IRC channel logs

<Guest72>Hey Im trying to set up suspension to ram both if I press the buttons on my keyboard, if I launch loginctl command, and if I close the lid. Which means I'm trying to set up elogind. I cant find any examples in the cookbook

<Guest72>Hence I dont know where to start. Id like to find some practical example, any references by any chance?

<Guest72>I have zero experience with the guix system config file

<Guest72>I just installed guix (no graphical interface)

<Guest72>Any suggestion is appreciated

<bdunahu>I was just about to respond :(

<Guest72>Im still online bdunahu

<Guest72>And I go check the logs if my ckient disconnects :)

<bdunahu>oh my client said there was a ping timeout

<bdunahu>I think this %desktop-services variable is what is currently bringing in the elogind service into your configuration: https://guix.gnu.org/manual/1.5.0/en/html_node/Desktop-Services.html

<Guest72>There aas indeed

<YAR_Oracool>does json modul ship with guix?

<Guest72>Agreed abiut the desktop services

<bdunahu>Guest72: Since it sounds like you want configure it beyond defaults, you can manipulate that variable with the `modify-services` function, which is here: https://guix.gnu.org/manual/1.5.0/en/html_node/Service-Reference.html

<Guest72>So wait the desktop services is "active" by default?

<Guest72>Cause loginctl is not a recognised command hence I deduce elogind is not active by default

<bdunahu>is the %desktop-services variable in the list of services in your config already? Or maybe %base-services?

<YAR_Oracool>(use-modules (json) this one.

<czan>Guest72: You could just add (service elogind-service-type) to your system services and see what happens. :P

<Guest72>But the manual suggests that elogind service has sth to do with the desktop Module

<Guest72>YAR_Oracool is your message about my query or other stuff?

<czan>elogind is included in the desktop services, but that's not the same as having "something to do with" desktop services.

<czan>elogind is also needed for docker to work, so I've got elogind on some of my headless servers.

<YAR_Oracool>other. I'm just trying to see if module for json ships with guix

<Guest72>czan does it mean that to active elogind I shall "activate" the desktop sefvices module? I wish I knew what a module is Im so amateur

<bdunahu>No, %desktop-services is simply a list of a bunch of services which are nice to have on a desktop

<czan>Can you send a paste (see the channel topic) with your config.scm?

<Guest72>czan shall I use your code or the modify services function?

<czan>Likely neither. If you're using %base-services then you just need to add a (service elogind-service-type) to your list of services. If you send your config in a paste I can show you what it would look like.

<Guest72>czan my config is the barest minimumest config possible. Im actually connected now to irc on my phone so I cant send that but its really the rightafterinstallation config

<czan>How did you do the installation? The installer lets you select options which change a few parts of the config, so that's ambiguous.

<Guest72>Ive never touched the config from the installation i did yesterday

<Guest72>Yeah a couple of things, keybaord language and a few other

<Guest72>I dont even have emacs yet to open the config file and see what inside

<czan>In barest terms: you should have a (services (append (list ...) %base-services)) somewhere in your config (where the "..." is some bunch of things). You want to turn it into (services (list (service elogind-service-type) ...) %base-services).

<Guest72>Ok question why now base services? I thiught elogind was in desktop serv?

<czan>You might also need a (use-service-modules desktop) at the top of the file.

<czan>%base-services is for a minimal system, %desktop-services is for a desktop system, and includes a bunch more stuff.

<czan>Feel free to just replace %base-services with %desktop-services if you'd rather do that. It does a bit more than just adding elogind, but if that's what you want then that's fine.

<Guest72>Oh I understand. So if I were to tinker around desktop serv then it would activate A biunch of things I dont need (but elogind would be one of them) so we decided I add it artificially to base services?

<czan>In particular, I believe it adds a graphical login manager, but I thought you didn't want a graphical interface.

<czan>It's not "artificial", it's just that %desktop-services is solving a different problem to yours. It's just a different configuration.

<Guest72>Can I just add elogind by itself, withiut making it partnof base services?

<czan>You should *already* have %base-services in the list.

<czan>If you're not using %base-services or %desktop-services then you are an expert and know more than me about Guix. :)

<Guest72>Ok let me see wait I start a live system and see whats in my config file

<Guest72>I guess im confused as to why add elogind to the base services modules, and then activate such module, instead of leVing it as it is and just activate elogind by itself a few lines below

<czan>I don't understand your question, and I'm finding it hard to help without knowing exactly what config you're working with.

<Guest72>Ok gimme a min. Apologies

<Guest72>And I appreciate your help so much

<Guest72>czan so I have:

<Guest72>(use-modules (gnu))

<czan>Don't put your config directly in the channel. Use a paste (see the channel description).

<Guest72>(use-service-modules cups desktop networking ssh xorg)

<Guest72>Ok

<Guest72>Ok so paste.debian.net/plainh/0859375e

<czan>Okay, thanks! I hope you didn't have to type it all out by hand. :)

<Guest72>And since were at it, do you know why the capslock-ctl switch and the trick to produce accent on the dvorak keyboard , as you can see from the config , dont work?

<Guest72>Ahahahah nono but had to start a live iso

<czan>Your paste has a typo in the keyboard bit, that might be relevant? ctlr -> ctrl

<Guest72>Ahh maybe thats why. And for the accent i guess i dont know what the compose button is, i suppose is the windows button, but I tried to produxe accents and failed. Maybe a just need to try better for tgat

<czan>But it looks like you're using %base-services for your services, and then adding network-manager, wpa-supplicant, and ntp. If you want elogind, then you just need to add it alongside them. Something like https://paste.debian.net/plainh/8e7091e3

<Guest72>Ok lets get to the suspend business

<Guest72>Ok let me see

<czan>I just added a single form: "(service elogind-service-type)"

<czan>Then you need to do a "sudo guix system reconfigure /etc/config.scm" to actually modify the system to match the config.

<Guest72>Ok im following u

<Guest72>However wait

<Guest72>Then I suppose loginctl will be available as a command

<Guest72>And will allow me to suspend with the suspend command

<Guest72>Xorrect? And will the susoend button on my keyboard also be active now?

<Guest72>And when I close the lid also?

<czan>I can't say with absolute certainty, but I expect yes.

<Guest72>Shall I try now and report back in 10 min?

<czan>Yup!

<bdunahu>Guest72: the power key will probably shutdown by default :(

<bdunahu>you need to configure the service to do a hibernate. At least that's what I do

<n|Phreak>sweet swaylock now works with niri

<czan>Did swaylock not work with niri? I've been using them together for a while...

<Guest72>bdunahu my keyboard has a sleep key

<bdunahu>oh, then maybe it would work. I only have a power key on mine, and that powers off by default

<n|Phreak>No I could not get it to work for the life of me , I had to use a different channel for niri because there was on desktop service for it

<n|Phreak>on/no

<czan>If you just install niri as a package then it shows up in GDM/whatever, and then I just have screen-locker-service-type configured with swaylock as it has in the manual.

<czan>I assume this other channel has a niri-service-type which just does both of those things for you?

<czan>I can't think of a reason something like that couldn't be upstreamed into Guix.

<n|Phreak>I was just using home-config for niri

<n|Phreak>but there wasn't a desktop-service for the system config

<n|Phreak>I probably did it the long way but it works now and if I can just reconfigure my home-config when I want to add stuff.

<czan>Niri isn't a desktop, so I wouldn't expect it to have a desktop-service. But yeah, how does your home service configure the screen locker?

<n|Phreak>sec

<n|Phreak>I use a home-shepherd service

<n|Phreak>for swayidle

<czan>Right, but to get swaylock itself to be able to log the screen I thought you needed some PAM config at the system level.

<czan>swayidle is just to lock the screen after a timeout, right?

<n|Phreak>so in the system config I use (service screen-locker-service-type

<n|Phreak> (screen-locker-configuration

<n|Phreak> (name "swaylock")

<n|Phreak> (program (file-append swaylock-effects "/bin/swaylock"))

<n|Phreak> (using-pam? #t)

<n|Phreak> (using-setuid? #f))))

<bjc>got 'guix pull' working on the hurd with remote guix =)

<czan>Don't paste your config in the channel, please. Use a paste (see the channel description) if you want to show multiple lines.

<n|Phreak>sorry I didn't think 5 lines would matter

<czan>Yeah, so that's the same as me (except I just use swaylock, not swaylock-effects). That has worked all along for me, so I still don't really understand what you mean about it now working with Niri. But if you've got things working: great! :)

<Guest72>bjc by the way how stable is the hurd if I were to use it on my x200 with no graohical interface? czan reconfig in progress.

<bjc>it is not stable

<bjc>i mean, it's not super crashy, but lots of things don't work and it can't really update itself

<bjc>it's also really slow because it's one cpu

<bjc>and if it does crash, you can't easily fsck the root fs and boot up again

<bjc>the debian hurd port is more polished if you're interested in playing with it

<n|Phreak>ACTION bjc I really want to play with hurd but I just don't have time ,I did a long time ago but I'm more wanting to use plan9 again.

<Guest72>czan lid doesnt work. Tho I wonder if there is a sensor in the x200 to sense that the lid was closed...??

<Guest72>Susoend from keyboard ok

<n|Phreak>ACTION czan I screwed it up , so it was me that couldn't get it to work on that it didn't work in general.

<Guest72>And loginctl ok

<czan>Guest72: Maybe check if there's a sensor before we go down this rabbit hole. If there isn't, there's not much Guix can do. :)

<n|Phreak>clamshell works for me

<n|Phreak>I just tried it , goes right to suspend

<Guest72>Phreak are you talking about my susoend issues with my guix?

<n|Phreak>yeah

<Guest72>What is clamshell?

<Guest72>czan works with thr lid too takes a minute

<czan>It looks like the default elogind is (handle-lid-switch 'suspend), (handle-lid-switch-docked 'ignore), and (handle-lid-switch-external-power 'suspend). I don't know exactly what "docked" means here, but is it possible your system considers itself docked?

<czan>Oh, okay, cool. :)

<Guest72>Ok so last two questions before I go sleep

<n|Phreak>oh sorry clamshell is like .. laptop

<Guest72>I dont understand n|Phreak ate you saying you hve an x200 and tjere is a sensor?

<Guest72>Ok so question 1: what do i need to do to produce accents

<Guest72>With my dvorak programmer and the config I pasted?

<Guest72>(in particular what is the system.button)

<n|Phreak>so when I close my laptop it goes into suspend

<Guest72>Question two I was recommended to use the modify function to activatr elogind, what happened with that?

<czan>For question two: that advice is if you are using %desktop-services, and even then only if you want to modify the elogind config. Since you're using %base-services you can add elogind directly without needing to use modify-services.

<Guest72>Oh bonus question for czan if i wanted to customise elogind with the options in tje manual, how does the config look like? (If you just customise one option and show me the past I will understand the dyntax)

<czan>For question one: I don't know.

<Guest72>Thanks czan

<czan>Something like this: (service elogind-service-type (elogind-configuration (handle-lid-switch 'ignore))). The options are documented in the info manual in "(guix) Desktop Services".

<Guest72>Ok thanks so much

<Guest72>Im going to sleep

<Guest72>You were very kind to me

<Guest72>And patient

<podiki>sneek: later tell efraim could you take a look at aarch64 failures on mesa-updates? julia and guile-fibers were pointed out (and might just merge the branch, nothing is building anymore unless there is actual fix for aarch64 stuff)

<sneek>Got it.

<podiki>sneek: later tell efraim nevermind about guile-fibers, there is substitute for it

<sneek>Got it.

<ieure>cccccckgcuedhkcihlkcvdlinlefuniudgullbukkihc

<ieure>whoops, sorry

<bdunahu>wkhkcofcgixjgluyhvqdjgh

<n|Phreak>hmm kitty terminal is super old from guix packages

<GreenRocky>Hello 👋 How do Guix containers compare to established sandboxing utilities like bubblewrap?

<GreenRocky>I hear they use similar technologies (Linux namespaces)

<czan>GreenRocky: I can confirm that Guix uses Linux namespaces for its containers. Assuming you mean "guix shell --container", it might be worth reading the manual for details of what you can do with it https://guix.gnu.org/manual/devel/en/html_node/Invoking-guix-shell.html

<czan>There are also a few blog posts about containers in Guix, but I don't think there's a meaningful comparison with other tools. They're mostly talking about what Guix does. https://guix.gnu.org/en/blog/tags/containers/

<ieure>All Linux containerization stuff uses Linux namespaces.

<ieure>That's like... what containers are.

<ieure>Never heard of bubblewrap.

<cnx>n|Phreak, kitty update is being worked on at https://codeberg.org/guix/guix/pulls/3513

<efraim>podiki: sure, I'll take a look at it

<sneek>Welcome back efraim, you have 2 messages!

<sneek>efraim, podiki says: could you take a look at aarch64 failures on mesa-updates? julia and guile-fibers were pointed out (and might just merge the branch, nothing is building anymore unless there is actual fix for aarch64 stuff)

<sneek>efraim, podiki says: nevermind about guile-fibers, there is substitute for it

<podiki>efraim: thanks! andreas pointed out the qa blocked build page showed like 11k blocked by julia (which much more than julia packages right?), though that cleared upon its own for i686

<efraim>podiki: probably due to the utf8proc package

<efraim>julia built fine on mesa-updates on my rpi5

<podiki>ah so maybe just a build farm issue

<podiki>i better get to bed but feel free to merge the branch then, coverage of i686, x86_64, and armhf are at least as good as master

<efraim>ok

<podiki>otherwise i'll double check things and do it tomorrow at some point

<cnx>how i stop the mail lists from referring to me as "via ... GNU Guix ..."?

<janneke>cnx: i'm seeing: "From: Nguyễn Gia Phong via "Development of GNU Guix and the GNU System distribution." <guix-devel@gnu.org> "

<janneke>so it looks to me as if that's what you use in your email client

<czan>No, that's mailman rewriting it.

<czan>But I don't know how to turn it off. It doesn't happen to my emails, but I don't know why.

<janneke>cnx: you might ask efraim what your From looks like, as he was cc'd on that original mail

<czan>I was, too, and the From that came directly to me looks fine.

<janneke>could it be that cnx's from doesn't pass DMARC?

<czan>Oh, actually, I wonder if it's because cnx's DKIM-Signature includes Reply-To, whereas mine doesn't. (Or maybe some other header.)

<czan>If mailman wants to change a header in the DKIM-Signature then it has to re-sign the message, which means it has to come from an address that the mailman server can make a signature for.

<identity>how can i use LLVMpipe on Guix?

<civodul>uh, we still have the problem on other distros that “systemctl restart guix-daemon” does not restart ‘gnu-store.mount’, and thus the daemon gets a read-only view of the store :-/

<sneek>civodul, you have 1 message!

<sneek>civodul, yelninei says: i could not reproduce your issues in commencement (on x86_64-linux)

<Rutherther>civodul: are you sure it's the last guix-daemon service variant? That one does that

<Rutherther>civodul: and the foreign test even tests for that

<Guest72>Hey I installed Guix System via the tui and guided partitioning. My laptop has 8gb ram but the swap recommended by the install is 4gb.

<Guest72>Should it not be more like 16gb?

<Rutherther>Guest72: why? It depends on the usage of the swap

<czan>I've heard the rule of thumb of "swap size as 1/2 RAM" before.

<Guest72>Well for one thing so that I can hybernate. I mean the install knows my disk is 250gb and redommends my / be 80gb plus separate home of 10gb so there is space to spare and it would make sense to rexommend 16gb awap

<Guest72>Ah i heard the rule swap = 2*ram

<bjc>some of us don't use swap at all…

<Rutherther>Guest72: right, for hibernation you need more than 100 % of the RAM. The installer's swap use case isn't hibernation though. So I do not think the installer's recommendation is wrong here

<Rutherther>currently the logic is: Use a 4GB size for the swap if it represents less than 5% of the disk space. Otherwise, set the swap size to 5% of the disk space.

<Guest72>Rutherther ok... But since there is space to spare then shiuld it not hVe recommended the amount that fits the most scenarios, hence also hybern?

<Guest72>ah...

<Guest72>Rutherther thanks for explaining the logic, but it seems to coarse a criteria.

<Rutherther>Guest72: does it fit most scenarios, though? also, what do you mean that there is space to spare?

<Guest72>Rutherther sorry I didnt remember that i astificially made the / partition smaller than recommended. Hence there would have been no spare unpartitioned space as per the installer partitioning recommendations. But nonetheless I think it shiuld ask whether I want to hybernate in which case adjust the swap suggestion

<Guest72>People still hybernate these days

<Guest72>Like a checkbox in tje installer i think would serve a purpose

<Rutherther>hm, but you did that after the auto partitioning, right? 1. auto partition, 2. you noticed / is too much for you, 3. you changed the size. Is that correct?

<Guest72>Rutherther yes correct, so the installer didnt know that there would have been unpRtitioned space, hence was not illogical to suggest that amount of swap

<Guest72>But I still think I would recommend a checkbox in the installer for hybernation, I think it would be very useful

<Rutherther>yeah, the logic currently doesn't support asking you questions or looking at the disk afterwards, it just sees. I am not really sure if it's worth changing much, though... mainly because if you know what you want you can change it afterwards or use manual

<Guest72>I mean, now I have to manually chNge my swap

<Guest72>Looking at the disk afterwards I agree not to implement by the way

<Guest72>Actually since we're at it I remember I also found something wrong with the bios partition

<cnx>thanks for the investigation, czan, I thought I was missing some configuration when I subscribed to the list

<Rutherther>Guest72: what was it?

<Guest72>So when I use the automatic part and use gpt, it prodices a bios partition for grub and a sentence next to it says bios partition. If you click on it as to see or edit properties but tou dont change anythibg and go back to the outline of the partition , the sentence bios partition has disappeared

<Guest72>So then I didnt know what was going on and had to exit the installer and redo it so that the sentence bios partition reappears

<Rutherther>you can just go back a page in the intaller and redo the partitioning

<Guest72>This will catch many nonprofessional users off guard. To be fair I still dont know what was up with it

<czan>cnx: To be clear, I'm just guessing. I reserve the right to be 100% wrong about my guess. :)

<Rutherther>but yeah, that is not really good, and I am not sure if it means the "esp" flag really disappears in this scenario or not

<Guest72>Rutherther yeah I guess it's a potential bug

<cnx>yea czan, i understand that, but some info would make it searchable on the interwebs

<identity>generally 2-3GB of swap is a good default (certainly much better than no swap at all) unless you want to hibernate. if you can, you should run your usual workload and check how much swap is used, and use at least that much, with some wiggle space

<identity>swap-size=ram-size×2 and swap-size=ram-size×2 are not founded in fact

<identity>s/×/÷/

<kestrelw`>Somehow `git pull` in a checkout is authenticating 123.000 new commits.

<gabber>kestrelwx: i think this means that you last authenticated them ~123000 commits ago?

<kestrelwx>Some sort of back to the future.

<kestrelwx>gabber: I've had to disable the authentication hook for some change, so that probably was part of it.

<gabber>probably. 123000 is quite a lot

<strawburster>I've set aside a few hours to try to make some merge requests. I ran into trouble last time but I couldn't remember exactly what went wrong, it was with trying to build the rust app system76-power. I'm going to try it again and see

<bjc>'guix build guix' is broken on hurd from: setting synchronous mode: unable to open database file

<bjc>and i wonder if that has to do with this, from earlier: unprivileged user namespaces not supported; running 'guix-daemon --disable-chroot'

<bjc>i wish building guix didn't take so long. forgetting the '-K' on the previous one has cost me ~2 hours =(

<janneke>bjc: seen https://codeberg.org/guix/guix/issues/4086?

<janneke>maybe you could add your info to that?

<bjc>i'll have a poke

<PotentialUser-71>Does ZFS-on-root Just Work(tm) with Guix, or is it more work than just specifying a ZFS filesystem as root and adding the module to kernel-loadable-modules?

<bjc>it does not work at all, and there are no current plans to make it work, and previous plans have all stalled out

<PotentialUser-71>Guess I'll just use btrfs then.

<bjc>yep

<nikolar>what exactly is the blocker for zfs on root

<ieure>nikolar, Take a look at some of the open PRs/issues around it. I believe there are multiple issues, but one of them is not having the ZFS tools inside the initrd, so there's no stuff to import the pool and mount the root FS.

<nikolar>i wouldn't imagine it being too hard to adjust the initrd though

<ieure>Also likely issues around how file-system records are handled and ZFS' system, which doesn't really interact with the normal mounting stuff.

<ieure>mra was working on this, they probably have a better idea of the current status.

<nikolar>and you can use normal mounting with zfs as well if you so prefer

<nikolar>i am just curious, i am no guix expert

<bjc>there's a patch for the initrd stuff, but it's held up on license contamination concerns, and the fix for that seems to have stalled out

<mra>nikolar: kernel drivers. it's ultimately a licensing issue

<nikolar>i see

<bjc>i had a zfs-on-root build at one point, but it had some shepherd issue, iirc

<bjc>it's been a long time since then though

<nikolar>mra: doesn't dkms work around it

<bjc>the concern is that the official substitute servers can be coerced somehow into building and distributing an initrd with the included dkms, which may be a license problem

<mra>see https://codeberg.org/guix/guix/pulls/3369 for my (stalled) attempts to a provide a fix for licensing issues

<nikolar>right so a fix would be to mark package as excluded from substitues or something

<nikolar>yea exactly like that

<mra>the solution that I put together does work, but it has unacceptable performance costs that I don't really know how to resolve

<bjc>oh, that's you who was heading that up =)

<mra>ayup

<bjc>i bet you can guess my take

<nikolar>what is it

<mra>the most promising progress on this seems to be coming from hako. see here for what they've been working on: https://codeberg.org/guix/guix/pulls/1917#issuecomment-12257826

<bjc>i think the solution is worth pursuing, but not worth holding up the patch for. it's been years now

<bjc>3?

<mra>something like three years. I've personally been working on it on and off for nearly two

<bjc>the problem is academic, and doesn't just affect this one patch. it's unfair that it's being held up for this

<nikolar>so you'll have to rebuild your own kernel locally?

<bjc>no, just the dkms

<bjc>not quick to do that, but it's not the whole kernel

<nikolar>oh on my laptop, dkms builds in under a minute i think

<nikolar>it's really not bad

<bjc>zfs is big

<nikolar>not *that* big

<nikolar>the patch you mentioned changes the kernel config

<bjc>well, if you're already building the dkms, the times will be equivalent, pretty much

<nikolar>seems to me that you're intended to rebuild the whole kernel (until the subsitute issue is resolved)

<ieure>I'm pretty sure that's not accurate? ZFS isn't even in the kernel, why would you need to rebuild it?

<nikolar>> I also use a customized kernel with ZFS module built in, since there's compatibility issue when loading the module from Guix initrd.

<bjc>oh, i didn't see that solution. i'm not sure what it entails. i've only used the dkms

<ieure>Ah, okay, I see.

<mra>my attempts have pretty well stalled out, and the lack of zfs support has been blocking me from actually using guix system, so I've been short motivation since finding out about the performance issues. at this point I'm just hoping that hako's solution proves workable

<nikolar>ieure: it's not in the upstream kernel sources, but you can still build it in apparently

<nikolar>i don't hate the idea of rebuilding the kernel with zfs locally, though that's going to take a while to do on a lower power machine

<mra>it's not actually clear to me how this solves the copyright issue. I guess the point is that if "kernel-with-zfs" is just a separate package, then it can simply be marked non-substitutable and bingo bongo?

<bjc>the point is that you have to do it locally

<bjc>there will never be substitutes

<mra>hm, I see

<mra>well, I'll be very happy if hako gets it working!

<bjc>i wonder what the initrd incompatibilities are

<yelninei>bjc: One of the issues with building guix and --disable-chroot is that you also get the tests that require networking and those that connect to the running daemon. This can somewhat be mitigated but imo tthe guix package should not rely on the sandbox to disable these tests

<bjc>right, but in this case it's saying it can't access the database, so i'm wondering if it's just a dumb path problem because it's not chrooted

<yelninei>who is saying that? the running guix-daemon or the tested one? i assume this is in a test?

<Noisytoot>How can I remove the isci module from the default initrd? My kernel is compiled without it, so linux-modules.drv is failing to build

<yelninei>also i think wed need another guix package update because of the libc ungraft

<bjc>this is from the copy-guile-bootstrap phase

<bjc>it has to copy it into the test store, and that's what breaks

<bjc>i don't know why this is necessary, but i have to imagine it's very important to go through so much trouble =)

<civodul>csantosb: ahem, maybe the hwloc change wasn’t fit for master: https://ci.guix.gnu.org/eval/2151660 :-)

<yelninei>hmm, on commit 74f536ef60cdf763efa9dd7c8a0bf3806788fcd7 i have built guix build guix -s x86_64-gnu --without-tests=guix. The only extra thing i have is I dont allow the guixbuilder group access to my running guix-daemon socket

<civodul>efraim: looks like bbb2235 caused a faire amount of rebuilds: https://guix.bordeaux.inria.fr/eval/9235681

<civodul>did it go through pulls.ci?

<Noisytoot>(initrd-modules (remove (lambda (module) (equal? module "isci")) %base-initrd-modules)) seems to have fixed it

<civodul>anyway, the situation is pretty dire today, with substitutes missing for a range of things

<efraim>it didn't, it was after the merge to get it to build again, after the cargo-build-system change in the mesa-updates branch

<efraim>it was also the first commit after the mesa-updates merge

<Noisytoot>building zfs into the kernel seems like a terrible idea. building an entire kernel with the default config takes hours to days

<kestrelwx>Should Librewolf use pinned inputs? It got built on mesa-updates, and on the recent PR with security fixes before it, but now on master there isn't a substitute for it.

<Noisytoot>could someone please review https://issues.guix.gnu.org/67403? it's sat unreviewed from 2 years, fixes a freedom bug, and is blocking me from packaging ectool

<Noisytoot>s/from/for/

<ieure>kestrelwx, LW already pins some important inputs like rust and nss.

<bjc>yelninei: without-tests would disable the copy-bootstrap-guile phase, which is what's crashing

<gabber>Noisytoot: i can not guarantee to review it, but could you re-open the issue on Codeberg? IIRC issues.g.g.o is deprecated since the end of last year...

<bjc>yelninei: since that phase is (add-before 'check)

<gabber>Noisytoot: feel free to tag or assign me in the issue (still no guarantee i can work it any time soon)

<bjc>i suppose i could just do that after this latest build fails

<Noisytoot>gabber: should I do it myself or ask the original author of the patch series (GNUtoo) to resubmit it?

<civodul>oooh, i see, it’s the glibc ungraft that’s causing so much churn

<gabber>Noisytoot: i don't think it matters much (if you rebase the commits onto master you would only change the committed-by line, not the author line). maybe you could also tag the original author? if you do it right now you could potentially fasten the execution path ;)

<yelninei>bjc: without-tests does not delete the check phase but rather skips its. https://paste.debian.net/hidden/8e75b17f

<bjc>oh

<bjc>hrm. why would it work for you then

<bjc>come to think of it, this has worked for me in the past, because i had test failures holding it up

<yelninei>ACTION is not looking forward to rebuilding 10 gccs

<kestrelwx>ieure: I should've just let the maintainers know this would happen, but I don't know how important it is.

<Noisytoot>gabber: actually it seems that someone already updated ifdtool and intelmetool since the last time I checked, so all that needs to be done now is adding bucts and nvramtool

<gabber>Noisytoot: hear, hear!

<Noisytoot>icedove is currently failing to build

<Noisytoot>what would be the correct command to update my entire profile except for a specific package (icedove, because it's failing to build)?

<efraim>I think it's `guix package -u . --do-not-upgrade icedove`

<Noisytoot>thanks

<test202020>hi

<ieure>Hello.

<test202020>ti have error Xproc_run_async: ofl no aucj file

<test202020>while to run greetd service

<test202020>hm, that glich, i solve problem

<PotentialUser-11>bjc Hello. Yesterday you mentioned that you used envrc alongside a manifest file or profile.

<PotentialUser-11>What variables do you store there? I'm afraid of missing some.

<Noisytoot>gabber: I opened a PR adding bucts, nvramtool, and ectool, and updating all coreboot utils to 26.03: https://codeberg.org/guix/guix/pulls/7576

<bjc>PotentialUser-11: i have a .envrc file at the toplevel of each project i use it with, which just contains the line 'use guix'

<bjc>that causes it to try to use guix shell, so you'll need a manifest.scm with something like '(specifications->manifest '("pkg1" "pkg2"))' in it

<bjc>(this is from memory. i may be off on details)

<bjc>you'll need to both: 'direnv allow' in that directory /as well as/ add it to a permission list for guix shell. it'll tell you how to do that after the 'direnv allow' step

<bjc>or, really, during it

<PotentialUser-11>Oh, that is a bit more elaborate than I though. But I'll try to make sense of it. Thank you.

<civodul>yelninei: hey! i’m thinking we should rebase ‘core-packages-team’ on ‘master’; WDYT?

<yelninei>ACTION thinks they should join core-packages team

<yelninei>civodul: there were some merge conflicts already mostly in copyright headers etc, but would be good

<civodul>yelninei: oh you tried already?

<yelninei>yes, when I rebased my coreutils branch against core-packages-team

<civodul>yelninei: a branch that upgrades coreutils?

<civodul>(should we have a team call sometime?)

<yelninei>civodul: well the one from guix/guix!4058 with all the other stuff

<yelninei>but i am not sure of all the things to convince coreutils-mesboot to compile as some things seem a bit fishy (but it works i guess)

<bjc>ah. i think i might know why my guix build was crashing: /tmp/guix-test was left behind from a previous build. cleaning up /tmp and now it's going through make check

<bjc>it's kinda funny that hurd, of all things, has to be run without chroot =)

<bjc>well that's progress. maybe i can coax a system build out of this thing yet

<yelninei>bjc: There is https://issues.guix.gnu.org/43857 but it would to get rebased

<bjc>interesting

<yelninei>bjc: i think a system build also depends on gdk-pixbuf (at least the manifest says so) and that depends on 1 million python packages with some fun test failures

<bjc>i saw one in setuptools, i think, which crashed with sigbus or something. i imagine that stops a lot of stuff from working

<bjc>oh, is that for the background image for the bootloader?

<bjc>i don't even know how grub installation works on hurd. hopefully the same as linux

<bjc>doesn't grub require rsvg now?

<civodul>yelninei: rebasing happened to be tedious, due to the “remove input labels” and “use inherited field value” changes…

<PotentialUser-11>Is `/media` used on Guix? Can I remove it? I think mounts are placed on `/mnt` or in `/run/media` .

<ieure>PotentialUser-11, I don't have /media on my nearest Guix System machine.

<ieure>PotentialUser-11, There was a change in the last ~year to move from /media to /run/media. Pretty sure this system lived through that change, maybe all you need to do is reconfigure your system with a newer commit?

<PotentialUser-11>ieure I recall the change. You mean running `guix system reconfigure` ? I have done so. Perhaps it did nothing because I have a user-named directory inside.

<PotentialUser-11>But if indeed it is useless now, I may as well remove it manually. Thank you.

<ieure>PotentialUser-11, Yes, that's what I mean. But yeah, feel free to remove it manually if you like.

<PotentialUser-11>Oh, and one more question related to the reconfiguration: Why has it, some times, not chose to install the latest kernel I see listed as available?

<ieure>PotentialUser-11, Do you have an example of that?

<ieure>It's always installed the latest kernel, if that's what I ask for in my configuration.

<ieure>PotentialUser-11, Are you doing a `guix pull' before your `sudo guix system reconfigure'?

<PotentialUser-11>Yes. I did a pull, saw linux 6.19 available but `system reconfigure` went with 6.18 again.

<PotentialUser-11>I will give you that I am using nonguix, which may have been a significant factor.

<ieure>PotentialUser-11, 6.19 is available, but not the default.

<ieure>PotentialUser-11, In (gnu packages linux), linux-libre is linux-libre-6.18.

<ieure>So if you want to run 6.19, you need to specify linux-6.19 in your operating-system config.

<ieure>Rather than `linux', which is the default, which is 6.18.

<PotentialUser-11>I thought the default was "whatever is highest". =L

<ieure>PotentialUser-11, When you use a specification -- yes, it is. In a

<ieure>...in an operating-system, you refer to a specific variable, so you don't get that behavior.

<PotentialUser-11>I imagine there is no way to say, in the configuration file, "linux-latest". It is OK, I can live without it. Thank you.

<ieure>PotentialUser-11, If you use the (gnu packages) module, you can use (specification->package "linux").

<ieure>PotentialUser-11, I personally would not advise this, but it is something you can do.

<ieure>There are usually good reasons why a given kernel version isn't the default. If you crave the excitement of discovering why that is, I won't stop you.

<PotentialUser-11>:')

<PotentialUser-11>With distributions protecting users from those situations, it is expected that we do not realize the potential problems that may come from running the bleeding edge.

IRC channel logs

2026-03-30.log