guys what window manager do you use on a laptop ?

i3

but like, thats such a personal thing lol.

wait, why would laptop or desktop make a diffrence with that question?

For me, a window manager for a laptop needs to have much better keyboard nav support, because almost all laptops have bad trackpads (and I don't pack a mouse)

i3 is a great example :)

wait, why would you want to control your wm or anything with a mouse?

i3 FTW! I use it both on the laptop and the desktop. But one must adjust the default keybindings because the original is definitely an unusual choice.

personal laptop uses dwm, work laptop uses kde with the same shortcuts to do the same things as the dwm one

nimaje: again, just for me, it sometimes comes in handy when opening remote desktops; I tried using connecting to an i3 remote session from a sway local session, and it got confusing

I almost never use remote graphical connections such as VNC. Almost never. Plain text over ssh is more than sufficient. Coupled with tmux/screen and a good editor and "Bob's Your Uncle" as they say.

But if I do need to do VNC to a remote host then I usually use a different window manager in the VNC session just so the look and feel is different to avoid the problem of which window manager in the nesting I am controlling.

Question: What do I need to know to run X with two graphics cards installed? I tried it and X does not start with a "Fatal server error:" bsd.to/3WMb/raw

Title: 3WMb

I don't have an Xorg.conf file at all and don't know how I should set it up to "Please specify busIDs" or whatever.

rwp: ordinarily I agree re. ssh/tmux. I use tmux to an absurd degree already. but remoting into VM consoles (if only during configuration) is one edge case I've just given in and used xrdp.

and yeah, same wm in host and remote is a recipe for madness!

The name xrdp makes me think MS Windows. Does it work with more than Windows?

it is the same RDP protocol from windows

A feature of tmux that I have started to appreciate is find-string in whatever window it is in. Appreciate because I have lost track of where I was doing what and can't find it now.

but it can serve an Xorg session to an RDP client on any platform; and RDP is a better, more functional experience than VNC.

I don't doubt that something could be better than VNC, by a lot even, as VNC is only incrementally improved over raw X.

I'm still drunk with power after getting tmux-resurrect running

I did not know that RDP could serve an X Windows session. I want to try it now. What do I need to know?

tpm is a great addon for tmux too.. if you are a vim fan (not meant to start stuff) vim-tmux-navigator is nice too

Most important is probably that it's my virt server and thus... *cough*kvm*cough*. On top of that, sometimes it got confused if a user has a local session when trying to connect to a remote session.

rwp i found that article quite helpful

Title: How to Set Up Remote Desktop in FreeBSD

I was unaware of tmux-resurrect until you mention it now and I go look. But when I reboot things I have a scripted startup to set up tmux and pre-populate panes as I want them.

I got about 1/3 of the way toward implementing my own tmux-resurrect (poorly) when I figured someone else must have done it. and they had.

I'm in awe of how it goes about reconstructing complex neovim sessions, but I'd use it even if all it did was restore panes in specific directories

voy4g3r2, What is tpm? As usual I enter here with one thing and then leave with twenty things I need to go investigate! :-)

haha

tpm is the tmux plugin mgr

it is a plugin manager for tmux

I am not using any plugins with tmux. It's fine for me as it is.

it's how you'd install tmux-resurrect... but I think you can do it manually too.

when i jump from machine to machine, i know everything is in sync an dhelps

need a theme.. boom tpm

want to load tmux-autoreload

tpm

I am browsing the list of plugins now. But so far nothing is reaching out and grabbing me saying I need it.

oh, I haven't used it to keep things in sync... though I do synchronize my tmuxrc across machines

penryu: i upload the latest tmux to git repo and then download updates

if i added anything.. i just do ctrl-s + I

and it just pulls everything

ahhh nice

i take a similar approach with my neovim and lazy

<3 lazy

find a great nerdfont and it is quite heavenly

i am a panes person, in vim, and i am told with buffer and some harpoon

you can be in heaven, have not quite gotten to there level.. so lazy + git allows me to move things around easily

harpoon?

Title: GitHub - ThePrimeagen/harpoon at harpoon2

you want some funny youtube/twitch stuff his personality is great

found him one day going down my lazyvim rabbithole

lua > vimscript

i was quite close to trying the mechnical keyboard he uses.. but he is a dorvak person.. cringed

then i had to spend all this money on a new file server... so mechnical keyboard goes on backburner

i guess my corsair k70 will have to do

it has been quite helpful when trying to edit files from src

especially freebsd14 tree and finding man page files

so instead of a huge :vs <location of file somehwere buried in src> i just load up harpoon and search for the file and it finds it for me

haha the license for fdwrite(1) is pretty funny... the beer-ware license

voy4g3r2: yaaa, that guy is pretty high-energy. will try out harpoon tho thx

yeah, i am still in the testing it out phase.. telescope does a lot of what i need so far.. just figured, he is funny and might as well give it a try

I really like fuzzy finding for finding a file in a project, but using it to switch _between_ open files is far from ideal. so I'd actually started doing exactly what he was mentioning: just :bnext/:bprev (mapped ofc). I think I almost have what I want with buffergator. so his rant about the mental pathway to harpoon definitely resonated with me

hmm, wonder where the pkg get the kernel version from, pastebin.com/3SchTYD1

Title: $ sudo pkg update -fUpdating FreeBSD repository catalogue...Fetching meta.co - Pastebin.com

phew, I had a bit of a slow migration back to CURRENT.

another reason to get a build server

yuripv, Did you set pkg to latest?

the problem is the "running kernel"

The paste seemed to indicate that it needed a running kernel later than the one you were running. It's a symmetrical relationship.

and at the end of the paste is kern.osreldate output showing that pkg got that wrong

Ah... Gotcha.

Run truss and grep to see where it gets that value from?

I have a /bsdisks.core no idea where it came from

is this a system file or just a crash from bsdisks

can anybody test -x /bsdisks.core so I know if I can erase this

Its not on my fs.

That's a core dump from bsdisks program. Bug!

wow, it's some black magic that pkg does, reading the version from uname and/or shell binaries (i wonder how it's related to "running kernel")

"freebsd-version -kr" ?? (shrug) Just grasping at straws. I would need to look at the pkg source to really know.

running current, and building with WITHOUT_CLEAN, so it's sometimes expected, i guess, still i wouldn't call that "running kernel" version :)

scary $#$@ too, bsdisks is a system program, amazing I didn't get a kernel panic

rsync is not automatically installed with the basic system ... Is rsync the tool of choice in the BSD world?

rsync isn't even the tool of choice in the linux world

and on freebsd? scp? sftp?

Try rsync first.

task is, download an folder recursive from another ssh-server

sound more like a task for 'scp -r' to me. rsync is only required if you have a 2-way sync inho.

Yea

oki, thank you

I use rsync -e ssh if I need to copy directory structure, scp -r has almost never done exacly what I need

or 'ssh host tar cf - path/ | tar xf -' if you have lots of files, scp can be QUITE slow

I like `rsync -a` for mirroring nontrivial directory structure, since it handles symlinks better than scp

you could probably find a comparable tar invocation, that might even be faster, but I haven't found anything to beat `rsync -a` for memorability

yuripv: cool I did not even know "tarcp" works over ssh

nmz: wait, bsddisks not only crashed for you but later caused a kernel panic too?

penryu: this? github.com/jeetsukumaran/vim-buffergator

Title: GitHub - jeetsukumaran/vim-buffergator: Vim plugin to list, select and switch between buffers.

voy4g3r2: yes, though I have now swapped it out for lspsaga's outline

er... no. I swapped buffergator out for fzf's :Buffers; I swapped symbols-outline out for lspsaga's outline

(in both cases the plugins were already there, I just had duplicate functionality)

If I understand the current status, then Raspi (1) should be supported as Tier 2 for the next 4.5 years while FreeBSD 14 is still in-use. Afterwards, it will no longer be supported. Am I correct?

so if i want to add hw.bge.allow_asf=0 and dev.bge.0.msi=0 to loader.conf, can i also just sysctl those without rebooting? what is needed to make them work?

rpi 1 is armv6 which means it is Tier 3 on FreeBSD 14

the man page says they are only settable by the loader, maybe it works if you can unload and reload the module, but I would expect it to be compiled into the kernel

bge compiled in kernel

is the rc.d script of wireguard-tools packages compatible with 13.2+ BASE if_wg?

I'm wondering why there is no rc.d script to setup wireguard

paulf, Oh. Thanks.

mage: it should be

mage: there probably should be an rc.d script in base, but i just use /etc/start_if.wg0, which is executed automatically by rc when the interface comes up

(it only has a single line, 'wg setconf wg0 /etc/wg/wg0.conf', and i do address config with ifconfig_wg0 in rc.conf)

at the least, it would be nice if you could setconf_wg0="/etc/wg/wg0.conf" in rc.conf

there is also blog.rlwinm.de/the-missing-wireguar…egration-into-rc-d-for-freebsd-13-2 and blog.rlwinm.de/wireguard-configuration-for-freebsd-13-2

Title: The missing WireGuard integration into rc.d for FreeBSD 13.2 — blog.rlwinm.de

hm, where is that /etc/start_if documented?

oh, that uses devd which is an interesting approach

nimaje: rc.conf(5)

yeah, I want to build something based on that blog, so that I can bring the tunnel up and down via ifconfig and have them configured in rc.conf

i'm always a bit suspicious of devd because it seems somewhat magic, but having the config loaded automatically would be easier than having to specify it manually. it might be confusing if users don't understand why that's happening though

do you know if the author of those posts submitted a PR?

afaik no, and as I read those posts it seemed a bit unfinished, no idea where he works on that, maybe only in private

Hello! How to specify multiple groups when using `pw groupmod newgroup -m user'?

I tried with `group1,group2,group3' and it generates an error `pw: unknown group'

and if I try with `group1 group2 group3', actually user only gets added to group3 and the first 2 are ignored

rockyh, I don't think you can do multiple groups, maybe you can, but you can loop over multiple groups and add them one at a time. for grp in group1 group2 group3; do echo pw groupmod $grp -m user; done

I didn't double check the syntax, I just modified the pw command you posted. Maybe it's correct. That's why the echo is there. Look first and remove the echo only when you think it is correct.

Thouh I know a lot of people just pipe the output to the shell when they think it is good. | sh

Probably "| sh -x" is better. I should have put a -x on the end. I usually remove the echo when I fire something like that for effect.

rockyh: you can use pw usermod user -G group1,group2,group3, but this requires you to specify the entire group list

otherwise, like rwp says just call groupmod 3 times

I admit that when I saw the object-verb-object combination that I did not realize why it confused me but now I do. The original example modified the group adding a user. But actually it needed to be modifying a user adding multiple groups instead.

The original form (groupmod) would have probably been suitable for adding multiple users to the group.

Always fun when I learn something new about how to think about the commands. :-)

well, technically modifying a user to add them to groups is really modifying the groups to add the user. but you don't always think of it that way in practice... probably depends on whether you just added a new user or a new group

"pw groupmod g1,g2,g3 -m user" wouldn't be a bad thing to support

Using pw really abstracts things and distances the command line from what's actually happening. Personally I would rather just edit the /etc/group file manually.

^^^

How are newcomers going to be learning how user accounts actually work when we put up such barriers? It makes it harder not easier.

rwp: do you really want to go back to using sed (or whatever) to edit /etc/group in a newuser script? :)

this is going to be an odd question, is it common for the freebsd handbook to reference older releases? such as the jails section.. it is referencing 13.2 stuff but not 14.0

voy4g3r2: if it says "this feature was added in 13.2" that would seem normal, if it says "you can only use this in 13.2" (when it also works in 14.0) that should probably be rephrased

handbooks updates are not really synced with releases, they happen whenever someone decides to do them

lw, I am certainly okay with using system tools for automation when provisioning a system or when rolling out a change to an entire network. But if I am just adding myself to another group then I just tend to use emacs and edit the file in that case for one-off things.

(this someone can be you! i know you're already working on manpages...)

And of course most large collections will use something networky to handle user accounts. NIS/yp, NIS, SQL, something. It's only small SOHO networks that don't.

s/NIS/LDAP/2

lw: it is referencing 13.2 images

I just totally blanked on LDAP there for a moment.

rwp: fairly commonly puppet, which i'm fairly sure uses pw(8). i mean, i don't disagree, i also prefer to use vipw because i can never remember the syntax for pw anyway. but i mean, i wouldn't support removing it to teach users to use vipw instead :-)

Title: Chapter 17. Jails and Containers | FreeBSD Documentation Portal

voy4g3r2: i think that should have a note added to tell the user to use their action release (which should be fairly obvious, but just to avoid confusion). i doubt someone will remember to update that example on every release, and even if it used 14.0, some people still use 13.2

s/action/actual

lw, s/vipw/vigr/ I don't know. It is useful to know how the machinery works under the hood.

voy4g3r2: it could perhaps replace '13.2-RELEASE' with $(uname -r), except i don't think patch releases get new distribution sets on the mirror

Puppet, Chef, Salt, Ansible, and so on. I am one of those people who wrote their own configuration infrastructure system from before those existed so I still use my own for obvious reasons.

voy4g3r2, The steady state of disks is full and documentation in slight need of an update. It always takes a while for docs to catch up.

nimaje: it didn't, that's why I was surprised

The best person to correct and improve documentation is the newcomer reading it for the very first time.

voy4g3r2, I looked at the example and in this case I think it is fine as it is. Because it works on either 13 or 14. If the example were updated to use 14.0-RELEASE then that would likely fail on a still supported 13.2-RELEASE host system. Jails should not be newer than the host due to likely potential use of newer system calls than the running kernel supports.

fair enough

Alternatively it could do as you say and use uname -r or some such to dynamically produce the same download as the running system. But that would make things more messy and possibly confusing for the new user reading it.

i am just walking through it and was expecting one thing and saw someone different

The docs could definitely be improved and expanded to explain the kernel version requirement. That would be a win-win all around. It just needs someone to do it.

hehe yes

i wonder if i just downloaded the src or also the docs area

first step, see if i can even get a jail to be operational

Also since you are able to read the current documentation and be able to interpret what it is conveying you can always adapt the command to match your uname -r and match your current system if you have upgraded.

The skill to read docs, examples, blogs, tutorials, and adapt them on the fly to a different OS, a different OS version, a different architecture, is a useful skill to learn and master.

yeah. i have been doing that over here: bughuntingfreebsd.wordpress.com/202…send-zfs-recv-you-say-lets-find-out

Title: sanoid and syncoid you say? easier than zfs send | zfs recv you say?.. lets find out. – FreeBSD Bug Hunting

for example

rwp: yes, suprisingly this is what i do for work all day.. take something someone told me, then translation for multiple stakeholders..

the art of meeting minutes, something i never thought would be useful.. they are quite useful

More useful as the brain ages. I miss my 20yo brain!

Fun! Lunch time here. BBIAB

yeah, i should head down tot he gym myself.. while this jails does a freebsd-update

i can't figure this out. in what must be connected, i previously had an entry in dhclient.conf that specified another host IP. the system was using a static IP, but i wanted another host IP for a jail to use (shared IP), and i thought this would work. It didn't. I abandoned the idea, but i accidently left the entry there until recently. anyway, now i have unrelated loopback jails whose requests are being given that IP on

their way out. my pfsense is blocking this unused address. the fbsd box is at 192.168.7.50. the jail is 10.0.0.50. when I tell the jail to pkg update, pfsense (192.168.7.1) is blocking those requests which it sees coming from 192.168.50.69 (the address that sat in dhclient.conf for a while).

it seems I need to just delete the route that stuck around, but I can't: bsd.to/qOgQ

Title: dpaste/qOgQ (Plain Text)

how.. bsd.to/qOgQ/raw looks much nice

Title: qOgQ

if i helps, i originally set up this box with a static IP (on that 192.168.50.0 network), but now i switched it to DHCP.

i can see in /etc/pf.conf that i am natting on ext_if, but that's set to 192.168.7.50, so i'm not sure why pf (is pf doing this?) would see 192.168.50.69 as the ext_if

i was thinking i could solve this by deleting the erroneous route in the routing table, but as the paste shows, it won't let me delete it because "gateway uses the same route"

scoobybejesus, Can you paste the ifconfig output? I think flags=UH means that is a host route. Which usually occurs because the IP is still assigned. (And yes I always only ever use the paste /raw URL.)

hi rwp, here you go: bsd.to/nhVy/raw .. yeah, you can see it there.. maybe i need to delete it using ifconfig...

Title: nhVy

which i did, but it didn't go away, and i figured that was caused by it sitting as an entry in dhclient.conf. but maybe i haven't deleted it from ifconfig since removing it from dhclient.conf

Right. You will need to remove the IP using ifconfig and that should remove it from the route table. It's an IP address in the route table which is effectively a /32 address IIUC.

Sitting in dhclient.conf just sitting there won't do anything. That's just a file. But dhclient reading the file dhclient might do something. But ifconfig and netstat list the kernel networking state as it is now and ifconfig and route can take actions to change the current state. (though dhclient might put it back again)

Be sure to look at the /var/log/messages system log to see if there are clues logged there from dhclient or other logging actions there.

So far while experimenting I have gotten my FreeBSD system into some pretty crazy networking states but have always been able to pop the stock and unwind the crazy and reversed things back to undo the crazy and have not yet needed to resort to rebooting to get to a clean state. So far.

pkg update working once again!

thanks for your help rwp!

Glad to have helped! Good luck!

anyone knows if hex0 or any other bootstrap process is available for FreeBSD?

unixman, are you the witch?

jbo: i am the witch

ack

lw, are you going to fix that other PR as we discussed?

jbo: uh... i think i forgot what we discussed, sorry :-( which PR was it? i can go look now

lw, 275947

oh, i did the other two, i must have forgotten that one. let me testport it

it's missing the category makefile entry, DISTVERSION vs PORTVERSION and ensuring that poudriere testport succeeds

I can't do the other(s) without this one first :)

jbo: new patch on the PR

ack

lw: thanks for all your suggestions. I find troublesome to specify the whole list of groups

rwp: thanks! Probably editing /etc/group is the best and more straightforward solution sometimes

it would be great if developers considered adding support for multiple groups in `pw groupmod'

I am very surprised that such support does not exist up to now

rockyh: can you explain your use case?

meena: I create a user `testuser', and its default primary group `testuser'; I also add `wheel' as secondary group. In another moment I would like to add some more secondary groups to that user: for example, `operator', `video' or similar. I would like to use a single command to add all the secondary groups, not to replace the list of secondary groups, but to append the new groups to that list. In

the example, I would like to keep `wheel' as secondary group, adding `operator` and `video'

it seems that this is not possible through `pw groupmod <group list> -m testuser', because there is no way to give a "group list". Only a single group can be provided. It would be useful if this functionality were added