duncan: a?

kerneldove_: those times are not fixed

key ketas man what's up

hey*

black2 dude

what? ketas

What!?

mej

mek

why features are added between b3/b4m tho?

what is a "beta" anyway then

ketas: it is normal that not all features planned for 15.0 are ready for BETA1, so they might be added later in the release cycle. after RC, the threshold for adding features is much higher unless it's one of the small number of specific issues identified beforehand

but this is strange as hell

redefining those meanings are weird

if i mentally explain myself it's a something before a real thing this of course works

must be the most Fallout thing I have seen for a long time. Lucas Deeco ST3220 youtube.com/watch?v=WSnk5ZbrWAo

fun eh

actual terminal

never seen one that is a plasma based touch screen

does it mean something wrong if a server hasn't had any additions to /var/log/messages for a couple days? everything seems to work but i'm used to having tons of ssh key exchange errors from brute force bots, limiting icmp unreach response messages, etc

reminds me what i did as younger one

these would be seriously cool for a home automation project

?

oh

the times the print screen actually prints screen on dot matrix

:p

i have 2 old elotouch devices

but i think the time moved on

have a vt520 on my desk to my server in the basement

so for automation status screens you have more fancy options, oled, lcd, eink

I guarantee these are still in use. Sitting on CNC machines, laithes etc

oh of course

tech that just works is so hot

I know they are outdated, but as a massive fallout fan, they strike a chord

i'm surprised how well freebsd 'just works' since it's C/C++ and not rust

question of testing I think

and just because something is rust - its not necessarily fault free. Its fault free from memory allocation issues.

I think the new uutils in Ubuntu is a really good example of that

I really don't know what the h... Cannonical was thinking

oh - and this is not a critique of rust btw. I like the language

but if it isn't broken - don't fix it

kerneldove_: i thought you like well oxygenated iron?

ya i love it, but that doesn't stop me from complimenting battle tested C

what might be cool is adding a bunch of freebsd tests to the harness in rust. a way, outside of the base OS, to let devs start using rust

i wonder how to make ws2812b driver in kernel

if we where to write uutils today - rust would be the tool

i'm curious if i could make it

ws2812b mcu looks cool

kinda neat idea, a driver for an LED strip that has a great form factor for putting on any 1u+ server so fbsd admins could easily configure extended physical readouts of status

something that would be cross hardware platform (dell, hp, supermicro whitebox, whatever)

it's more like shift register in it iirc

most servers have a front usb or 2, could plug it in, affix it with velcro or some other adhesive, and be able to easily see OS-level status readout just by looking at a rack

servers are great for hw level status readout but something integrated into OS level would be cool

like something i lost when i switched from hardware raid to hba mode + zfs is being able to see neat amber light on failed drive. well what if the led indicated drive error by integrating with smarttools?

kerneldove_: lots of dell servers have a front LCD display

for usb you have other options

ya but like i said that's for hardware shit

you can customize it with programming I believe

even if so, how cross hadware compatible is that?

across what and what?

i'm thinking something way simpler, a few LED panel that's like a half inch by a couple inches in size

any brand of hardware

that LCD is like non-removable, only available in select dell server models

so a usb screen?

do we have them already

or just a buzzer or speaker coupled with tts engines onboard

(installed in the OS)

that's far more than enough for a status indication

those led strips are very bright. I think this could easily turn into a bit of a disco in the server room. Most of them are designed for lighting

unless you can get some more low output strips

nah speaker/tts would suck, needs to be visual. imagine trying to hear which server it is in a noisy DC

i think they sold gamer status displays

but you can make it

using i2c screen, maybe strips

uc and usb

speaker/tts would be fun for a workstation tho

you could turn brightness down in leds

diffuse them

they also consume a ton of current so you need some limits to not blow up

unsure how many dc's even need statuses on server itself

this is not a hollywood movie

tho some display somewhere could be useful

in industry they are hmi's

well it's not like it'd replace monitoring systems, but rather tying those into the physical world would be cool

being able to make a led flash on the server and on the disk is really nice, if you need to make a tech in a remote server room find the correct device though

there's a reason dell servers have blue light amber light

yep

put single rgb there

'just verify the serial number label on the server bro' <- hasn't relied on remote hands a whole lot

:p

yea that's what i mean, a very few function led

no need for a whole display

do any normal mobo have a gpio btw?

why do i recall those fancy hetzner servers

which look like mine

who needs case eh

ya if your DC has air filtration and strict rules on no liquids could be worth skipping the case

i wonder tho what about like, paperclip falling out of shirt pocket of a tech while bending over or smth and shorting a server?

unlikely

yeah it was total fun to lool those photos

just bare minimum hw

they did order custom mobos tho

url to pic?

there are so many

hetznet dedicated is right query

but

on other photos i see basic atx power contol bridge pcb's etc

looks good. i woulda been a customer but too bad they demanded a photo of my gov id stupid weirdos

secret birds

ya i don't need to compromise my opsec just so nerds can jerk it to my pic

found other hosts with just as good prices just not the big name

an actual enclosure must also give better protection against EMP

FYI, hetzner dedicated bottom tier "servers" are actually just trash tier desktop hardware

I've had to debug a misbehaving service running on one, and it turned out the hardware was somehow malfunctioning - but in a way that wasn't trivially diagnosed. hetzner turned out to be incredibly unhelpful in responding to the issue after their first level engineers failed to diagnose the fault

of course they are

ecc ram tho?

not if it's Intel, they don't support that on desktop skus

it wasn't Intel

the great war of ecc

btw

rescue - usr.bin/id/id.c:149:16: error: use of undeclared identifier 'Aflag'

wtf?

there was id.c change indeed

i'll see what ml says :p

oh yeah i see the error

that's because desdotdev forgot to test without USE_BSM_AUDIT :p

ketas: instead of blaming people for a mistake, it would be more useful to submit a patch

i'm looking at code but can't come up with good idea

ivy: less drama pls

yeah i thinki found right patch but

i'll let des look at it?

#ifdef USE_BSM_AUDIT\ncombo += Aflag;

Koston: it is not "drama" to suggest a way of engaging with bug reports that doesn't hinge on blaming people

and remove from general

should i file that as bug too :p

or it's so little

and nevermind current

it does break customized build tho

just poke des

ketas: it is probably redundant to post on the mailing list and also file a bug, but you could do if you wanted

well it's not needed indeed

if the preference is to avoid calling attention to it, perhaps you don't want to file a bug and announce it on mailing lists. sigh

des might fix it before i do that

i mean i guess there was extra eyeball

there's nothing wrong with "calling attention to it". i assume you're deliberating misrepresenting what i just said to make some sort of point?

and tester

it's a bug, report a bug, you don't need to call out a specific person to do that

ivy: ok drama queen

are there drama kings too

Koston: how come you know that ivy is a queen?

maybe princess

like is that name so feminine as to suggest a user's gender?

like ivy could be ivy league, poison ivy and stuff, none of which is particularly gender-specific

:p

was I being delusional?

no

ever had that issue?

oh god it needs another change

black2: a simple solution to this is to simply not call people "bro", "dude", etc. if you don't know their gender, then you never have to worry about this

drama monarch

hah

some song had line he was a she

lola?

Was there a change in bridges between 14.1 and 14.3 where it won't pass traffic unless it's up?

if ((cflag || Aflag || Mflag) && argc > 0) usage();

how to take aflag out :)

if ((clfag || \n #ifdef WHATEVER Aflag || #endif || MFLAG) ...

you can do THAT?

ifdef is parsed before?

maybe

they are, but it doesn't matter in this case, becuase you can put a newline anywhere you want in a C expression

but it looks like how to make c code unreadable

you could write if \n ((cflag \n || \n Aflag \n || \n ... if you wanted (although you obviously wouldn't do that)

yes, this isn't a particular nice construct, but for a one-off case it's fine

combo = Aflag +

yeah ifdef here looks confusing

even if working

haven't even seen it

ketas: the alternative would be to always define Aflag, even if it's always zero when #ifdef WHATEVER isn't enabled

yeah

maybe it's nice and optimized now

?

unless compiler takes it out

id(1) does not need to be optimised, that shouldn't be the primary concern when deciding how to do this

it's extra bytes in rescue :p

but yeah

because machines have 640kb memory

ketas: it's not extra bytes, because Aflag is an auto (stack) variable

meaning?

at most it might be an extra instruction to set it to false, but that is not something that realistically anyone cares about

ah, it wasn't a change in 14.1-14.3, it was a change in Bastille. I was previously using jib to add epair's which made sure the bridge was up. Current bastille just uses ifconfig to do it, so no handy up being automatically added.

(rescue is already ~20MB)

10744180

bytes

in 16

tho

i may have excluded stuff?

main, amd64: -r-xr-xr-x 149 root wheel 19958328 Oct 30 19:52 /rescue/rescue*

16787576

armv7 tho

14174088 in 13 amd64

but the exact size doesn't matter, the point is, the size of rescue is irrelevant to how you fix this issue in id

no one cares about 4-8 bytes being added to rescue

yeah

ecc win, amd win, nvme win

Koston: funny hetzfault

but yeah

that's entry level hosting

price is matching too

funnily it's still six figure sum of hw even on tiny photo crop

because hw costs apparently

ok i closed that and should go to home or so

winter mid street phone c coding

ok i'll sleep and if noone has fixed or moaned tomorrow i like try it on my own

no

i fixed it

nice

:p

ketas: your PR title says w(1), but the patch is for id(1)?

also, your patch appears to include the entire file, which is not desirable for bugzilla patches since it makes it more difficult to see what actually changed

i have too many w's in my brain

somebody asked for fulls :/

when is it needed

that is only when you submit a patch to phabricator, not bugzilla

and you should be using git arc to submit phabricator patches, so you never need to create a patch like that manually

was phab for committers only?

no

ye, git arc is your friend there:)

but, i don't recommend non-committers use phabricator since there's no workflow to get patches committed

i mean i gave a patch and receiver couldn't see where it applies to

or how

once

meh

fun in w continues

and in gpart

xo related

guys, guys. have you finished your IPv6 migrations yet? for the No NAT November

the problem is not that I don't have working v6, there are to many others don't having v6

Koston: i've got v6 but still use v4

best I can do some vlans with v6 only and dns64 + 464xlat

v6 is great but if your ISP doesn't open the firewall for individual addresses, its greatest power is being wasted.

ivy: so the shorter patch in bugs.freebsd.org/bugzilla/show_bug.cgi?id=283427 raised issues like can't see where it applies by glance, i assumed people would just apply it and check where it applied... maybe midpoint context length is required, anyway how do i know how people work

bunch of, is this implenting a function, the code already has a function

hardcoded 2 runs copy is wtf but i got that (bad) idea from rc.conf :p

ketas: i think both your patch and the rc.conf version that it's based on are kind of wrong

wrong may be too strong

while loop with 2 passes? other detections?

loader's semantics are a little more predictable because we would instead pick up any modifications after each file, rather than letting later files in a default set override what you might view as a 'request' to load an extra file from an earlier one

s/default/given/

(i.e., for loader_conf_files="/boot/defaults/loader.conf /boot/loader.conf /boot/loader.conf.local", each of those has its own chance to set a loader_conf_files and have more picked up)

hmm

probably a while loop and a "files_sourced" counter that resets each iteration is what i'd end up going with for shell

how many passes i wonder

(to be clear I wouldn't object to this going in as-is since it does match rc.conf semantics, just pointing out room for possible improvement)

well, until files_sourced=0

until no more files appear?

should it have limit or we could just expect people to not bust it

i wouldn't expect to need more than 5-10 in the absolute worst cases, I imagine

much better indeed

maybe no limit is needed

just let it run

i wonder if it slows something

in periodic.conf, no

in rc.conf, well

hmm how do i implement it now

oh yes

I'm running FreeBSD on a laptop that I need to connect to other wifi networks. In order to do this, I'm trying to use wifimgr, but am getting an error when I try to connect to the mobile hotspot on my phone: Invalid key for network. WPA keys must be 8-63 characters long or a 64 character hexidecimal number. My password is 12 characters. What is going on here?

ke8cqc: I don't know, but you can still use /etc/wpa_supplicant.conf

look what it wrote

I got wpa_supplicant working. I had to add Google's DNS servers to resolv.conf, but it's working now.

ehm

that's not wpa error

I may have used the wrong terminology, but I now know how to switch wifi networks from the CLI. It's not as difficult as I thought it was going to be.

kevans: managed to create first version of that, it allows infinite recursion and stops after it can't find any new files... that's what i actually wanted since beginning

ke8cqc: wpa_cli will let you choose profiles

ketas: i think you were on a better track with your previous notion that there should be *some* arbitrary cap on it as a safe-guard

if it takes 25 iteration to resolve, your product is probably poorly designed or we have a bug

!OPS

kevans: ok, runs max 25 times, cuts off before if no new files come out of $rc_conf_files... but should it give error if it ran max allowed times?

that's also in now

Hi! I am setting up a new machine with FBSD. On my linux systemd I have always used luks for backups but FBSD does not support luks. There is a library but it is not even in alpha state. Linux cannot hangle geli... And to complicate things further, FBSD does not manage XFS or F2FS, leaving me with NTFS as the best FS for backups, and using file-encryption rather than block.

Do you have any suggestion?

for what exactly?

backups.

data and configuration.

freebsd has all mentioned from linux world. backups, encryption and so on.

you do not need luks, ntfs, xfs, f2fs

It has to be encrypted because the backup destinations are external disks. And the FS has to support checksumming and repairs. ZFS would be too mucho for an external drive.

you can setup zfs encrypted pool for backups

why it would be too much?

qop: pkg install restic :)

pertho: hey, that looks like a great suggestion. Thanks man!

trying to install to newer ThinkPad's SSD/M.2/NVMe but 'gpart bootcode -b /boot/boot nda0s1' says 'invalid argument'

trying to install to that partition says same

i read FreeBSD disclabels won't be used for FreeBSD 15... what does that mean? What if I want to install in the old (MBR/non-UEFI) way?

or of course upgrade all my installations that only are that way

hmm where does it even say that

you mean you have to use gpart?