00:30:53 Ope. Time to move to Linux for all tile cool new features https://social.treehouse.systems/@astraleureka/115844218496725551 00:31:09 s/tile/the 00:42:56 Honestly a fine enough idea if it didn't take the form of "surprise! There's a network service you won't notice because your usual tools don't list it and PID 1 listens on the socket so you don't even have a 00:43:17 suspicious process to investigate or google 00:43:53 ... hmm. rc.conf option? 00:59:25 zip: not a surprise that systemd would do something that wacky 01:05:51 I could do worse 01:33:27 has the package kitty been removed? this is a fresh install of freebsd and I have got hyprland up and working but the package kitty is non existent 01:34:17 i googled quite a bit to see what's happened and i couldn't find anything about this 01:35:47 seems like it failed to build on 15.0 https://portsfallout.com/fallout?port=x11%2Fkitty%24 01:36:31 that must be the case i was just looking at fallout 01:37:13 there was a commit yesterday that should fix it, but no idea when it will be in the repos https://cgit.freebsd.org/ports/commit/?id=fa18e9953d38cc0aaf90db71c50442156358ced2 01:37:39 thank you <3 05:33:36 pkg: sqlite error while executing INSERT OR REPLACE INTO packages (origin, name, version, comment, desc, arch, maintainer, www, prefix, pkgsize, flatsize, licenselogic, cksum, path, manifestdigest, olddigest, vital)VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7, ?8, ?9, ?10, ?11, ?12, ?13, ?14, ?15, ?16, ?17) in file update.c:175: not an error 05:35:27 rtprio: Sounds like desc is a reserved keyword. 05:36:21 well, that's up to whoever was hacking on pkg 05:36:37 also the irony of pkg: error ... ... not an error 06:31:06 rtprio: it's harmless 09:15:57 If Chimera Linux can try to port the BSD userland to Linux perhaps it's time to port systemd to BSD. 09:16:47 Actually that sounds like something you can start as a joke and then be stuck running for two decades 09:55:51 Getting the error as well, also for packages.path. Others reporting the same at https://github.com/freebsd/pkg/issues/2584 09:56:37 It's not harmless, it's preventing installation of packages into a new jail 10:02:25 zip: why would you want to port that unportable and badly designed piece of software? I give them that they implemented some good ideas like proper service management, but rewriting the good parts in a good designed and portable way without all thoses interdependencies should also be easier than to port it, especially as parts are already there, like in the form of the s6 tools suite 10:03:11 As a joke 10:05:47 fortunately they decided to write it in the most unportable way they could find 10:06:23 Which thankfully makes it a high effort joke and not something you could end up building and being expected to maintain 10:06:59 s6 seems like the sort of thing you could add to rc services as a flag tbh 10:13:20 well, someone made the experiment of replacing init completely with s6 tools http://static.bultmann.eu/s6-talk/ 10:15:44 yop 11:47:14 well 11:49:47 Nice trololololol zip 11:49:49 part of me suspects that for a lot of use cases, the service monitoring loop doesn't need to be any tighter than having an alert to the sysadmin during working hours 11:50:06 barely trolling, more goofing :P 11:50:45 I think it's a valid point that you gotta walk carefully without it becoming heated. 11:50:52 I wonder if there's stats out there for how often httpd, nginx, sshd etc actually crash on their own 11:52:13 I'll hate systemd all day but I'm quite open that my main reason is mostly that I learned all this shit before systemd existed and I hate that it's new and different 11:53:26 I like how things work today and I'm not a systemd fan either but at the same time, I can see why you might want to implement something more "modern" 11:54:35 it feels nondiscoverable and it feels like I have to go and look things up every time I interact with it. Part of me also recoiled at freebsd's rc system for feeling too much like an old fashioned in-order rc script and for `servicename_enable="YES"` feeling a little bit magic... but I've very much warmed to it 11:55:39 no small part of that is that `man rc.conf` and `less /etc/rc.subr` are most of the tools I need to work out what something is doing or how to do something, including making new services 11:55:47 I agree. systemd feels like a black box you can't penetrate where as BSD is generally much more transparent 11:57:15 I don't really know or understand systemd at all, and I've learned it twice already. 11:57:21 o0x1eef: according to the systemd devs, systemd is super simple. I mean, I've gotten used to systemd here at work over the years. But I've yet to find the simplicity of it :D 11:58:31 Any system I learn and then still have not learned is one I'm inclined to dislike 12:00:09 you just need to place your socket/timer/service files in the right place, fill the content with WantedBy,RequiredBy,NicetoHaveBy statements and it just works™. At least to the devs 12:00:49 and of course it's quite nice to know (or at least believe) that reviewing `{/usr/local,}/etc/rc.conf{,.d/*}` and `/boot/loader.conf`are going to give you a very solid idea of what a system is going to load and start 12:00:51 without being nix 12:00:56 it's a clusterfuck, but given who wrote it, it's no surprise. 12:01:27 that was in regards to systemd, not rc 12:01:27 I don't have the words except that it sounds like the Ruby on Rails of init(8) systems (not a compliment...) 12:01:37 I'm also half-sympathetic to building binary logs but again, it's a matter of learning the incantation for reading them 12:02:12 at least put `/var/log/README` in place with a line or two that says "you want to run `journalctl -xu `" or whatever 12:02:37 zip: i actually like journalctl. but the amount logging data on a modern linux desktop system (fedora in my case) is insane 12:02:45 I think I'd hate it about 50% less if I found the directories where things were configured and it was a short step from that to the manpage I need to read 12:02:46 but that's not journalctl's fault 12:03:22 maybe that'll be my first debian package, eh 12:04:35 just something that dumps a bunch of `README` files all over the filesystem so that sysadmins like me who are likely to try to figure out how a thing is configured with `find /etc -name "*service*"` get pointed to the right manpage 12:04:48 "hi, you've reached `/var/log`. Run `man journalctl` for more information" 12:06:26 zip: systemd is just one of the absurd things common in modern day Linux ¯\_(ツ)_/¯ 12:06:46 my filesystem is littered with notes to myself 12:06:52 `jsconfig.json` says `"comment": "You can have this file and have Command-T searches or you can delete it and have 50% of intellisense"` haha 12:07:23 clearly, the last time I said "it's fucking annoying that intellisense doesn't work" I spent a couple hours discovering this 12:08:08 To be fair, Linux itself is just a kernel, and the init system is whatever you choose. That's the theory. In practical terms it doesn't work quite like that. I don't think debian or ubuntu provided options. It's systemd or nothing. 12:08:20 zip: my huge amount of notes is in ~/Notes and synced with nextcloud everywhere. i'm too old to remember stuff. 12:09:01 o0x1eef: i think there are only about two distros without systemd left. 12:09:20 o0x1eef: gnome without systemd is also nearly impossible 12:09:34 yup. that's why I've ended up on Void for the tinkering system, although I'd probably use Fedora if I wanted a Just Works™ laptop and to never attempt to run any services 12:10:01 Void / runit is nice. 12:10:08 zip: i actually run plenty of service in my user session with systemctl --user 12:10:19 zip: i find that rather handy 12:10:27 it's pretty good. mostly discoverable, helpful to understand service control directories, otherwise pretty nice 12:10:47 and again discoverable, the bits that run before services boot and for shutdown are in nice shell scripts 12:11:27 what runit doesn't give you is dependency management better than seeing if it starts, sleeping one second, and trying again in a loop 12:12:21 zip: the thing about shell scripts on linux is, they are mostly written for bash, which even today seems quite buggy to me 12:12:23 I have a runit service directory in my homedir for that and it's mostly okay because most of the services don't have dependencies. What's annoying is I've not worked out what the dependency is between the thing that tells me what music is playing and something in the system services such that if I auto-login it loads juuuust fast enough to break 12:13:18 well, maybe it's not the best idea to have my ridiculous hand-crafted artisinal swaywm-on-runit setup in the first place 12:13:42 systemd is, unfortunately, the compromise you have to make for the system mostly "just working" most of the time 12:15:21 running through the process of building a modern desktop without Gnome or KDE has been a lesson in how DBus _also_ has a bunch of logic for service activation, how it's mandatory now, how you need seat management to make sure it's up and running before your shell, how to tell it about environment variables otherwise set after it starts (such as the X or wayland display...) and so on 12:16:38 and then you'll want your desktop portals so flatpak works, and a secrets management service is nice, oh, and remember that native clipboards are just a channel between two applications so you want a third one that sits in the background and whose entire job is to sit in the middle so you can close the first application and still paste whatever it was you copied 12:16:59 well, the thing about shell scripts is, that they are buggy if they are more than one simple line 12:18:03 What's the way to install freebsd in a chroot (host machine: Arch linux) 12:18:10 and of course if your wifi configuration is more complicated than simply starting up and connecting to a known network you probably want NetworkManager, and so on 12:19:04 I find myself constantly torn between understanding why the complexity exists and not having any constructive suggestions for improving it, and yearning for the good old (and frankly, FreeBSD-shaped) days 12:19:05 avid: I don't think arch linux provides some freebsd ABI, maybe you have some luck to compile the freebsd userland for arch linux 12:19:38 No I meant only the chroot installation not for usage as a container 12:19:48 what are you trying to achieve? 12:19:50 i.e: installing an OS the arch way 12:20:03 aka as rootfs installation 12:20:23 mount /dev/sda1 /mnt; sudo tar xvf freebsd-rootfs /mnt 12:20:28 I'm trying to achieve something like this 12:20:44 well, that bit doesn't really need the chroot 12:20:45 Use the VM images and mdconfig(8) with mount 12:21:30 and I'm assuming here you're trying to do this on ZFS because otherwise you won't be able to both write it from linux and boot it from FreeBSD anyway 12:21:37 does freebsd offer official torrents? 12:21:39 but yeah, use a VM image 12:21:41 I found 3rd party foss torrents 12:22:03 if you would like to use FreeBSD and also have it access your linux filesystem, that's what plan9 filesystem is for 12:22:30 If you do use third party make sure to download the checksum files and to confirm. But otherwise, no 12:23:28 No does not provide official torrents AFAIK but a freebsd dev maintains their own. 12:23:36 which option should I go with? zfs.qcow zfs.raw zfs ? 12:23:41 raw 12:23:44 depends what you're trying to do 12:23:58 you want raw for it to work with mdconfig 12:24:02 oh, right, yeah 12:24:25 but this does sound like a "why are you trying to do that" kinda problem 12:24:40 if you want freebsd on bare metal hardware, install it from the USB stick installer 12:24:50 if you want to dual boot FreeBSD and Linux, probably, don't 12:25:02 or at least, don't try to do it on the same hard drive 12:25:09 It may be doable, but not without substantial pain 12:25:31 I find installing the arch way very educational 12:25:39 I want to learn the exact boot process for bsd 12:25:52 i.e what would be the equivalent to vmlinuz and initrd.img 12:25:56 FreeBSD has its own ways 12:26:01 If you wish to learn the dark arts it is a fine place to start, but if you wish to try out FreeBSD and see what it's like, do it in a VM or install it on a separate hard drive 12:26:02 does it support uki? 12:26:26 FreeBSD doesn't really have init ramdisks 12:26:34 nor do we have kexec 12:26:44 The man pages describe the boot process, and the handbook as well. 12:27:00 we _do_ have running the shutdown script, switching root, and then booting (with your original init, I believe) from that new root 12:28:37 Yup. Multistage bootloader off a msdosfs partition. Written in lua, I believe, although there's an older forth version too. That'll load up the kernel and any modules, set the command line options (of which there are approximately five) and then kick off the kernel. 12:28:58 It's not like linux where you have a million command line flags. The flags are basically "use the serial terminal as the console" and "boot in single user mode" 12:29:18 other things can be set with `/boot/loader.conf` such as which modules to start with and kernel environment variables 12:29:25 zfs on dmcrypt supported? 12:29:35 or is dmcrypt linux thing 12:29:37 I think there's some magic for zfs boot environments 12:29:41 dmcrypt is a linux thing 12:29:47 we have GELI and zfs native 12:29:54 and GEOM but don't 12:30:35 but a lot of neat boot stuff from linux doesn't exist, but also, it doesn't hurt as much that it doesn't exist. Mostly. 12:30:39 man loader, loader.conf, init, rc.d - etc, unlike Linux, all this stuff is plainly documented as part of the OS distribution. 12:31:42 I'm not even sure we can do FDE 12:32:04 and if you want the TPM involved you'll be writing it yourself 12:33:44 we might be able to do GELI encrypted root? If we do I bet the bootloader is prompting and then injecting the key into the kernel 12:35:41 is there a "gentoo" equivalent to bsd? where everything is source compiled? 12:36:10 otherwise you can encrypt some of your zfs partitions and then late-mount them, but again you're probably hand-writing anything that kicks off services that rely on them being mounted... or at least, that's how it looked last I looked 12:36:18 yes, it's BSD 12:36:22 avid: You can compile everything if you want to 12:36:45 ofc, but there a system for source compilation or is it LFS style 12:36:52 Gentoo's ports is modelled on BSD ports 12:37:07 yup, ports 12:37:31 or if you want to build the whole thing you can technically do it with port but really everyone uses poudriere 12:38:00 if you're curious, Absolute FreeBSD is a great book and it's how I got started. Do get a recent edition. 12:38:12 Ya I build my own ports with poudriere 12:38:48 but I wouldn't suggest coming into FreeBSD expecting it to be a drop-in replacement for a linux desktop 12:39:25 I hear that's a lot better for 15, but as a tool I'd say FreeBSD is what I'll reach for if I want a server 12:39:47 I wouldn't use FreeBSD on the desktop either (I use OpenBSD on that front). 12:40:34 I use it on desktop 12:40:49 My current view is: FreeBSD for servers, Void for linux Tinkering™, Fedora for "this is my main computer and I just want a working web browser and Steam" 12:41:58 well, I do want to try FreeBSD on desktop 12:42:49 as a rule I don't put files or my browser's cookie jar on a non-encrypted filesystem 12:43:15 and I'll want to run sway, so... 12:43:22 I expect getting myself to a FreeBSD desktop will be a A Journey 12:43:42 IMO FreeBSD is not optimized for the desktop, especially when it comes to security, and relative to the other options like OpenBSD. 12:43:45 You can use zfs encryption on your $HOME 12:44:03 I think there's maybe even pam modules to make that less painful? 12:44:14 which is probably better on desktop where you're using a password, and very annoying for ssh 12:46:27 I do need to remember that while it's neat to be super sekrit squirrel and have the best possible setup, tm, my real threat model is having my laptop nicked on a train and someone having a go at ordering things on ebay 12:48:13 which, as a side note, is why you want biometrics on your phone and a PIN on your banking app, because your threat model there is either (a) someone surfs your PIN, pinches your phone, unlocks it, possibly has a go at your online accounts or (b) someone mugs you and then holds your phone up to your face to get into your banking app 12:48:47 That's a double edged sword. I'm not in favor of it. But maybe more suitable for #freebsd-social 12:48:53 ah, good idea 12:49:02 I wasn't in there 12:49:27 well. I'm ADHD'ing my way out of doing my job anyway so I shall close IRC 12:49:50 * rtj just gives all the passwords to the dog. he can't tell anybody else but other dogs. they are not malicious. 12:50:50 rtj :D 12:50:55 zip thanks for the chat 12:50:56 dog -- the best password manager 12:52:28 I'm more than happy to refine my view on this, being right is a process, not my default state, after all :) 12:52:34 unix dog and I say good morning. hope that everbody has a good day. 12:52:41 dog is completely invincible to post-quantum cryptoanalysis 12:53:51 my mum did at some point apologetically tell me she has a password book. I told her it's great security: she's a retiree, not someone in an office that someone might break into, and having a different password for each service stored in a way that works for her is far better than using the same password or something that's hard to operate or might break 13:00:22 yeah realistically is there any research on how many accounts are broken into vs how many are lost forever because a person loses password? 13:15:33 well, given password recovery via email I suspect a good appproximation for that is how often that happens to email accounts 13:15:53 there's also the matter of people forgetting they have an account and creating five more on the same service 13:16:43 the upside of using an email address as an account identifier is that it doesn't happen, but then the downside is that you can't change your email address as easily... 18:15:04 huh. pkg is saying it has errors updating the sqlitedb for "packages.path" :-/ How screwed am I... 18:15:58 wavefunction: your system is fine, it's an issue with the repository 18:16:24 see PR 292214 18:27:15 https://github.com/freebsd/pkg/issues/2575 <- probably the same issue 18:27:20 it happens 18:34:32 Oh. Well then. Thanks ivy & mzar 18:36:27 Requiring people to "remember" a password is a recipe for all the problems we propose to solve 20:17:40 ah, came here to see if anyone had this pkg bug and the answer is right here. thanks. 20:18:19 markmcb: speak of the daemon, that's what I'm here for too. 20:20:39 markmcb: what's the solution? 20:21:04 dgriffi: I don't think there is one yet. Seems to be a wait and see situation. 20:21:33 markmcb: ah... you said "answer", not "solution". 20:21:45 Any idea what's going on? 20:22:36 It must have started since yesterday afternoon US Pacific time. I made a fresh install of 15 then. 20:23:26 that github link has a bit of info, but the root cause isn't clear. yeah, same for me, i just noticed it today for the first time. 20:24:00 on my side, there's no github reference in the error messages. 20:24:12 I did get this: pkg: Failed to fetch https://pkg.FreeBSD.org/FreeBSD:15:amd64/quarterly/All/Hashed/git-2.51.0~42542afbdd.pkg: Not found 20:24:32 I thought it was my local mirror barfing 20:24:49 I meant from mzar earlier in this channel, here it is again: https://github.com/freebsd/pkg/issues/2575 20:30:53 Three weeks??? 20:32:48 I've done updates on 14.x machines in the past three weeks without issue. 20:37:11 I posted a comment mentioning my recent install of 15 and doing a pkg update just now on a 14.3 box (no problems). 20:46:59 The answer is "wait for maintainers to fix it." The FreeBSD package builders seem to not be building the pkg dataset with the correct values emitted for 15 20:47:15 https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=292214 <- 20:47:34 last update was around 2 hours ago 21:25:09 Is anyone from the postmaster team available who can send me a private message?