Hey all! Is there a way to make FreeBSD sleep an attached (HDMI) screen after inactivity, when just using the ttys, and no desktop environment? I just want it to stop outputting a signal without me having to unplug the cable or turn off the screen

Axman6: Despite the channel name, I'd ask the folks in #freebsd-xorg on EFnet.

giusb: I have `rdr pass inet proto {tcp, udp} from !<this_host> to any port {80, 443} -> $jail_caddy_ip` where <this_host> is a table containing the host IP's (host plus vnet jails). otherwise all http/s traffic pipes into caddy instead of heading out of the box. caddy is a loopback jail though. is yours vnet? anyway, going to bed. hope this helps

mason: Thanks, I'll connect sometime soon in that case

good morning

Do you guys know of a good BSD licensed version of twitter bootstrap?

lisbeths: you mean like this thing here getbootstrap.com/docs/4.0/about/license ?

Title: License FAQs · Bootstrap

MIT License != BSD License

Mit License < BSD License -> 10

MIT license > BSD License => 10

s/->/=>

u what mate

MIT License != BSD License.

BSD licensed? copyfree she means?

out of curiousity what do BSDers use for services?

given that theres no systemd

init

hmm

well I was hoping to start some rclone mounts services as a user, not as system

Then you'll probably use FUSE.

i'll use runit

I was hoping there was something I could use crontab @restart from

...

xa0z: whats wrong?

you still need service supervision, FUSE has nothing to do with it

lol

whats simplest/easiest, perp,runit,s6,daemontools?

I am using newsyslog to rotate several Apache log files

I want to rotate all of them @ 23:00 , how can I do that and only have newsyslog generate a single sigusr to Apache

aka, I don't want multiple graceful reloads for each log file

or do I need to set each log file at different times, aka 23:01 , 23:02, 23:03, with the N flag to not signal Apache

and only let the last rotate log to signal Apache ?

asked differently: If I have 5 entries in newsyslog, all running @ 23:00 , do they run in parallel or one after another ?

dch: have you ever tested building tailwindcss on arm64?

drobban: yes, it doersn't work, the nodejs stuff is a catastrophe

fuck me.... as i suspected. I hated node before, it gets further up on my shit list

dch.... seems like I perhaps should start looking on how you have used simple.css instead.

you can npm install tailwind on arm64 just fine, then softlink the name back in

ummmm

drobban: I have one here, let me look what I did

O already found it ;)

no, for arm64, I made no notes for that IIRC

npm install -D tailwindcss@latest postcss@latest autoprefixer@latest @tailwindcss/forms

oo okey

ln -s ~/node_modules/.bin/tailwind _build/tailwind-freebsd-arm64

dch: thx!

just those two

thanks!

and should work with node20 but this box used node18

slap me with a wet fish next week if I didn't write a blog post about all this garbage

dch: hahaha ! i will.

ofc with @latest everywhere who knows what garbage gets pulled in and we are back in zero reproducibility land again

dch: yea. Im planning on using my ampere altra to spin up a service using elixir/phoenix... Cant understand how a freaking css framework is so complex it has trouble to build on freebsd and arm64 :D hahaha its freaking crazy

drobban: most of my stuff is ampere + freebsd here, it might be worth cooperating on at least a couple of private packages for tailwind etc

its not possible to do as an official port, but a private one would be quite doable I think

bbl, family stuff

! sure. cya

Hi all - I'm struggling a bit with understanding freeBSD's network tunables for a mellanox card - is there a utility for finding out acceptable values or docs somewhere?

I'd like to change forward error correction, but I'm not sure what values to use for something like `dev.mce.<N>.conf.fec.mode_active: 4`

is there something better than trial and error?

Fhttps://network.nvidia.com/pdf/prod_software/Mellanox_FreeBSD_User_Manual_v3.4.1.pdf

Alloysious seems to be gone, but if they come back...

Also mlx5en(4)

nmz: for service supervision, daemon(8) can do most things

is there rationale for why some ports have a more recent version in quarterly than latest? or is it up to the maintainer to decide?

markmcb: quarterly builds less stuff, so it builds faster

this doesn't match what the handbook says

"the Latest branch provides the latest versions of the packages to the users"

the text in the handbook makes it sound like it's impossible for quarterly to ever be more up to date than latest

unless I'm fundamentally misunderstanding

markmcb, I think what happens is that the Latest build queue is always full and so it takes some time (days, weeks) to get through it while Quarterly has less churn and less build work to do. This creates a situation where a version 123 is placed into both and Quarterly will have it finished while Latest is still working it through the queue.

Which package in particular are you waiting for to get through the build queue?

rwp: thanks. makes sense. no specific concern, i just notice it frequently. it's more of a new user experience issue. it makes freebsd look like it's unstable or broken.

random question prompted by a discussion on another channel

does anyone else still use config(8) to build the kernel?

lw: other than the build-system?

meena: what

it;s still used underneath

oh

yeah i mean if people actually type 'config' at the keyboard

Does anyone know if it's possible to write a keymap entry for a key that produces more than 1 byte? -- The atkbd keyboard driver has a few hardcoded bits for specific keys with a 0xe0 prefix, but i'm having to process keys that have a 0xe0 prefix not covered by the atkbd driver -- moreover, it seems that it never hits that switch statement anyway because state->ks_mode == KS_RAW by default (which seems separate from the keyboard mode set with the normal ioctl(0,

KDSKBMODE, K_RAW) -- which i'm assuming is because it's intercepted by the kbd driver and not passed on to the atkbd driver...