IRC channel logs


back to list of logs

<nckx>This is the first time in my life I've encountered SIGXCPU.
<lfam>Never even heard of it
<nckx>I wonder if it's legit or just the misfiring synapses of a dead daemon.
<nckx>I've restarted it. Three of the ‘In progress’ evaluations now actually have build numbers, the rest is ‘failed’.
<nckx>Talking about
<nckx> is an absolute bloodbath but I don't know if that's new.
<lfam>core-updates doesn't have to work
<nckx>But nor should it be left in a non-evaluating state for months.
<nckx>Assuming that's the case which isn't certain at all.
<terpri>hm, no obvious way to enable luks trim support afaict
<vagrantc>does full disk encryption still requires luks format v1 these days or has grub learned how to read luks v2?
<nckx>terpri: You could add an allow-discards? option, with a big fat warning.
<terpri>maybe with a 'make-luks-device-mapping' function or something (that allows you to pass random options)
<nckx>vagrantc: Last I read the GRUB support, even with the latest patches, was incomplete.
<terpri>vagrantc, yes, but there's no obvious (to me) place to put the option
<terpri>not that important atm, i only want to be able to run fstrim eventually, not even necessarily from guix (debian livecd would probably work just as well)
<terpri>i noticed that my new ssd is locked into 512-byte (or whatever, not 4 KiB) emulation mode, so in theory i could make a 32 MiB ESP partition. probably not ideal though
<terpri>is grub2 really forcing luks v1 still?
<nckx>That is my understanding. The citiation is in my mailbox, which is off-line for maintenance.
<terpri>hm, yeah, version 1 according to "cryptsetup luksDump"
<terpri> but...
<nckx>Yeah, & I recently read a mail saying ‘akshually it's still incomplete’. I didn't read it closely. I would now 🙂
<terpri>only PBKDF2 supported, because libgcrypt 🤦
<nckx>terpri: Our cryptsetup defaults to LUKS1 because I deliberately pass --with-default-luks-format=LUKS1 for now.
<nckx>That's not going anywhere until it's been well-tested, but it's not because cryptsetup is querying GRUB or anything. You could experiment if you were feeling brave.
<nckx>And use a git GRUB.
<nckx>I presume.
<terpri>i livepatched lvm support into guix during installation, maybe i'll feel similarly brave soon :p
<terpri>but i'm supposed to play in a bzflag tournament in a few hours, so it's probably a bad time to potentially break my entire system
<nckx>I also presume this is the mail you're referring to?
<terpri>i must have done something similar (with the luks format) during installation, but forgot
<terpri>maybe i should just switch to unencrypted /boot
<nckx>ret = grub_error (GRUB_ERR_BAD_ARGUMENT, "Argon2 not supported");
<terpri>granted then i may not get so many comments like "Terpri youre badass man, like opsec to the max", but life would a little easier
<terpri>quick, write your own argon2 implementation asap! nothing could possibly go wrong
<terpri>a noble KDF for a gnoble operating system
*nckx giggles.
<terpri>does PBKDF2 vs Argon-foo really matter (to people not being targeted by the NSA or FSB)? i haven't been keeping up with modern crypto that well
<terpri>pbkdf2 is relatively new at least, and has an (informational) rfc
<nckx>I don't pretend to know.
<terpri>"One weakness of PBKDF2 is that while its number of iterations can be adjusted to make it take an arbitrarily large amount of computing time, it can be implemented with a small circuit and very little RAM, which makes brute-force attacks using application-specific integrated circuits or graphics processing units relatively cheap."
<terpri>ah, argon2 was specifically designed as a successor, it sounds like
<terpri>so, better pick a really good passphrase, i guess ;)
<nckx>I think it tries to defend against GPU attacks too.
<nckx>Whether or not that's just a buzzword I don't have the chops to say.
<nckx>Did you mean to call PBKDF2 ‘new’? That made me do a double take. I feel like it's been around ‘forever’ now.
<terpri>no, i got confused for a minute by the RFC date (2018)
<nckx>RFC2898 says 2000.
<terpri>but the contest to replace it was in 2013, so it must be ancient
<nckx>So at least it's Y2K-proof.
<terpri>almost old enough to drink
<nckx>‘PBKDF2 - I drink to forget.’
<nckx>Do you play bzflag on Guix? Had to look up what it was, and lo, we have it.
<terpri>well, libgcrypt does at least (appear to have) have blake2, which argon seems to be based on. so maybe i'll code it in the most naïve way possible (i've done simple implementations of aes, etc.)
<terpri>and then everyone will make fun of me for missing basic but non-obvious issues
<terpri>stand-up cryptography
<terpri>i tried it once and found it unplayable on the public servers
<nckx>Well, I guess for all its flaws home-grown crypto does have the element of surprise...
<nckx>Unplayable how?
<terpri>but some of my local friends have been doing game nights (mostly board games -- pioneers, dominion, etc.) in lieu of irl parties, which also conveniently helps test jitsi as a future service
<terpri>either people using bots, or only world-class players still bothering to play it, or both
<terpri>un-fun rather than unplayable, really
<nckx>Oh, not Guix's specifically, gotya.
<nckx>HCoop is cool.
<terpri>a while ago i was tasked with installing pioneers on a friend's computer so they could join in; it was an ancient m*cos system so i assumed i could just use homebrew. nope, too ancient to build successfully, similarly with they have to play pioneers on wine on m*cos :p (yes, they'll be getting a gnu/linux box once they can afford parts with unemployment $, the iMac was just an old corporate system that
<terpri>was headed to the dumpster)
<terpri>it'll be cooler when we have full guix support :p
<terpri>(jackhill -- who i think hangs out here usually -- is a sysadmin, the other two don't know how awesome guix is yet)
<terpri>relatedly, expect to see some patches for openafs support in guix sometime...
<nckx>(I had no idea. Neat.)
<nckx>\o/ patches.
<nckx>First real-world use of AFS I'm aware of too.
<brendyyn>just wondering if anyone here has a made a guix package for gccemacs. it would be fun to play with. If you don't know what it is, its an experimental version of emacs with native compilation that is said to load up and run a lot faster.
<terpri>it's still commonly used at universities, though not always obviously so (e.g. both UNC and NCSU, near me, use it). also some bank (citi?) uses it for some reason
<terpri>but is the only public-access afs system i know of
<terpri>brendyyn, not that i've heard of, but i'm eager to compare it to guile-emacs once the latter is updated
<nckx>Don't use pastebins in e-mail: a lesson.
*nckx pings Formbi.
<brendyyn>hmm ok so there is some interest
<terpri>will be interesting to see whether AOT or JIT compilation works better with typical emacs usage
*nckx 's uni was a total Windows box...
***catonano_ is now known as catonano
<Formbi>nckx: hmm?
<sneek>Formbi, you have 1 message!
<sneek>Formbi, quiliro says: Mother what?
<brendyyn>im pretty sure guile emacs is very slow, and emacs developers seem unintersted in using guile.
<Formbi>brendyyn: I couldn't get gccemacs to work properly
<terpri>it is very slow, mainly because i was aiming for 100% compatibility before doing any kind of optimization
<brendyyn>terpri: are you a developer of guile-emacs?
<Formbi>it did compile Emacs itself, but there were some settings that I couldn't get right and which prevented the compiler from compiling other things
<terpri>brendyyn, the primary developer, yes
<terpri>or i should say "was", as it's been dormant for years (getting BigInt into firefox was more profitable...)
<brendyyn>terpri: That's amazing! *kowtows*
<Formbi>I updated it recently and some stupid check fails even though it shouldn't
<Formbi>guile-emacs is very slow and it also crashes a lot
<brendyyn>I wish I was a better programmer. Everytime I try to work on anything slightly difficult it just fries my brain cells and I don't get very far. I think I'm not cut out for it.
<terpri>it is very slow, but shouldn't crash a lot... :(
<Formbi>well, it did for me
<Formbi>(with -q too)
<str1ngs>terpri: not technically emacs, but I've been working on Nomad which is a emacs like browser using guile.
<terpri>Formbi, if you have reproducible examples, please email them to me (, or the guile bugtracker (i guess, haven't asked permission to use the emacs bugtracker)
<str1ngs>terpri: though it uncanny how much like emacs it is sometimes.
<brendyyn>terpri: I'm not an expert at all so I was curious; Do you think if emacs doesn't get ported to guile, it would still be realistic to develop the elisp implementation on guile and turn useful elisp emacs extensions in to guile libraries for use outside emacs?
*nckx also tips hat to terpri, also had no idea.
<terpri>str1ngs, neat, is there a public repo?
<Formbi>terpri: it was crashing as soon as I was opening things like eshell or dired
<terpri>Formbi, on what os?
<Formbi>BTW would it be possible to move Guile Emacs to Guile 3?
<terpri>that's the plan
<Formbi>Arch, I think
<Formbi>I installed Guile Emacs with Guix though
<terpri>ok, i'll make a note to try that, maybe something broken with external processes (though gnus, etc. worked)
<terpri>str1ngs, sounds a little bit like the next browser from the brief description (but that one is CL-based i think)
<str1ngs>terpri: yep, . devel is the current development branch. most of the emacs features are done using emacsy a guile library. and the GTK aspects use g-golf a gobject introspection library for guile.
<terpri>brendyyn, it would be less useful, but there could be some niche applications (say, using org-mode stuff from within guile)
<str1ngs>terpri: next CL based yep. I do at time ponder how things would be done if emacs was written in guile. side effect of working on Nomad I guess.
<Formbi>it's very unpleasant that all these Emacs-like thingies don't support C-m as RET
<str1ngs>Formbi: what should C-m RET do?
<terpri>also, guile-elisp will be the basis of (well, more like inspiration for) a full CL implementation for guile, which will be more generally useful, so not a waste of effort imo even if the emacs maintainers veto it
<Formbi>when you do C-m in Emacs, it's equivalent to pressing RET
<brendyyn>CL = Common Lisp?
<Formbi>it doesn't work in Next and I don't think it does in Nomad
<Formbi>terpri: :O
<Formbi>interesting thing
<terpri>my top-secret plan is to use guile-emacs to unify elisp and cl as much as possible (as they're quite closely related already)
<brendyyn>I'm using nix on guix but getting this error of nix not being able to find bash
<brendyyn>what is useful about it? 0.o this stuffs way over my head
<Formbi>terpri: hehe
<Formbi>Scheme/Emacs/Common Lisp
<terpri>one lisp to rule them all, one editor to combine them
<terpri>one guix to build them all and in the bootstrap compile them
<str1ngs>Formbi: pretty easy to add to Nomad in ~/.nomad with (define-key global-map (kbd "C-m") (lambda _ (insert #\newline)))
<Formbi>ok, thanks
<Formbi>I wanted to try it out, but it's not working :/
<str1ngs>for some reason 'newline does not exist in Emacsy I should add that.
<terpri>str1ngs, oh, that's really neat, i was a fellow gsoc intern with the emacsy developer. hadn't followed its development after that summer
<Formbi>(process:10983): GLib-GObject-CRITICAL **: 02:36:41.185: g_type_add_interface_static: assertion 'G_TYPE_IS_INSTANTIATABLE (instance_type)' failed
<str1ngs>terpri: Emacsy is well designed, right now i'm kinda maintaining it. with the help of others @
<nckx>terpri: Damn, I expected you to end that with an evil mode binding reference.
<str1ngs>terpri: err
<str1ngs>Formbi: you are using a resent guix pull?
<Formbi>from yesterday, I guess
<str1ngs>Formbi: guix show nomad should be @ 0.2.0-alpha
<terpri>nckx, :D
<Formbi>it's 0.1.2
<str1ngs>Formbi: 0.1.2 was mainly written in C and scheme 0.2.0-alpha is mostly scheme now. also it's still alpha so expect some issues.
<str1ngs>Formbi: what dos your guix describe show?
<Formbi>wait a sec, I'm doing a pull
<terpri>lisp poetry is a very small genre these days
<terpri>i keep a copy of this blake riff (by dsk) on my desk: "I will not cease from Mental Fight, / Nor shall my Sword sleep in my hand: / Till we have built the Lisp Machine, / in Linux' green & pleasant Land"
<nckx>terpri: Is that a ‘fortune’? It looks recently familiar.
<brendyyn>Any idea when the next big core-updates update and merge will occur?
<nckx>Is ‘soon’ enough?
<Formbi>IMO a properly gccemacs with the blocking things replaced would be a nice basis for a Lisp-centered setup
<terpri>nckx, i quoted it here a few months ago, apparently
<Formbi>having several emacs-like things isn't so great
<nckx>Ah, that'd do.
<Formbi>str1ngs: hmm, it still shows 0.1.2
<brendyyn>shows 0.2.0-alpha for me
<terpri>Formbi, well, we no longer have (s)xemacs really, but guile-emacs, gccemacs and remacs all exist
<Formbi>I mentioned gccemacs specifically because it's quicker
<brendyyn>are you updating to the very latest guix or do you have some channels.scm hack that makes it update to the latest one with substitutes?
<Formbi>so making big things like browsers would be more feasible with gccemacs
<terpri>i'm skeptical -- frankly remacs is a faster path to "browser-like" things -- but who knows
<Formbi>maybe if they would be written in Rust
<Formbi>but that wouldn't be very lisp-machine-ish
<terpri>for the record, guile-emacs came first by far, so i'm not to blame for the duplication of effort in this case :p (although i haven't been a good maintainer lately)
<raghavgururajan>Hello Guix!
<terpri>i'll have to take a closer look at gccemacs sometime, but i'm skeptical that it would be notably better than guile-emacs for browser-like programs, though, if it is what it sounds like (assuming i can speed up guile-emacs significantly, which is "just" a SMOP that will probably involve rewriting some of the emacs C innards in guile...)
<terpri>(it's also possible that gccemacs and guile-emacs could work together in some fashion, but again, i'd have to learn more about gccemacs to figure that out -- guile doesn't really do native AOT compilation yet afaik)
<terpri>(though it does perform AOT compilation to fairly low-level bytecode)
<terpri>thanks Formbi, i've poked around the git branch but hadn't read the docs yet
***terpri_ is now known as terpri
<terpri>well, i only have a few tasks left before i can focus on larger things -- some small projects for $client, navigating the unemployment bureaucracy, moving house, etc.
<terpri>then i suppose we'll find out whether i still have the strength to push this million-line ball of mud over the hill
<nckx>Ohai raghavgururajan.
<str1ngs>terpri: I'm still interested in guile-emacs myself I think it has great potential. using guile's compiler tower would be great for emacs IMHO.
<apteryx>terpri: did you say you lived patch lvm support at install time?
<terpri>apteryx, yes, guix lacked lvm and btrfs subvolume support back when i installed it
<apteryx>are you still using LVM?
<apteryx>in case you haven't already done so, it'd be nice to contribute those patches, if they add something more than
*terpri reads
<apteryx>would anyone have an idea of how to slip a (canonical-package coreutils) reference into a Gexp? I tried the naive #$(canonical-package coreutils), but that failed with a confused backtrace.
<apteryx>this is in a snippet
<apteryx>which is defined using a Gexp
<apteryx>here's the snippet I'm working on:
***Guest93955 is now known as KE0VVT
<paul__>I wonder if folks have seen this problem before: i'm trying to `sudo guix system reconfigure` and it starts working, but fails halfway through the download of nss-certs-3.52.1 with a `apply-smob/1` backtrace, mentioning likely network issues. If i curl the same url it appears to work fine.
<paul__>It repeatedly fails at the same point in the download, whereas curl manages to download it.
<paul__>Hm, it appears that doing `guix build nss-certs --fallback` and then trying `sudo guix system reconfigure` again has helped me past that blockade.
<pkill9>why does downloading substitutes from the build server ci.guix... max out at 3mbps
<brendyyn>mine goes at around 34KiB/s, consider yourself lucky
<pkill9>well, i'm downloading via a VPS
<pkill9>i get less than 3mbps at home
<pkill9>but the vps is a lot faster
<paul__>pkill9: FWIW yesterday i set up a caching CDN in front of it for my own use. Much faster.
<pkill9>what do you mean in front of it?
<pkill9>like a personal proxy?
<paul__>I'm guessing lives in the USA somewhere, so having a geographically closer cache is also super helpful. I'm not sure if it'll end up being expensive, otherwise i'd be happy to share.
<paul__>pkill9: I mean it's a transparent caching proxy so i point Guix at it, and it pulls from
<nckx>paul__: Did you save the backtrace?
<paul__>It's CloudFront - at the risk of being kicked out here ;)
<nckx>ci.guix is in Berlin.
<paul__>nckx: ok.
<nckx>Transit to the US is known-horrible ☹
<paul__>nckx: I did not - it's in a VM window, and copy pasting is super awkward. Apologies.
<pkill9>paul__: if you could share that then that would be cool, but no worries if you don't want to
<paul__>nckx: I'm even further than that, imagine ;)
<pkill9>i'm building every available package on my vps
<pkill9>(building meaning downloading also)
<pkill9>to get .desktop files
<nckx>paul__: I didn't mean to accuse you of being an USian, just that that's been confirmed to be particularly poop over there.
<paul__>nckx: No offense taken P)
<paul__>* :)
<nckx>paul__: So since your ‘personal CDN’ can only be reactive -- start downloading when you request a URL -- does that mean it's connecting to ci.guix from elsewhere, say Europe, effectively replacing our connection to you with a better one?
<nckx>That's clever.
<paul__>I believe that's what's happening.
<paul__>I initially just hoped for the geo-proximate caching to help me, but indeed throughput looks much better too, anecdotally.
<paul__>Although that last bit surprises me a bit. I'd be curious to know where, from's point of view, it looks like i'm connecting from.
<nckx>paul__: We've actually used CloudFront ourselves in the past. No kicking will be done.
<paul__>Ah, i'm surprised on both counts ;)
<nckx>We might've held our noses. It was expensive though, so we stopped. Throwing lots of money away instead of fixing the obvious bug is silly. Problem is, nobody has a clue what the bug is.
<paul__>I see. When you say 'bug', you mean there's a legit problem with network traffic?
<paul__>As in, it's not just slow because of low bandwidth?
<nckx>Downloading from Germany I get ~30 MiB/s reliably.
<paul__>And you're not in Germany?
<nckx>So it's not ‘our’ bandwidth anyway.
<nckx>I have a VPS there.
<paul__>I see. I guess it could be any peering connection in between.
<paul__>FWIW i was surprised `traceroute` didn't manage to do anything against
<paul__>nckx: Out of curiosity, do you use this VPS to host a Hydra you control, to have quicker package installs or something?
<vagrantc>i'm curious if the fsf would consider hosting a caching proxy in the US ...
<jackhill>sneek: later tell terpri I have a working kafs-client package. The services need more work:
<pkill9>hmm, i don't suppose there be a way to get a list of all packages that have been built by the build servers?
<pkill9>guix weather finds out percentage of packages built, so there must be a way
***paul__ is now known as toothbrush
<pkill9>yea there's a bunch of code
<pkill9>but it's 5am
<nckx>toothbrush: The firewall at the institute blocks ICMP.
<nckx>toothbrush: <VPS> That's the short gist of it.
<nckx>The public VPS is a caching nginx proxy. The build farm is at my home.
<nckx>In my Nix days it ran Hydra, now it runs a bash monstrosity tailored to my every quirk.
*nckx needs sleep, good night #guix.
*toothbrush waves
<str1ngs>pkill9: guix weather can take a manifest with the -m flag. don't know if that helps.
<Kimapr>Hello, i cant boot guix system again
<brendyyn>Why does inherit not inherit the package name?
<Kimapr>I can't guix pull on stable guix installation image
<Kimapr>It gets stuck at 'Updating channel 'guix' at <...>...'
<Kimapr>Also does stable guix support btrfs subvolumes?
<Kimapr>Worked around it by cloning the repo manually with --depth=1 and instructing guix pull to pull from local clone
<efraim>hmm, suddenly my gpg-agent isn't working on my desktop, didn't change any configs for it
<efraim>says it can't find pinentry
<guix-vits>Hi sneak bot!
<efraim>sneek: botsnack
<guix-vits>sneek: \o/\o/
<Kimapr>I'm confused. Is sneek bot or not?
<guix-vits>sneek: seen sneek?
<sneek>I think I remember s34n in #guix 2 months ago, saying: (I think. I can't look at history right now and I'm brand-new to guix).
<guix-vits>sneek: seen that bag of happiness, guix-vits?
<sneek>Sorry, haven't seem 'em.
<guix-vits>But he is there!
<guix-vits>OK: I did updated Guix System on the RockPro64 SBC. And all worked well. Then i added %desktop-services to config.scm, and get stuck on boot (but not in rescue shell). The error was about some filesystem called "freezer". I think i'll fall back to Armbian on the SBC. Am not ready to this fun yet being wanting an working SBC, now :)
<guix-vits>I think the culprit is elogind (adding just this service maked the System unbootable).
<guix-vits>Also iwd didn't builded (failed on make check for me and nly).
<guix-vits>str1ngs: ^^ (Best wishes)
<efraim>it seems something changed between my reconfigure a week ago and now. the config change I made was to add the qemu binfmt service
<str1ngs>guix-vits did you save your config for rockpro64?
<guix-vits>str1ngs: Yes
<str1ngs>guix-vits: do you mind pushing to a git repo? also how did you boot off of emmc or sdcard? and how did you install the rootfs?
<guix-vits>str1ngs: one sec, there were some instruction i'd followed on mailing list..
<str1ngs>guix-vits: it looks like you just partitioned emmc then installed to a partition. other then that looks like a straight forward config.
<str1ngs>guix-vits: pretty straight forward. thanks for the details.
<guix-vits>short: Have some system with Guix on it (for `guix system init`)
<guix-vits> make a GPT partitioned SD || eMMC, with one partition (start from 32768, ext4)
<guix-vits> mount, create a config.scm, `guix system init`.
<guix-vits>str1ngs: BUT!
<guix-vits>Use UUID or label for the /. That should be enough.
<str1ngs>theoretically it might be possible to just use guix system build and create disk-image and dd it after.
<guix-vits>str1ngs: elogind-service *possibly* makes system not boot. And nor gdm, nor sddm, nor slim worked __for me__, also.
<guix-vits>If the 5.8 kernels will make the graphics distorted, try 4.4 or 4.19 (not sure now).
<guix-vits>str1ngs: And not linux-libre-arm64-generic kernel (linux-libre-4.19) wasn't bootable too.
<str1ngs>guix-vits: okay, I have not received my rockpro yet will test it out once I get it.
<guix-vits>*linux-libre-4.19 was unbootable
<guix-vits>*linux-libre-arm64-generic was bootable
<guix-vits>str1ngs: In the worst case there is Armbian, which work OK (if not count that it isn't Guix). And installed with Guixp PM apps have no GPU acceleration (scrolling a man-page, within a graphical session, in terminal emulator takes seconds).
<str1ngs>guix-vits: is armbian based on bionic of focal do you know?
<guix-vits>str1ngs: You can choose
<str1ngs>gotcha thanks
<guix-vits>str1ngs: last time i'd check'd the /etc/apt/sources, there were http_NOT_S by default, maybe it's changed now.
<str1ngs>guix-vits: right that's pretty easy to fix
<guix-vits><please enter a joke about Debian and stability, then press Return>
<str1ngs>this is a lot of pressure to be funny first thing in the morning! :P
*str1ngs sips his Americano
<str1ngs>guix-vits: looks like you made some headway with rockproc64, thanks for the info.
<str1ngs>guix-vits: also I need to do all of this when my pinephone arrives. that should be intresting.
<nckx>Morning everyguix.
<guix-vits>Hello there.
<nckx>brendyyn: But it does.
<brendyyn>when i removed the explicit (name ...) i would get a syntax error
<brendyyn>It would be great if the crate importer included the versions for all the inputs
<brendyyn>just about every crate could be fully automatically imported then
<nckx>Adding them all (after looking them up) by hand is annoying.
<brendyyn>yeah im looking at the importer code, and its a bit strenous on my brain cells to figure out how to do it, but i think i could get there eventually
*nckx wonders if tackles that too.
<brendyyn>ooh awesome. i had search "rust" but not "crate" and didn't see that one
<nckx>I knew there was a ‘semver importer’ floating around so searched for that, probably wouldn't have found it otherwise.
<brendyyn>i have a habit of trying to make things that it turns out someone else has almost already done
<brendyyn>i also spent a couple hours crafting crates, but anyway
<str1ngs>sneek: later tell nly, I purged merged branches on Savannah. If there are any missing branches you need on Savannah can you push them back. Encase I missed something.
<str1ngs>sneek: later tell nly, feature-release has been renamed to release btw.
<sneek>Will do.
<raghavgururajan>Hello Guix!
<raghavgururajan>nckx: Oh good you are here.
<raghavgururajan>nckx: How do you initiate a `./pre-inst-env guix buid foo` on bayfront, that does not terminate when ssh disconnects?
<raghavgururajan>guix-vits o/
<brendyyn>raghavgururajan: probably the screen tool is available?
<raghavgururajan>nckx: A while ago, leoprikler told me to do, after starting the build, [1] Ctrl+Z [2] bg [3] disown -ar. But that does not work. When I reconnect after few hours, the package was not built.
<leoprikler>you should probably build in screen tho ;)
<nckx>raghavgururajan: I use screen.
<raghavgururajan>brendyyn: Hmm. I thought of using GNU Screen or something like that. But I haven't used those kinda stuff before. So gonna take a while to learn.
*raghavgururajan is getting screened
<brendyyn>raghavgururajan: in your case you just run screen, run the command with `;exit` on the end, then i think Ctrl-d detaches you
<raghavgururajan>May be, quick and dirty command please :-D
<nckx>raghavgururajan: If screen is too much to learn right now look at mosh, it's an SSH session that survives disconnection. As long as you don't kill it locally it should keep running on the server. Needs to be installed on both ends.
<nckx>But learning screen (or dtach/tmux, whatever you prefer) is a better investment that learning disown/nohup incantations by heart.
<raghavgururajan>Okay, so [1] `screen` [2] `./pre-inst-env guix build foo ;exit [3] `Ctrl+D`
<raghavgururajan>Is that correct?
<efraim>curl fails to link with libssh2 on mips64el in guix-0.13
<nckx>raghavgururajan: C-a d.
<efraim>I connect to bayfront with 'ssh bayfront -t screen -RD' the 'screen -RD' is really-really detach and/or create a new screen session for me if there's isn't one
*raghavgururajan got frightened about screen and looks into bosh
<raghavgururajan>Is mariadb failing on master
<raghavgururajan>After to rebase of wip-desktop to master, mariadb fails to build.
<nckx>mosh will sufficiently protect against accidental disconnection.
<nckx>mariadb@10.1.45 from 825b2d9 builds fine.
<raghavgururajan>Tried building 6 six times with different core options
<raghavgururajan>nckx: Would you be able to copy the susbtitute from to batfront please?
<raghavgururajan>Test timeout
<raghavgururajan>Just one
<raghavgururajan>Failure: Failed 1/500 tests, 99.80% were successful.
<raghavgururajan>Failing test(s): binlog_encryption.encrypted_slave
<raghavgururajan>nckx: ec2f375c90c58fba9a8708d71b2efe9aa985c12e at wip-desktop
*nckx deleted their bootstrapped guix-wip-desktop clone too soon. Sure. No idea how long it will take.
<raghavgururajan>nckx: Thanks! About 20min with `--cores=24 --max-jobs=1`
<schonlank>Hello, I am a new Guix SD user with the following problem: I don't know how to suspend my computer in Guix.
<brendyyn>sudo loginctl suspend
<brendyyn>personally i configure my power button to suspend
*raghavgururajan just closes his lid :-P
<brendyyn>i cant remember the last time i did that
<schonlank>Okay wow that worked. Thanks! How could I have learned it from the documentation?
<schonlank>I mean: I searched the manual for suspend, I looked through all my %desktop-services, and I found no clues.
<brendyyn>Perhaps it is missing and should be added. It comes from the fact that guix uses elogind, which comes from systemd
<nckx>I don't think it's documented in the manual since it's not a Guix-specific thing.
*nckx uses /sys/power/state 4 lyfe.
<brendyyn>hmm my nix on guix is broken now im getting error: opening lock file '/nix/store/jb7sxp3n9pcyc37j0j7na8a2356yc1h0-nixexprs.tar.xz.lock': Permission denied
<brendyyn>maybe sudo is required but i never needed it in the past
<peanutbutterandc>Hey guix
<guix-vits>nckx: Is there any way to return to OS after `echo freeze > /sys/power/state'?
<nckx>brendyyn: That's the (Nix) daemon's job, and the (Nix) daemon should already be running as root.
<brendyyn>i only just added the nix service to my configuration. i never used that in the past and was still able to install nix packages as my user
<brendyyn>but recently nix broke so i added it, and now nix needs to be used with sudo apparently
*nckx makes the o_O face but hasn't used Nix for years and never on Guix.
<nckx>guix-vits: I'm juggling too much state to comfortably test that right now but I'd expect it to just work. What do you mean by ‘return to OS’? What happens?
*pkill9 is getting all the desktop files
<nckx>guix-vits: Does ‘mem’ work? (Don't try ‘disk’ yet.)
<guix-vits>nckx: on this laptop mem work. 'freeze' makes the machine "hang". IDK how to return to the session (WM, emacs, anything).
<guix-vits>I wonder what freeze is, and where it can be used.
<guix-vits>Good bue.
<nckx>raghavgururajan: Yeah, but it's starting from cmake, so we'll see. Slow build is better than no build.
<bluekeys>Hi guix. Is it possible to make guix output a console to serial on boot? So, as long as the server is booted and I can get a serial cable to it then I'm able to access a shell. Ssh went weird for the last few days and I don't really know why, but a serial console is now feeling like a good idea as a backup.
<bluekeys>At the moment, after booting the serial console is just stuck saying "Welcome to GRUB!". I'd love to be able to see grub in my serial console too.
<bluekeys>When I have a monitor connected, I can control the grub screen via the serial connection, but just don't see any feedback *except welcome to grub*. Once the system is booted, I just get left with the welcome message, but would love to see the system booting, as if I was watching the screen, and then a console, as if I was connected via ssh
<bluekeys>And... I think the problem with ssh may be that the system is suspending itself. I've installed a gui, but don't want to use it. Today is the first time ive reconnected the monitor and It's just popped up a message saying Power Automatic suspend Computer will suspend very soon... I gues that was the cause of the ssh hangups
<apteryx>bluekeys: about SSH, highly likely!
<bluekeys>I think I'll remove the WM because I'm not really using it, I hope that would stop the suspend, I don't remember it happening before that.
<apteryx>I think another user was bitten by the same situation. For headless systems, seems safer to not install desktop environments.
<guix-vits>bluekeys: Is there any USB involved in the connection? Like USB-modem? Linux loves to autosuspend those.
<sneek>Welcome back guix-vits, you have 1 message!
<sneek>guix-vits, nckx says: ‘freeze’ is Suspend-To-Idle, ‘mem’ is Suspend-To-RAM. S2I is *less* deep than S2R and relies less on hardware support; it's basically the kernel saying ‘OK, I'll put the CPU in deep power-save mode & just... stop doing stuff for a while’ but the system is technically ‘running’. S2R actually powers down the CPU and lets the motherboard refresh RAM. So it's at least strange that it shouldn't work where mem
<bluekeys>I'd like to make use of the screen, as I can see it from my desk. I would like feedback, like current IP, cpu, other metrics etc...
<bluekeys>No usb modem, there is a serial to usb connection, don't hink that would affect ssh
<bluekeys>99% it's the wm hibernating as the screen has now gone blank and although the server is making noise, it's not responding to pings etc and it was before. This also perfectly coincided with the warning message and the screen blanking so....
<bluekeys>Anyone got any ideas on making use of the screen without a window manager? Maybe an autologin to a webbrowser with metrics. Maybe somthing with stumpwm, I could have a whole lisp machine thing going on?
<guix-vits>bluekeys: is SSH works with Wayland? sway seems to not autosuspend.
<bluekeys>I don't understand whay you mean guix-vits. You may have to dumb it down a little ;)
<bluekeys>*a lot!
<guix-vits>bluekeys: You can try to remove the '(screen-locker-service slock)' and '(screen-locker-service xlockmore "xlock")' from %desktop-services. I doubt that any WM do autosuspend on it's own.
<guix-vits>.. or i can step out, to not flood. IDK SSH, anyway.
***terpri__ is now known as terpri
<bluekeys>I think you might be right guix-vits. No need to step out, I appreciate the help. I'm really new to this and don't know scheme. How do I remove elements from %desktop-services? No harm in trying.
<guix-vits>bluekeys: I do not remember, there is the whole thing:
<bluekeys>Thanks guix-vits. I'm gonna grab some lunch and read that. The server just suspended in the middle of a system reconfigure, so I'm going to have to figure this out for my sanity. Bye for now.
<guix-vits>bluekeys: 8.1, System services has some with "(remove"
<guix-vits>bluekeys: Also in "Scheme Variable: slim-service-type" is an example of how to use %desktop-services, but swap gdm with slim.
<pkill9>web applications should be an oxymoron
<nckx>Aren't they?
<nckx>raghavgururajan: Great success.
<nckx>Don't forget to create a GC root.
<nckx>raghavgururajan: Which exact test (‘Failing test(s): xxx’ in the log) failed for you?
<voidie>Hey I have recently heard about the guix package manager and it seems interesting. I just have some questions concerning it, if anyone can help me out
<voidie>My first question is: I'm using musl Void Linux. I know that a lot of software is compiled against glibc. If I install a software with guix, will it install glibc as a dependency for that software?
<voidie>My follow
<voidie>My follow-up question is: what if I install multiple software which have the same dependencies? Will guix pull that (version of) the dependency for each package that I install? So let's say I install package A, which requires glibc. Then I assume guix will install glibc as a dependency for it. But when I install another package (package B), which also requires glibc, will I end up with glibc twice on my system now; one as part of each package?
<voidie>Thanks in advance for the help!
<vagrantc>when they use the same glibc, there will only be one instance
<vagrantc>in some cases, different packages may use different versions of libraries
<voidie>I see. That's what I wanted to know. Thanks!
<vagrantc>and guix packages use libraries from guix, not from the host distribution (e.g. void linux)
<voidie>Ah, so I will probably end up with 'duplicates' in some way or the other
<vagrantc>each time you update guix, you'll likely get some new packages and libraries ... and interactions between them .. e.g. when a dependency of a package changes, guix creates a new instance of that package ... essentially
<vagrantc>it's a very different model from a typical software distribution... takes some getting used to
<voidie>Yeah I see. My concern was mainly with disk storage. I've read that guix stores (almost) everything in /guix/store, and my root (/) partition is a little less than 25 GB in size
<voidie>My concern is*, rather. Not was
<vagrantc>guix definitely eats a lot of disk space
<voidie>Hmm that's difficult :/
<vagrantc>it saves as much as it can by hard-linking identical files and such
<vagrantc>but the additional disk space usage is one of the costs of a functional package management paradigm
<sneek>Welcome back nly, you have 2 messages!
<sneek>nly, str1ngs says: I purged merged branches on Savannah. If there are any missing branches you need on Savannah can you push them back. Encase I missed something.
<sneek>nly, str1ngs says: feature-release has been renamed to release btw.
<vagrantc>e.g. every time you update glibc, all the packages that use glibc are rebuilt, rather than just hoping it still works with the new version like most distros do
<voidie>Could I hypothetically configure guix to install everything on an external hard drive that I plug in?
<voidie>I see. That seems like a nice way of dealing with it, from a functionality perspective
<vagrantc>voidie: it's all (for the most part) installed in /gnu/store ... so you mount however you want
<voidie>Hmm I'd have to find a way to make it install on /mnt/gnu/store then?
<leoprikler>as long as that external hard drive is always present on boot and can be mounted to /gnu/store, there should be no problem
<voidie>I'm sorry if I ask dumb questions but I'm very new with non-standard ways of installing packages
<vagrantc>voidie: well, if you're only using guix from the hard drive, just mount your external disk on /gnu or /gnu/store ... no need to use /mnt
<leoprikler>you shouldn't really unplug it while your guix system is running tho
<leoprikler>alternatively, you could just store some guix packs on that hard drive
<leoprikler>that way you're not depending on it physically being there all the time, but stuff is by default still installed on the other drive
<voidie>I see, that's a good idea
<voidie>Could you just give me a very general outline how it would look like? For example. what directories (if any) do I create on that external hard drive, where would I mount it, would I have to change something in a config file or can I just mv stuff to the hard drive?
<leoprikler>using guix packs might also come in handy if you also have to deal with non-Guix systems and want to use Guix software on them
<voidie>That's what I'm trying to do actually
<voidie>I'm not going to use the guix distro
<voidie>I want to use the guix package manager on musl void
<leoprikler>okay, first you invoke `guix pack <stuff you want to pack>`
<leoprikler>then you copy the result of that somewhere on your external hard drive
<leoprikler>where you mount it etc. is not relevant for guix packs, they are self-contained
<voidie>I see. So do I copy it or can I move it? Because my concern is with saving disk space
<voidie>By the way, I'm on a very unstable internet connection so I might randomly disconnect. Sorry in advance for that
<leoprikler>if you're concerned about disk space, have a look at `guix gc`
<leoprikler>though you should not need it so quickly
<voidie>Is guix gc a command or is it something I should browse for online?
<voidie>Nvm I found a page about it in the guix reference manual. Gonna read it later
<voidie>Thank you very much for your help leoprikler and vagrantc. I'll try to install the guix package manager and I'll see if I end up low on storage
<vagrantc>good luck!
<voidie>Thanks :)
<voidiee>I disconnected
<voidiee>The last message that I saw was vagrantc's "good luck!". Did I miss anything after that?
<leoprikler>voidiee: Nothing but your own thanks ;)
<voidiee>Hehe thanks
<pkill9>is the guix data service running on the build machines?
<vagrantc>pretty sure it's separate infrastructure
<pkill9>vmpk fails if it can't load the default, ridiculous
<nckx>voidiee: The channel is logged on the WWW, see the topic.
<nckx>pkill9: GDS is totally separate from berlin/CI, not even in the same city.
<voidiee>Oh that's super handy. Thanks nckx
*vagrantc needs to remember ho wto make the guix tarball ... make dist and a bunch of random workarounds
<joshuaBPMan87>Hello #guixoids
<joshuaBPMan87>Work has nothing going on.
<nckx>Work has nothing on Guix.
<joshuaBPMan87>nckx: I second that. Though work does manage to feed me every month. They are stubbornly good at that. :)
<nckx>They wouldn't if they could get away with it.
<nckx>Has anyone made progress on the IceCat media sources robot voice?
***rupicapra[m]1 is now known as rupicapra[m]11
<nckx>I tire of watching sportsball in silence ☹
***rupicapra[m]11 is now known as rupicapra[m]12
<joshuaBPMan87>nckx: Icecat media sources? It that a text to speech thing?
<pkill9>try posting a feature request to the mailing list for guix to feed you, maybe someone will implement it
<joshuaBPMan87>pkill9: haha. Not a bad idea. I would probably need to be a better contributer for that though.
<nckx>There's a bug in Guix's IceCat that distorts video audio in some cases. The cases seem to corrolate with something called ‘media[ ]sources’ in about:config.
<nckx>I don't think it's DRM.
<joshuaBPMan87>nckx: have you tried the nix-service type to install firefox? Yes it would be better if icecat was fixed properly....
<nckx>I'm not sure I trust our Nix service enough to install it on a non-throwaway machine but thanks for the idea. I might get more desperate, and it beats installing Chromium.
*nckx is actually quite sure.
<joshuaBPMan87>nckx: may I ask why you do not trust it? And how about the flatpak?
<raghavgururajan>nckx: Thanks! It was Failing test(s): binlog_encryption.encrypted_slave
<nckx>Ah, I got a different one. Pity.
<joshuaBPMan87>nckx: there is also ungoogled-chromium
<nckx>joshuaBPMan87: I'd rather build everything from source based on Guix's bootstrap binaries.
<joshuaBPMan87>nckx: I fully support you decision! That's probably the "best" way to do it.
*nckx go watch man hit ball with hitstick in silence, later all.
<lfam>Shouldn't epiphany work?
<nckx>Epiphany doesn't work at all. Type URL, RET, nothing.
<lfam>Works for me!
<nckx>bwrap: No permissions to creating new namespace, likely because the kernel does not allow non-privileged user namespaces. On e.g. debian this can be enabled with 'sysctl kernel.unprivileged_userns_clone=1'.
<joshuaBPMan87>nckx: I think I've had similar problems.
<nckx>Mangle my kernel to appease some browser? That's a hell no from me cap.
<nckx>I can't find how to disable this Bubblewrap thing.
<lfam>The kernel has already been mangled by some amateur ;)
<lfam>But I don't think you should need unprivileged namespace for epiphany, if that's what you meant
<apteryx>nckx: on a Guix System? I thought non-privilege user namespaces was already enabled.
<apteryx>epiphany works well from my testing too
<nckx>lfam: I certainly agree, but this is all it says: Could well be unrelated.
<nckx>apteryx: It is by default.
***caleb_ is now known as KE0VVT
<nckx>I tried Epiphany last year and it didn't work with MLB at least.
<apteryx>this bwrap thing is only trying to run epiphany in a sandbox. It's probably more safe to enable unprivileged namespace than try to run Epiphany without it.
<lfam>That would make sense due to DRM
<lfam>Where are unpriv user namespaces enabled in Guix?
*lfam afk, will read the log
<nckx>lfam: CONFIG_USER_NS=y in gnu/packages/aux-files/linux-libre/5.8-x86_64.conf.
<nckx>apteryx: But I don't run Epiphany. So I need to recompile my kernel just to test it. That'll take a while.
<pkill9>i have a request for anyone with access to the build server: to run a script (i'll write it) that gets all the .desktop files in the store from share/applications in every store path
<nckx>That's going to take ages.
<joshuaBPMan87>pkill9: May I ask what service that script would render? I actually have no idea...
<pkill9>i don't think it would take that long
<pkill9>using fd
<pkill9>a superfast alternative to 'find'
<pkill9>similar to the silver searcher
<apteryx>the silver finder
<joshuaBPMan87>watching ghost busters....
*apteryx is still wakaran about why my Gexp snippet fails to resolve #+(canonical-package some-package) -> error: canonical-package: unbound variable
<apteryx>the module has #:use-module (gnu packages base) at the top
<apteryx>which provides canonical-package
<apteryx>is it possible to pass the equivalent of --no-offload when using the (guix derivations) API? namely, (build-derivations ...)
<pkill9>also could alternatively get the output store paths of all the packages and just check those, that would cut down the number of store paths to check
<pkill9>i forgot the build farm has many older packages cached
<leoprikler>apteryx: use set-build-options?
<apteryx>leoprikler: cool. Can I make this call anytime?
<apteryx>in the context of building stuff at the REPL with ,run-in store from (guix monad-repl)
<leoprikler>I'm not quite sure about the specifics, but you should probably collect your arguments in some manner and make one call at some point before invoking build-derivations
<apteryx>I've never understood how to use ,enter-store-monad, perhaps that could be useful
<pancak3>is there a way to create a package that downloads files from multiple locations? Like have multiple origin definitions or something like that?
<apteryx>leoprikler: I made this call: (with-store store (set-build-options store #:offload? #f)), and it seems to stick!
<apteryx>thanks for the tip
<leoprikler>pancak3: you can use origins as inputs or try your own computed-origin
<pancak3>what's a computed-origin?
<apteryx>a neat way to describe a file hierarchy programmatically, using a Gexp
<pancak3>I couldn't find the exact term in the manual. Is it under the gexp section?
<apteryx>ah, sorry, that's computed-file
<apteryx>I don't know much about that one, except that it's probably using Gexp to describe an origin. See the icecat, linux-libre or chromium package definition, which makes use of it
<bluekeys>Is anyone here using stumpwm?
<apteryx>seems my snippet is valid (tested OK from the REPL), so I suppose it's a cyclic dependency problem at run time. I see the computed-origin uses a (delay ...) to avoid those, but I don't see how this can work on a snippet field.
<apteryx>the code above fail because of (delay... and otherwise prints bogus errors (unbound canonical-package variable)
<alexshendi>Hello from guix!
<jonsger>I really don't understand it. `guix weather` says it has 148 of 151 substitues but `guix system reconfigure` wants to build 720 drvs...
<vagrantc>i vaguely recall an email to guix-devel from ludo describing the process for making the release tarball ... but i'm having trouble finding it... sound familiar to anyone?
<apteryx>oh nice, seems rekado had that problem in 2016:
<plasma41>Why did I and loads of other people get kicked and banned from this channel on Friday by an overzealous ChanServ?
<apteryx>it was a mistake, although we're not sure how it happened.
<alexshendi>I have a question: As with every GPD device the GPD MicroPC has a landscape/portrait orientation problem. What I don't understand is: GDM login window orientation is correct, but the X session has wrong orientation and I must correct it with xrandr???
<alexshendi>Also I believe I currently have the base system and kernel from 1.1.0 and the user profile for guix-current. How do I correct this?
<NieDzejkob>alexshendi: confirm that sudo which guix points to /home/$USER/.config/guix/current/bin/guix and then sudo guix system reconfigure /path/to/your/config.scm
<apteryx>so, to answer myself about why the gexp snippet fails to work, it's because it's eager and fails to resolve coreutils-final (cyclic dependency). The solution as proposed by civodul in 2016 would be to make snippet a thunked field.
<NieDzejkob>plasma41: the logs from just after the incident suggest that this was caused by nckx's userscript going berserk
<nckx>This is both the most likely and best reason.
<nckx>I would not type *!*@* by accident 😛 But I have some (in retrospect, years later) dodgy ‘helper’ scripts that I collected when I first started IRCing. Half of them have broke by now. I threw out the rest.
<nckx>(Now I'm a 13373 ircop, obviously.)
*nckx goes back to watching telly.
<apteryx>still unsure if the snippet field of an <origin> record would better be delayed, or thunked? What is the difference?
<apteryx>are both transparent to users?
<apteryx>thunked means wrap in an argless lambda, while delayed means wrapped in a (delay ...) call.
<jackhill>terpri__: I have a kafs-client package:
<terpri__>jackhill, awesome!
<sneek>terpri__, you have 1 message!
<sneek>terpri__, jackhill says: I have a working kafs-client package. The services need more work:
***terpri__ is now known as terpri
<apteryx>continuing my monologue: yes, both are transparent to users. The define-record-type* implements wraps the accessors to thunked or delayed field in a way that they are either evaluated or forced, respectively (c.f. (guix records)).
***sneek_ is now known as sneek
<apteryx>ah, thunked probably preserves the context (captures the closure), while promises probably not.
<erc>I'm trying to change the login shell of my user account to fish. The manual says we need to make a G-expression for this. How can I do this?
*jonsger would be very happy if someone could show him a way to *successfully* set up certificates with certbot-service for multiple domains :)
<vagrantc>when trying to run "make dist" i get: guix- /gnu/store/cq0cmv35s9dhilx14zaghlc08gpc0hwr-tcc-boot0-0.9.26-6.c004e9a/lib/libc.a: error: 'sigprocmask' defined twice
<vagrantc>error: store file names embedded in the distribution
<vagrantc>i've seen similar things before ...
<vagrantc>i guess that means need to update guix- /gnu/store/cq0cmv35s9dhilx14zaghlc08gpc0hwr-tcc-boot0-0.9.26-6.c004e9a/lib/libc.a: error: 'sigprocmask' defined twice
<vagrantc>guess need to update assert-no-store-file-names:
<vagrantc>it looks like it's only a comment...
<efraim>is there anyone online who uses the dovecot service?
<nckx>efraim: I use an opaque-dovecot-configuration.
<efraim>nckx: have you seen the dovecot improvement patches?
<efraim>for patch 2 I like my version better, 5-end I don't feel confident reviewing
<efraim>3 looks like it should be safe
<nckx>Your 2/10 is beyond better.
<vagrantc>ok, i gots a tarball!
<nckx>efraim: Thanks for the ping. I'll deploy them & see what breaks (and if/how your patch 2 changes the resulting binaries).
***caleb_ is now known as KE0VVT
<apteryx>so, I tried making the snippet field of an <origin> record thunked, but this seems to introduce yet another cycle when I tried to build my package :-/.
<vagrantc>janneke: it looks like you added the /gnu/store/... thing in d9484fba91bc360ad18fe82cb9f4963ccb32c580 gnu: commencement: Add bash-mesboot0. ... does it look to you if the full path is required to be in there, or can it be truncated to /gnu/store/…-tcc-boot0-0.9.26-6.c004e9a/lib/libc.a ?
<apteryx>i.e., building any package hangs the 'guix' command, spikes the CPU and eventually exhaust all the memory.
<vagrantc>as replacing the hash with /gnu/store/…
<vagrantc>seems to fix it
<rekado>we must not add literal store file names to the source code
<rekado>IIRC there’s a check to complain about that
<rekado>apteryx: I don’t think you can do fancy things with just snippets because of the lack of delayed evaluation
<rekado>apteryx: if you need fancier things the next best thing is the icecat / linux libre source code hack.
<vagrantc>rekado: yes, that's exactly what's breaking
<jonsger>do I need to understand certbot-service or is it just random what it's doing?
<vagrantc>but i don't know if it's somehow required for bash-mesboot0 somehow
<janneke>vagrantc: you don't mean in the comment?
<vagrantc>janneke: it looks to me like it's a C comment, yes ... but wanted to double-check
<vagrantc>janneke: make dist balks at it
<janneke>vagrantc: that's totally silly -- yes, and make dist /should/
<janneke>having ".../tcc-boot0-0.9.26-6.c004e9a/lib/libc.a" wouold be more than enough
<vagrantc>janneke: ok, then i guess i should just push the fix ... although a little at a loss for writing the commit message
<vagrantc>wait, will bash-mesboot0 trigger a rebuild-the-world scenario?
<NieDzejkob>if you modify the build-side code
<lfam>`guix refresh -l -e '(@@ (gnu packages commencement) bash-mesboot0)'`
<janneke>something like commencement: bash-mesboot0: Remove store file name. \n\n This is a follow-up to commit xxx-the-hash; fixes make dist.
<lfam>"Building the following 6209 packages would ensure 16854 dependent packages are rebuilt..."
<janneke>vagrantc: yeah...i guess that's a problem
*janneke feels pretty bad about this
<vagrantc>janneke: just have to do it before the next release, then ... which will possibly involve some world-rebuilding anyways
<lfam>I don't know the context, but we al lmake mistakes
<vagrantc>what lfam said :)
<janneke>lfam, vagrantc: sure, and thanks :)
<vagrantc>i guess ... i'll report a bug with a patch, and then i can have a message commit message :)
<vagrantc>er, messy commit message ... and someone else can betterificatize it
<janneke>vagrantc: good trick, maybe "someone" can has a good idea about the release/make dist/world rebuild thingy
<lfam>Yes, I'm sure that someone will have advice ;)
<lfam>"That" someone
<janneke>lfam: yeah, preferrably
<janneke>vagrantc: just for fun, try making bash-mesboot0 public (define-public bash-mesboot0 and then
<janneke>./pre-inst-env guix refresh -l bash-mesboot0
<janneke>Building the following 6186 packages would ensure 16824 dependent packages are rebuilt:
<lfam>You can do it with the -e option, like `guix refresh -l -e '(@@ (gnu packages commencement) bash-mesboot0)'`
<lfam>"-e / --expression"
<vagrantc>janneke: i trusted lfam's test of that well enough
<lfam>It's always useful to try something else and see what happens!
<vagrantc>true enough
<janneke>lfam: nice, i use that for build -- good to know it also works for refresh
<jonsger>lfam: nice hint
<jonsger>is there a reason why guix refresh ignores hidden packages? I mean it's a developer tool not something for the user who should not install hidden packages...
<str1ngs>I'm not a big fan of hiding packages. take for example you can't do this guix edit gcc-10
<vagrantc>janneke: oh, i'm totally using your commit message. :)
<vagrantc>way more succinct than i was writing
<str1ngs>janneke: BTW I have some local sraged work for getting emacsy working with guile 3. I'll try to push something to a guile-3 branch sometime this week. if you want to double check it before I merge?
<lfam>jonsger: I don't there is a good reason. Hidden packages are meant to be hidden from the "UI", but that really means from `guix show`. I think there is effectively a different UI for development, and `guix refresh` should mention hidden packages
<lfam>`guix edit` is another case where you should be able to use it for all public packages
<lfam>There's a distinction to be drawn between public hidden packages and packages that are private (not exported)
<janneke>str1ngs: sure, i'll have a look; thanks
<PotentialUser-15>Hello! How do I make it so that when I run `guix system reconfigure`, it'll use the latest 5.4 Linux-libre kernel?
<str1ngs>janneke: more then likely I'll have everything working smoothly for guile 2 and 3 before pushing anything.
<lfam>PotentialUser-15: In your config.scm, you'll want to include a kernel field, like (kernel linux-libre-5.4)
<lfam>If you do that, you'll also want to import the module where the linux-libre packages are defined. That means you'll add it to the list of imported packages modules, like (use-packages modules linux)
<lfam>I'm curious PotentialUser-15, were you having problems with the 5.7 or 5.8 kernels?