azonenberg>
that's the research project
<zid>
if arm can't triplefault you're already finished though aren't you
<azonenberg>
as a followup to the original antikernel which absolutely needed custom silicon
<azonenberg>
anyway so the idea is, coresight bus master in FPGA that would talk to the debug APB
<azonenberg>
every ms or so, halt the A9, serialize all the registers out to a tiny on-die SRAM
<azonenberg>
load a new context, then resume
<zid>
just don't set up syscalls properly, and don't preempt sched in your irq handlers, gg done
<azonenberg>
at this point you've got preemptive multitasking without ever entering kernel mode (in theory)
<azonenberg>
i can't see a reason it wouldn't work but have not had the time to build it
<azonenberg>
and test
<GeDaMo>
This is vaguely reminding me of something but I can't come up with any keywords to search for it :/
<bslsk05>
eprint.iacr.org: Antikernel: A Decentralized Secure Hardware-Software Operating System Architecture
<GeDaMo>
No, it was something else
n00by has quit [Quit: leaving]
<nikolar>
zid: i am pretty sure arm does have a triple fault equivalent though
<azonenberg>
anyway, the details are irrelevant the main thing that would have ruined my plan is if jumping from kernel mode to userspace always re-enabled interrupts (but the bogus vector table is insurance against that leading to a useful attack vector)
<bslsk05>
Ruddle/Fomos - Experimental OS, built with rust (25 forks/964 stargazers/MIT)
<zid>
nikolar: have you install idiot handlers that just ignore all faults I guess then
<zid>
s/you/to
<nikolar>
yeah what i suggested basically
<nikolar>
trivial handlers that just iret
<nikolar>
or the equivalent on arm
<zid>
you suggested that did you
<zid>
[13:10] <zid> you can just set them all to iret
<nikolar>
well, i didn't scroll up :P
<nikolar>
> OS development is extremely hard, Rust makes it more bearable.
<nikolar>
LOL
<zid>
you're supposed to be tesla, not bell
<zid>
alexander graham "forgot to scroll up" bell
<nikolar>
lol
<zid>
invented the telephone btw
<nikolar>
yes i know
<zid>
I know you know
<zid>
that was not the point of that statement
<nikolar>
lol
<heat>
nikolar: that's a bit of a fact
<heat>
but not in the way the author probably meant
<nikolar>
kek
<heat>
author just wantz to use dem cargo cratezzzzzzzz
<nikolar>
so basically he wants to play lego
<nikolar>
and not actually write an os :P
<zid>
duplo
<zid>
lego is C
<nikolar>
kek
<heat>
yeah lego is unsafe
<heat>
you can accidentally eat them
<zid>
'accidentally'
<zid>
you can't tell me what to do
<heat>
accidentally, on purpose
<zid>
it's my lego and I think it's a tastey
<heat>
what's the difference?
<heat>
you get a tasty lego down your stomach
<nikolar>
indeed
karenw has quit [Ping timeout: 252 seconds]
karenw has joined #osdev
Yoofie6464463823 has quit [Read error: Connection reset by peer]
Yoofie6464463823 has joined #osdev
<kof673>
there were those mindstorm robot things but no idea if there are emulators etc. i think relatively, as far as lego goes, they were sort of pricey, but i am not up to date on lego, and this was a long time ago
<kof673>
i mean you can dev on lego, its not a metaphor
sham1 has joined #osdev
k4m1 has quit [Ping timeout: 244 seconds]
k4m1 has joined #osdev
Left_Turn has joined #osdev
k4m1 has quit [Ping timeout: 260 seconds]
k4m1 has joined #osdev
k4m1 has quit [Ping timeout: 260 seconds]
k4m1 has joined #osdev
k4m1 has quit [Ping timeout: 260 seconds]
<Ermine>
glib is blursed
<nikolar>
glib is
<zid>
blursed is correct
karenw has quit [Quit: Deep into that darkness peering...]
goliath has quit [Quit: SIGSEGV]
k4m1 has joined #osdev
tronexte has quit [Ping timeout: 260 seconds]
tronexte has joined #osdev
vdamewood has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
k4m1 has quit [Ping timeout: 272 seconds]
Turn_Left has joined #osdev
Turn_Left has quit [Remote host closed the connection]
<bslsk05>
wccftech.com: AMD Shares Gain ~7% As Analyst Believes Firm Can Close Gap With NVIDIA
<zid>
nvidia just has to release the new 1080 ti
<heat>
why do stocks always go up *before* I buy them
<zid>
if amd ever threatens them
<heat>
that sounds intel-esque
<zid>
intel fucked it 10 years ago
<zid>
and amd caught up
<zid>
nvidia are ahead here, just releasing garbage because of policy
<GeDaMo>
I thought Nvidia were all about the AI now
<zid>
ai policy, mainly :P
<zid>
nvidia's stock is through the roof cus they're selling most dies as AI ACCELERATORS
<zid>
but it's not like their actual tech is lacking
<heat>
what if I got an Arc A770
<heat>
what sounds like a terrible idea
<zid>
the 4xxx and 5xxx cards are just.. pretty anti-consumer, they wanted their 10% benchmark figures so they upped the power and voltage by 20% each gen
<zid>
I'd use an A770
<heat>
GeDaMo: all about the datacenter
<heat>
they sell shovels.
<heat>
particularly expensive shovels, but they do sell them
<zid>
nvidia cares so little about gaming segment it's kind of funny
<heat>
it's a very small percentage of their revenue
<zid>
shit overpriced cards and a lot of the 3rd parties pulled out, and their stock is up 8493943% anyway cus AI
<bslsk05>
www.tomshardware.com: Intel Arc B580 review: The new $249 GPU champion has arrived | Tom's Hardware
<zid>
heat: Yup, pretty much exactly what I expected
<zid>
and yea, I'm still sat here fingers crossed that intel don't axe their gpu efforts
<zid>
they're finally hitting their stride
<zid>
it's probably losing money still, but it's finally decent enough to start to compete
<GeDaMo>
I'm doing OK with integrated graphics so far :P
<zid>
but how will you play... roadcraft
<GeDaMo>
I don't know what that is :|
<zid>
a video game about driving heavy construction equipment
<heat>
crafting roads
Terlisimo has quit [Ping timeout: 252 seconds]
<GeDaMo>
Ah
<zid>
surprising little actual road crafting
<zid>
it's honestly more of a gimmick :(
<zid>
Lots of filling potholes with sand and saying "good enough" though!
<GeDaMo>
Tea breaks? :P
Terlisimo has joined #osdev
<nikolar>
zid: wrong
<nikolar>
a video game about filling holes with sand
<zid>
that's what I said
<nikolar>
i hadn't gotten to that part yet :(
<zid>
Alexander Graham Nikolar strikes again
<nikolar>
lol
n00by has joined #osdev
Turn_Left has joined #osdev
<Ermine>
heat: why 5080 and not 70 or 60?
<Ermine>
They are also power hungry as hell afaik
Left_Turn has quit [Ping timeout: 252 seconds]
<geist>
i'm still using my A750 for living room gaming
<geist>
it's pretty okay
<heat>
Ermine: idk i'm seeing the benchmarks and getting spooked
<nikolar>
Ermine: as zid said, they are all overclocked to shit
<nikolar>
so yeah, power hungry
goliath has joined #osdev
n00by has quit [Quit: leaving]
<geist>
nice having a space heater when it's cold
<heat>
the VAX or the A750?
<heat>
or both?
<geist>
well not the A750, the big VAX uses more power (though surprisingly low, about 380W)
<geist>
i use the ultrasparc for warming my feet, it pulls a solid 200W and is under the table
<heat>
time to attach the A750 to the VAX via serial?
<heat>
PCIe-to-serial converter
<heat>
just sayin, it's a good idea
<geist>
A750 to tape drive
<geist>
write out a frame of textures and draw commands, move between computers, load it up
<geist>
render frame
<nikolar>
heat: hey vax can do better than serial
<nikolar>
pcie to qbus let's go
<geist>
but yeah actually qbus is fairly sophisticated, and it's a simpler version than unibus
<nikolar>
do you still have to hardware the interrupt request passthrough thingy :P
<geist>
there's some sort of termination thing on it yeah
<nikolar>
i know that you had to actually hardwire that line on pdp-11
<geist>
but actually it's more silly than that, it's because there's doublewide QBUS on both the PDP11/53 and the big vax, there's some rules about half width cards and which side they can go in
<geist>
and if you have to terminate
<nikolar>
i wonder if/when they made it automatic
<geist>
so it's actually kinda complicated because the 'right side' of the slots are not always just another qbus
steelswords94361 has quit [Quit: Ping timeout (120 seconds)]
<nikolar>
yeah and the userspace side is still totally proprietary lol
<heat>
the 9070 XT is a little too expensive for my taste
<geist>
i think that's a difference in unibus and qbus too, the request lines are chained in qbus, but the vector itself is another cycle the card provides
<nikolar>
heat: i got a 7800xt iirc
<nikolar>
wait let me check lol
<nikolar>
> Radeon RX 7800 XT
<nikolar>
yea
<nikolar>
> [Radeon RX 7700 XT / 7800 XT]
<nikolar>
lol thanks lspci
<nikolar>
very helpful
<heat>
christ
<nikolar>
why
<heat>
i guess it's the same PCI device ID but they differentiate in IP blocks or something
<nikolar>
presumably
<geist>
but also dont forget qbus only goes up to 22 bits of addressing, and is 16 bit wide, so when using it on this microvax it can only address the first 4MB of ram
<geist>
so there's that whole bounce buffer stuff and whatnot
<nikolar>
oh silly
<nikolar>
did they not extend it at all for the vaxen
<heat>
BOUNCE BUFFERS!!!
<geist>
remember vaxen started off with unibus, which is a full 32bit
<geist>
qbus is a cheaper, cost reduced smaller thing, so it was part of the tradeoff when they made cheaper microvaxes
<nikolar>
how much of a saving was it really
<GeDaMo>
nikolar: apparently the 7800 has more memory
<nikolar>
it's got 16gb iirc
<nikolar>
mine does i mean
<GeDaMo>
"The AMD Radeon RX 7700 XT features 12GB GDDR6 with memory speeds of 18GBps. For comparison, the AMD Radeon RX 7800 XT features 16GB GDDR6, with slightly faster speeds of 19.5GBps."
<geist>
a lot, you could get one of these microvaxes and roll it around or put it on your desk, vs it being a large mainframe
<nikolar>
GeDaMo: i think there's a difference in compute units or whatever amd calls them
<geist>
a qbus card is like 12 inches wide too, relatively tiny
<heat>
CUDA CORES
<nikolar>
i think amd calls them compute units, yeah
<heat>
yeah but i call them cuda cores
<nikolar>
they are not quite equivalent iicr
<nikolar>
i think amd groups a lot more or something
<heat>
also one gpu can run CUDA and the other can't :p
<nikolar>
so you get fewer CUs for equivalent cards
<geist>
oh interesting, actually unibus only has like 18 address lines
<nikolar>
wasn't there some work (in progress) to make cuda work on amd cards too
<nikolar>
geist: so even fewer lol
<geist>
was from like 1969. may be massbus i'm thining about that's bigger
<geist>
well no looks like massbus is more for storage
<geist>
anyway, there were lots of busses on DEC things but they were kinda common among families, so that was nice
<GeDaMo>
nikolar: I think that was dropped by AMD
<nikolar>
what was
<nikolar>
GeDaMo: so, pcie to qbus when?
<geist>
i think AMD has their own thing that can deal with cuda now right?
<bslsk05>
www.theregister.com: AMD lawyers claw back CUDA compatibility layer ZLUDA • The Register
<geist>
looks like there's ROCm and SCALE
<nikolar>
i wonder what the throuput of vaxen {uni,q}bus was
GeDaMo has quit [Quit: 0wt 0f v0w3ls.]
Left_Turn has joined #osdev
Turn_Left has quit [Ping timeout: 276 seconds]
tigerbrother has quit [Quit: Ping timeout (120 seconds)]
tigerbrother has joined #osdev
goliath has quit [Quit: SIGSEGV]
karenw has joined #osdev
<geist>
i'm guessing low MB/sec, which would have been pretty good in the 70s
<geist>
i remember benchmarking a 1990 era 386 with a 40MB drive and it was in the hundreds of KB/sec transfer rate, for example
<geist>
to sort of put it in perspective
Turn_Left has joined #osdev
averymt has joined #osdev
sprock has quit [Ping timeout: 244 seconds]
Left_Turn has quit [Ping timeout: 260 seconds]
Yoofie6464463823 has quit [Ping timeout: 260 seconds]
<nikolar>
that's the hdd read speed then, no?
<nikolar>
not the bus throughput or whatever
<heat>
i mean the ISA bug was real fuckin slow
<heat>
s/bug/bus
foxsecu has quit [Ping timeout: 252 seconds]
Lucretia has quit [Remote host closed the connection]
<nikolar>
i mean, sure
<geist>
yeah a bit of both
<geist>
like getting 300-400KB/sec off your HD on an 8 bit ISA bus is not half bad
<nikolar>
iirc, isa had to go through the cpu right
<geist>
you're getting like 50% bus utilization
sprock has joined #osdev
<geist>
yep. so my point is the busses in the 70s that VAXen were using and whatnot were actually a lot closer to PCI
<geist>
in that everything could bus master, etc
<nikolar>
yeah, very fancy
<nikolar>
(thus the bus grant line thing i mentioned earlier)
<nikolar>
though it wasn't autoconfig and you had to hardwire it :P
<nikolar>
at first at least
sprock has quit [Remote host closed the connection]
<geist>
but OTOH unibus had few address lines, etc so even at the time of the 11/780 it was already too small
<geist>
since it only had like 18 bits of addressing (256Kwords)
<geist>
it had originated on PDP-11 which only ever got up to 22 bits of addressing at the max
<geist>
but started off 16, then 18, then 22 over its life
<geist>
so to a certain extent the later and cheaper QBUS was more powerful because it at least had 22 address lines, though it appears it was multiplexd with the data lines and thus required more cycles
<geist>
i dunno what the later, bigger vaxen did for busses
<geist>
looks ike later vaxen (6000, 7000, 10000 series) had a new bus called XMI
<geist>
at least for IO stuff. i think a lot of these vaxen had various internal cpu to cpu and cpu to memory busses that were a bit more proprietary
<geist>
or at least specific to their family
Left_Turn has joined #osdev
<nikolar>
did you need to care about them, as a programmer