*bandali gets busy sending a couple more emacs-xyz patches
<nckx>lfam, bandali: Then I probably have bad news. It will depend on what the disclosed vulnerability turns out to be, but Guix's opensmtp service still uses an OLD version because it hasn't been ported to the new syntax yet.
<nckx>If the vuln was introduced between 6.0.3p1 (Guix's service) and 6.6.2p1 (opensmtpd-next, which I run in prod but does not have a service) you're lucky. Otherwise someone will have to sweatily adapt our service now. It won't be me, sorry.
<pkill9>how difficult would it be to take an operating-system configuration, and modify it with guile?
<pkill9>i'm thinking it would be neat to have a command line interface that does common changes, like adding a user, and it would add the user to the config and then reconfigure the system with that config
<pkill9>i've heard lisp languages are good at doing things like that
<dctrud>Hello all. I think I read a while back about 'guix pull' taking a fair amount of RAM. Does anyone know what the worst case is?
<dctrud>... before I waste time with a RAM constrained VM
<terpri>pkill9, the difficult part would be serializing the changes in a useful way
<g_bor[m]>there are ready-made solutions to read and write os configs, and a modification is usually as easy as set-fields on the record.
<g_bor[m]>or use inherit and the operating-system constructor.
<apteryx>is there a way to skip the "Network selection" step in the graphical installer? I've just set a "shared" internet connection over a wire using NetworkManager, and have internet access on the box (ping 22.214.171.124 works, for example), but the installer refuses to recognizes that.
<drakonis>the installer generates an config i think?
<drakonis>you can back off that step to reach a list of available steps i think?
<apteryx>yes, I can go back to the installation menu, where "Network selection" is at the end of the list of possible actions (there's also an "Abort" button).
<leoprikler>raingloom: search-patches accepts patches at your channel root. If you put them elsewhere, you're on your own.
<ng0>ok, not that easy. but it's on a list which seems to have many people I try to avoid because I have a life, so getting an email with it was positively surprising and irritating (because i have just been uploading releases)
<smithras>the 'time of file birth' attribute isn't being scrubbed :/
<numerobis>Hi #guix! Noob question, but where in my directory tree is guix located (the place where the 'git pull' happens). I'm asking because, up until now, I always go online to view the git log, and it'd be more convenient to view it locally.
<efraim>I forgot to print out keysigning info for people registered to come
<daviwil>Anyone else here encounter a lot of unexpected errors after switching from "emacs" to "emacs-next"? Some parts of my config are exceeding both max-lisp-eval-depth and max-specpdl-size just by changing over to Emacs 27
<daviwil>I searched around online but didn't find anything conclusive about whether certain packages I use have issues with 27
<lfam>daviwil: Usually if there is a -next package it means that the new package isn't totally ready for prime-time
<lfam>I don't use Emacs so I don't know the specifics of this case
<daviwil>I would imagine Emacs 27 is pretty stable by this point, was mainly curious if anyone saw any issues specific to Guix's management of EMACSLOADPATH, etc
<daviwil>I'll probably just have to step through my config and enable things one by one until the culprit is discovered
<Blackbeard[m]>daviwil: I think you have to adapt your config and packages for emacs 27 a bit
***apteryx_ is now known as apteryx
<NieDzejkob>bricewge: it's possible to just set the env var when starting xss-lock
<NieDzejkob>leoprikler: see build-etc/profile in guix/build/profiles.scm
<leoprikler>I actually meant the real /etc/profile in gnu/system.scm
***apteryx is now known as apteryx_
***apteryx_ is now known as Apteryx
<leoprikler>It's fine, I already patched, what I wanted to patch. I'll wait until FOSDEM is over before submitting it, though, given that I already wait on two patchsets.
<NieDzejkob>is it not generated by that procedure anyway, though?
<leoprikler>I think you mean $GUIX_PROFILE/etc/profile whereas I mean $OS_ROOT/etc/profile.
<leoprikler>i.e. the thing that sources all the $GUIX_PROFILE/etc/profiles 😉️
<NieDzejkob>oh that's confusing. There's /run/current-system/etc/profile and /run/current-system/profile/etc/profile
<leoprikler>bricewge: Having both in literally the same file is a bit overkill.
<NieDzejkob>bricewge: I mean that putting your wrapper in a separate file and keeping both in a separate git repo is IMO a good solution
<bricewge>leoprikler: The thing is I need xss-lock to forkexec when starting as usual but one argument should be a guile wrapper. And passing my script as an argument seems a bad idea. That's why I was looking for (guix gexp) to get a path from my wrapper.
<bricewge>Okay thanks guys, I'll do it your way then. (:
<leoprikler>To be completely honest, I still don't understand your use case.
<NieDzejkob>he wants to set env vars (theme) based on, say, time
<apteryx>what target should be given to grub-configuration for a GPT table with a /dev/sda1 boot partition?
<apteryx>this old tower is like 13 years old. So it probably wants BIOS?
<nckx>apteryx: Hard to tell, UEFI had been out for a while by then but there were plenty of BIOS systems being sold (or uselessly bad UEFI implementations) IIRC.
<nckx>apteryx: It will definitely support ‘BIOS’ in any case.
<leoprikler>bricewge: If your wrapping code is simple enough, you could do something like (define (my-constructor . args) [code to determine env vars] (fork+exec-command ...))
<nckx>apteryx: If you want to use GPT+BIOS (which works fine) you'll have to create a ~1 MiB ‘BIOS boot partition’ for GRUB to live in. It won't be a mountable file system, you don't need to do anything with it, you just need to reserve a few million bits for GRUB.
<apteryx>last time it failed because the target machine is running the Guix installer (off a flash drive), and I had forgotten to start the cow-store service.
<apteryx>grub-install was complaining about None mounted on / or so.
<apteryx>it might still not like it: mount says: none on / type overlay
<nckx>Interesting. I'm almost certain I've forgot the cow-store before (knowing me, it's just a likely bet) and all I got was an error when RAM eventually filled up.
<apteryx>yeah, the RAM was filling, but that bare-bone installation seemed to fit in it :-)
<nckx>apteryx: What do you mean ‘might still not like it’? That mount line seems correct.
<nckx>As long as your real drive is mounted at /mnt, both guix system init & GRUB should DTRT.
<nckx>(Or wherever, even /mnt isn't a hard-coded requirement.)
<bricewge>leoprikler: Unfortunately it won't work in my case. xss-lock is a daemon that listen on X events and execute it's first positional argument when he want the screenlocker up. So xss-lock is a service that should never stop, but in my case I need him to call FOO=bar xsecurelock and some time FOO=baz xsecurelock. That's why I was looking to path a wrapper to xss-lock instead of xsecurelock directly, since stop and starting such a
<mfg>So i'm trying to update all necessary packages to get adb 9.0.0_r52. I'm currently trying to get it's liblog dependency built. But the make step says "nothing to be done for all", but it builds nothing and crashes afterwards because of obviously missing files ...
<apteryx>nckx: I was referring to a grub-installerror I saw yesterday, about apparently looking at teh mount for / and being unhappy that it found 'None'.
<nckx>But that's… isn't that always the case for / in the installer?
<nckx>Could you have specified a partition as the target instead of the device, causing GRUB to look for a mount point?
<apteryx>perhaps. I had tried /dev/sda1 as my last attempt. We'll see, now it's running with cow-store enabled and /dev/sda as a target.
<NieDzejkob>mfg: Could you post a link to the relevant Makefile?
*NieDzejkob kinda wants to punch whoever wrote the docs for guile-git in the face
<jackhill>+1. I took the plunge after my disk failed, and I had to install something. I'm a former gentoo user, coming by way of debian, and it has a lot of the nice things I miss about gentoo, but with so much more awesome stuff.
<drakonis>shepherd's docs are a little inadequate right now
<bricewge>leoprikler: It would work, but I would need to restart xss-lock at some time.
<leoprikler>so you would need to modify the env vars of the xss-lock process?
<bricewge>Anyway I'm near done with the wrapper. Just need to understand how to get from ("foo" "bar") to "foo" "bar".
<drakonis>shepherd's start up process seems a little slow though, it can sometimes take a while to actually construct the system to a point that logging in functions
<leoprikler>starting xsecurelock with some var alone is not enough?
<drakonis>ie: user data hasn't been made available in their correct locations so it says it belongs to the guest user
<drakonis>hrm, is there an easy-ish way to switch out which cups version is available to the service?
<drakonis>my printer only functions with an older version of cups from years ago
<bricewge>leoprikler: Yes modifying the env vars of the running xss-lock process would be enought, but I have been told it's not the UNIX way.
<bandali>y’all, anyone know the status of texlive packages in guix?
<leoprikler>hacking into the environment variables of a running process sounds very evil
<bricewge>It is engouh to start xsecurelock with some var, the issue is that they ought to change.
<bandali>i remember seeing related discussion on one of the guix lists a while back, but can’t remember off top of my head what the recommendation is for installing and using texlive packages
<leoprikler>bricewge: which should not be an issue if you make a separate action for that
<leoprikler>either one for each set of variables, or one + logic to determine which variable set you need
<apteryx>nckx: had forgotten the dependencies field, so yes, something was smelly there indeed.
<apteryx>now it still fails, but with: guix/scripts/deploy.scm:115:22: Throw to key `srfi-34' with args `(#<condition &store-protocol-error [message: "creating directory `/gnu/store/nix-18347-0': No such file or directory" status: 42] 7f8b9cd149f0>)'.