uskerine: based on that report, that looks to be the case, except for bluetooth

on 13.4, i'm getting some daemons killed with the log message "failed to reclaim memory". top reports that almost all memory (7.5G out of 8) is wired, and almost all of that (6.3G) is ARC. would limiting the memory for ARC likely prevent daemons from getting killed? (i'm not entirely sure what "failed to reclaim memory" indicates)

thorongil: limiting arc would probably fix that, yes. there were some major fixes to ARC memory reclaimation recently, i don't remember if or when they went into releng/13 though

thanks. will the machine need a reboot, or will setting the sysctl take effect immediately?

iirc, you can set it without rebooting, i don't remember whether it'll immediately shrink the arc or not

thanks

for the record, after setting it with sysctl, top shows the ARC has shrunk, but wired memory has not. well, not quite: wired memory is shrinking veeeeeery slowly (like 4M every 10s)

Can you search the old PR system or were all those those migrated to bugzilla bugs?

ok nm. A simple search says yes they were.

i'm starting to play around with jails and i'm trying to figure out whether i want exec.start = "/bin/sh /etc/rc". i want this jail to only run a single daemon installed from ports, which i intend to start manually rather than from /usr/local/etc/rc.d. what benefit would exec.start = "/bin/sh /etc/rc" have in this case?

thorongil, It is perfectly okay to not use rc within a jail and simply to start whatever processes you want running in the jail. That's one of the many normal configurations.

cool, thanks

I'm trying to run something under linux emulation and it's complaining about missing libcrypt.so.1 which I gather was part of glibc but has been removed, what the right thing to do here

crb: the easiest method is to create a chroot environment containing the version of Linux you need and install the software there, that way you're not tied to whatever version ports currently supports

anybody done arduino development on freebsd?

I have an inkplate device (wifi esp32 + e-ink screen) that I want to set up

inkplate.readthedocs.io/en/latest/get-started.html if people are curious

dch: a long time ago

I've done micropython from freebsd as recently as yesterday

oh that is good to know

and inkplate mentions micropython

yes it does

we have a few wiki pages on arduino related stuff, so this looks like its going to be worth trying to make it work

nah its awful

oh dear

micropython is just python3

I'm assuming your goal isn't hack on a microcontroller, but "do something"

yeah, initially i just want to put a handful of words on the display

and update them over wifi

your life will be much simpler by using micropython

and I will be helpful rather than condescending

how will I know the difference over irc tho?

cool, so my first step will be to find appropriate cables. I may as well look for wifi chips while I'm in the cellar.

I keep bicycles in the coal seller and wifi in the attic

hmm, i was hoping i could use pf's new af-to to redirect ipv4 traffic to an ipv6 address for a single host, but this doesn't seem to be supported

well, freebsds pf forked from openbsds pf a good while ago "This implementation is derived from OpenBSD 4.5."

i'm not saying it would work on openbsd (i have no idea) only that it doesn't on freebsd :-)

(specifically because af-to seems to be designed only for 464XLAT and doesn't seem to support translating to a single destination IPv6 address)

well, with "pf's new af-to" you probably mean openbsds pf there, so it is no wonder that it isn't in freebsds pf and as those implementations diverged quite a bit, as freebsd wanted SMP support there and openbsd implemented other features, …

no, i mean the feature that was added to freebsd a couple of months back

hm, why does it say "FreeBSD 13.2" when I search for the -CURRENT man page of pf.conf on man.freebsd.org?

I'm a bit confused setting up WiFi

net.wlan.devices says iwn0

iwn0 is also in my dmesg

ifconfig doesn't show it?

runxiyu: iirc you need to create subinterfaces, like 'ifconfig wlan0 create wlandev iwn0'

ah, thanks

i'm new to freebsd (came from openbsd and linux), got myself a bit messed up there

ivy: seems like it landed sometime in december; are you on -CURRENT?

nimaje: yes, and i'm using it for NAT64 already and it works great :-)

(well sort of, i reported one bug and i need to report a second one... but the basic functionaliy is there)

err, wpa_supplicant is rejecting mac_addr=3 (use dedicated/pregenerated mac address in mac_value) even though the example config file seems to support so

ivy: as I understand it, it is mostly meant for nat64 but your usecase should be supported to (just from what is written in git log, no idea how it would work)

nimaje: well, i tried 'pass in from any to 1.1.1.1 port 53 af-to inet6 from 2001:db8::1 to 2001:db8::2/128', but it seems to always treat the 'to' address as a /96 and replaces the lower 96 bits with the IPv4 destination address

is it possible to do a keydisk encryption from the installation menu

or does it have to be done by geli manually?

iirc you can only do passphrase via the installer

hi everyone, how do I store my music in hdd? #newintothis

42

There must be something the FreeBSD foundation can try doing to reach out to companies like Zoom, Salesforce/Slack, AWS Chime, Microsoft Teams etc.. to have native clients on FreeBSD for conferencing products. I've tried using FreeBSD as a day to day driver, but always run into a blocker when it comes to actually talking with my co-workers (yes unfortuantly we do need to do that occisionally). I end up having to run both Windows and FreeBSD for that pu

that would be the dream

how do those things work in a modern chromium browser for you?

tuaris: the problem is that google doesn't really care about anything that's not linux, windows or mac when it comes to chromium-related software, so there's not much you can do.

until upstream starts caring about freebsd, it will always suck

it's a real shame that MSFT went with Ubuntu instead of FreeBSD for the WSL (which, I guess, would need a different name too)

it is all about business, i use Windows and learning Freebsd witch I love it but programms its a bottleneck'; and if you can find one on FreeBsd Port/Package collection then they abandon or no do more updates :)

yeah, community software projects can require some effort

well, "some effort" for things like chromium is at least 2-3 people on FTE working on the port :P

sometimes there are newer patches available in the Bugzilla site, bugs.freebsd.org/bugzilla

and if not (I know it's a lot of reading/yak shaving), there is good documentation about how to submit new/updated ports in the Porter's Handbook: docs.freebsd.org/en/books/porters-handbook

mfisher: chromium isn't my thing, but the experiance between it and Firefox isn't that diffrent.

Now, I'm running into another blocker. I got to get an updated version of the Teleport client working on FreeBSD, and... well they have not been to friendly with supporting FreeBSD in the past.

When I talked to them, the basiclly told me to pound sand

yeah, sounds about normal :(

firefox is probably less diverged from upstream on freebsd than chromium is

the reality is that if we want a properly working chromium on freebsd, the FF will just have to pay people to do it. at least until upstream decides to accept patches

this also applies to electron, CEF, and to some extent, nodejs

The last time I tried updating the teleport port was a few years ago and ran into some rate limit issues while generating the GH tupples for go mods. Maybe I might have better luck now using the same approach I took while making the grafana tempo port

I just have a hard time understanding why browsers need so much development work on them

what is it about chromium that needs fixing? what new feature does there need to be? the more features the more websites get uppity and start abusing them

ring0_starr: everchanging standards (not just web but also graphics), more stuff getting added and people wanting it to be really fast while remaining reasonably secure since it's arbitrary code execution as a feature

The HTTP protocol is being used so far beyond it's intended purpose.

exactly.

it can be boiled down to "abuse of a thing that started off with good intentions"

and "support of said abuse"

as for freebsd, a good start would be having any kind of sandboxing in chromium

since right now chromium on freebsd runs without any real sandbox

i thought chromium itself had the sandbox!?

yes, on linux, mac and windows via seccomp & friends

it doesn't use capsicum, though

it's being abused. They really should have created an entirly new protocol. In some strange way they did, but they called it HTTP2

It should have been called BTTP, lol

so in effect, the rate of 1-2 ACEs per week in chromium (via v8 or something else) result in a proper ACE as opposed to needing a sandbox escape first. unfortunately capsicum support requires a lot of work to maintain, which is why i say that the FF will basically need to pay someone. it's unfeasible to do as a hobby

imagine how great the world of tech could be if it weren't subject to external pressures

every project is incredibly overcomplicated. I just clicked the button in github on some function in llvm to see where it gets defined, and i got this "<repo> cannot be searched because it is too large" message

github with all its mighty microsoft resources can't even bear the weight of llvm, which is smaller and simpler in design than gcc

it is all about business, i use Windows and learning Freebsd witch I love it but programms its a bottleneck'; and if you can find one on FreeBsd Port/Package collection then they abandon or no do more updates :)

i don't know that llvm is smaller and simpler these days, it's probably about the same. llvm has around 12 million non-blank lines of code in total based on running tokei on it. that also includes tests & all, but that is all indexed by git

and the solution to all this complexity is to throw more tooling and more resources at it, when any real person would stand back, look at the mess, and stop

it's like everybody in the world has a meth habit

ring0_starr: yeah, Docker to solve the complexity created by all the tooling for language package managers, then k8s to solve all the complexity created by Docker.

What will solve k8s? :3

The world will finally come to it's senses and solve the problems cuased by k8s by using FreeBSD everywhere.

hah

i don't get why those are even used

and how widespread the usage is

the idea isn't bad actually

k8s is basiccly a distibuted operating system. It even has it's own package manger (helm)

but it's a disaster for. every org I've seen using it has no idea how to do system administration