IRC channel logs
2022-11-08.log
back to list of logs
<antipode>munen: To reduce network traffic, you could set up a caching HTTP proxy <user>Someeonee already go libgcc_s on guix? <antipode>Is this a response to ???Hey all, does anyone know what package has libgcc_s.so.1? Thanks!??? <antipode>If so, I'd assume the "lib" output of "gcc". <user>lib output off gcc doest notcreated libgcc_s here <PotentialUser-42>I don't know what language they speak here so I will leave this message in English and Spanish <PotentialUser-42>still keep Guix distribution active? I ask because I barely bump into it and according to wikipedia it came out in May of 2021 <podiki[m]>PotentialUser-42: guix is very active, growing every day <lilyp>We invented time travel very early. That's how Guix could be ported the Hurd in 2015. <cnx>is there a way to subscribe to a patch on issues.guix? <noob_tea[m]>I think they mean the last version was in 2021 may according to wiki <cbaines>ph03n1xaim[m], Guix changes multiple times a day normally, most users update fairly regularly I think <unmatched-paren>ph03n1xaim[m]: Guix is rolling-release, so a guix pull will update to the head of the master branch <unmatched-paren>ph03n1xaim[m]: Which is why there's so much confusion regarding the manual; the release manual is the one most people seem to click on, but it's also very outdated <ph03n1xaim[m]><unmatched-paren> "/ph03n1x/aim: Which is why there..." <- Ohhhhhhhhhhh <rekado>the logs don???t display ??? correctly <abrenon>is there a way to determine how a particular package made its way into the current environment ? <abrenon>I thought find was from coreutils and discovered it was actually from findutils: I have it in my environment but I never mentioned it in my configuration <abrenon>I suppose it was propagated from another package, but I'm at a loss to find out which one <iyzsong>abrenon: you can search it in the manifest file (eg: ~/.guix-profile/manifest) <abrenon>iyzsong: thanks for the tip ! It's actually not in my user profile but I found it in the system's profile looking at the corresponding file <abrenon>I had never looked into the generated manifest, this looks useful <abrenon>oh, yeah, there's a %base-packages in there too??? ^^ <florhizome[m]>basically v40 still works with libsoup2 and if any input propagates libsoup3 it doesn???t work <civodul>mothacehe: i think you mentioned reconfiguring berlin yesterday <civodul>that leaves us time should something go wrong <mothacehe>i'll take care of it now. at start up there's some nginx/anonip sheneningans preventing nginx service startup <civodul>right, i changed the anonip service to DTRT <civodul>commit f83622f17d21d684e431fe14731349b952411368 <civodul>so if we're running a recent commit, the anonip will create the FIFOs in /var/run/anonip, *or* it will fail if those files exist but are not FIFOs <civodul>IIRC it takes up to 20mn (!) to boot due to Btrfs <civodul>oh i see a reproducer for the Parted bug, neat! <mothacehe>yes, i spent some times in gdb/libgc yesterday <mothacehe>i have a better understanding of the situation <mothacehe>but not of any obvious fix, i'll keep thinking about it today :) <mothacehe>just merged konrad guix-install.sh patch, down to 5 blocking issues for the release! <mothacehe>well rebooted 40 minutes ago and no sign of life <mothacehe>i'll see if i can connect to the serial monitor <stevenroose-phon>Hello. Yesterday I configured hibernate and this morning when booting my laptop, it entered GRUB rescue mode, saying /boot/grub/x86_64-efi/normal.mod not found. I set resume= to my swap partition and I have an encrypted /home <stevenroose-phon>No idea what to do, never used grub rescue mode before. Whatever I find online doesn't work, as it seems the grub files are just plain missing :/ <civodul>the other day bayfront took quite a while to come back <civodul>long enough for me to start panicking <stevenroose-phon>Hmm, `set` seems to tell me that the boot partition is `/` (the only btrfs one), but it should be another. <mothacehe>sadly i don't have a user account on hydra-guix-129 yet to access the berlin iDRAC console <civodul>cbaines: funny bug that you found with r-minimal cross-compilation <nckx>This has happened before, to apteryx & me. <nckx>We rebooted then. Let's reboot now. <nckx>stevenroose-phon: The ???unknown file system??? might just be due to missing modules (drivers) in the minimal rescue mode. <nckx>I'm not sure though. As a feeble compliment to my own admin skills, it's been many years since I've seen the GRUB rescue prompt. <nckx>civodul: I think that predated 6.0. <nckx>We couldn't reproduce it so chalked it up to goblins. <nckx>Well, also rebooting berlin for hours on end was an unattractive proposition after it was finally back up. <nckx>We'd need a twin that can safely access the SAN, if that's even possible. <stevenroose-phon>nckx but this laptop has been working fine before. I changed the kernel arguments to set resume= the swap partition. That's all <nckx>Can you manually insmod your way to a working GRUB? <stevenroose-phon>nckx, can I execute guix system reconfigure on /mnt somehow if I mount my partitions to /mnt and /mnt/boot? <nckx>So, this a thing that happens, apparently. This would be the second time. It has something to do with the QLA (fibre channel, I think) controller driver locking up. <civodul>so, shepherd heap starting at 21 MiB, let's see... <nckx>I haven't found other reports, but I didn't search hard. <nckx>stevenroose-phon: To reinstall GRUB? If so, I'd just manually reinstall GRUB. <nckx>(???or run dmesg on berlin. :) <nckx>Why is Enterprise Storage Technologies doing this at all. <nckx>Somebody poked the reset button an uneven number of times, when everyone knows it has to be an even number of time. Classic mistake. <nckx>stevenroose-phon: I'm sure you can bind-mount/chroot/guix-daemon & your way into a state where you can mostly run ???guix system reconfigure???, but this to me is overkill. <nckx>Adding or removing Linux kernel arguments does not affect GRUB installation at all. <civodul>nckx: i don't want to badmouth but there's always btrfs in the call stack :-) <nckx>stevenroose-phon: So while I can't say what happened, I can still say it's not caused by the presence of ???resume=??? itself. You could put old English poetry in there and GRUB would install fine. What might have happened is you added that string, ran guix system reconfigure, and then (for some other reason) GRUB mis-installed itself somehow. <civodul>mothacehe: actually i wonder if the anonip activation snippet is doing the right thing; had to manually rm /var/run/anonip/*.log because they were regular files (empty) <mothacehe>civodul: that's what i did last week, must be more to it i guess <nckx>stevenroose-phon: You're in a rescue system now, right? You can mount /boot. Does /boot/grub/grub.conf (1) still exist (2) look sane? <nckx>If so, running that grub-install command with the same arguments, pointing it to your mounted /boot, should just work. <nckx>Are you sure you mounted the right device? <nckx>Sounds like you mounted the ESP if it has (uppercase, vfat) \EFI. <nckx>If you don't have a separate boot partition, it'll be on /. / was encrypted, right? :-/ <nckx>So, separate from your ESP? <stevenroose-phon>I mounted the boot partition and it has /EFI. No idea what an ESP is.. :/ <nckx>ESP is the EFI System Partition, that contains EFI executables (usually boot loaders like GRUB). A boot partition is something else. It doesn't sound like you have one, so let's move on. <nckx>You should be able to do what you mentioned above: mount your broken root at /mnt, then mount the ESP at /mnt/boot/efi. <stevenroose-phon>K mounting thst one failed so that might explain. I guess the btrfs is corrupt. I'll look into that <nckx>There's a #btrfs channel if you think this might be recoverable/want to try anyway. <nckx>See, this is why I'm extremely hesitant to recommend things??? ??? <nckx>civodul: It's ???normal??? that it shows up here though. It's not purely the messenger (it's doing a lot of work and making a lot of choices) but it's also the messenger. <unwox>do you use ext4 or something else? <nckx>But I feel like I talk a weird amount about bcachefs here already. <unwox>i've been using btrfs for 3 years without a problem. but seeing today's discussions makes me think there may be more reliable options <nckx>unwox: I've been using it for about as long without problems ever since I gave up trying to use any of its ???raid??? features. My home server runs btrfs on top of md. The way I interpret the anecdotes here is that btrfs is stable in stable conditions, but curves sharply towards catastrophic failure when something else wobbles. <nckx>That's not a recommendation to switch. It's not a recommendation at all (see above ????) because it does offer some nice features. <stevenroose-phon>Yeah tbh I think it wasn't random. I had just configured hibernation and might have accidentally done a hard shutdown while either a hibernation wake-up or a hibernation attempt was ongoing <unwox>never heard of it tbh. i'll take a look, thanks <unwox>does guix system support it though? <nckx>stevenroose-phon: Exactly what I think happened. <nckx>What might have been a ???oops, not enough swap/some other configuration issue/your laptop firmware is the devil??? failure to resume, followed by a reboot, turned into corruption. <nckx>unwox: bcachefs? No, not officially. <nckx>Don't use it in production (that includes at home). <nckx>unwox: By ???switch??? I meant switch away from btrfs to another well-known one like ext4. <unwox>i'm not going to switch until btrfs fails me :) too much of a hassle <nckx>sneek: later tell the_tubular: Yes, file:line:column. But don't take the precision too seriously: it's the exact spot where Guile *guessed* the error might be. :) <Phil51>quick question - if I want to remove the authorization I've provided for a substitute server using: <Phil51>guix archive --authorize < server.key <Phil51>Am I safe to just manually remove the key from the acl file under /etc/guix? <Phil51>Or will this leave the system in an inconsistent state <Phil51>Note I have not installed anything from the server in question - only added the key. <nckx>Even if you had installed anything, ever, it would be fine. The store is trusted. Signatures are checked before putting things into it. <stevenroose-phon>That was fun. Just extracted my config file (that stupidly was still only on /etc/config.scm and not copied to my /home) from the raw btrfs device using grep ???? <stevenroose-phon>I guess I'm gonna cross my fingers that the /home is still intact and reinstall the rest and then restore my config file :) yay for deterministic systems I guess :) <nckx>stevenroose-phon: If it's any consolation, I've done that before (and thanked my lucky stars I didn't enable compression). I won't preach about automated back-ups, but??? automated back-ups. <Phil51>nckx thanks for confirming - so I just remove the s-expression for the acl file??? It will only effect downloading of substitutes going forwards? <stevenroose-phon>It has a variety of others, but invoking just fsck on an ext4 partition says no such file or directory even if the device exists <anschlussy>Wasn't hoping for this, but when I select "install using the shell based process" on my machine, nothing happens. <anschlussy>I turned off my monitor and turned it back on; that gave me the 'background' terminal output, but then put me back on the gnu guix system installer bit, albeit with a different option highlighted, but still unresponsive. <rekado>ACTION restarted the issues.guix.gnu.org sync <anschlussy>20 minutes left on downloading an installer from today instead of my "ancient" couple-day-old one <civodul>rekado: is there something we can do so it always restarts properly? <gnou_liber>Hello, I have two problems in Guix GNU with Linux-libre: 1?? There are Adwaita icons that the applications cannot find and therefore do not show them to me. To fix this problem I had to change the icons with lxappearance, but I would like to use the adwaita icons. 2?? Now I am using StumpWM, I would like to install some modules for this window manager that are available in the Guix repositories, but I don't know how to do <gnou_liber>it, on the GitHub page they say to edit the configuration file ~/.stumpwmrc, but I don't have this config file. <anschlussy>HAHAHA, my installer has network and is overall working <anschlussy>it just won't output to my display, suspiciously <anschlussy>unless the installer has sshd running or something <rekado>civodul: it needs RSYNC_PASSWORD to be set; we would need that to be loaded from a file <nckx>/bin/sh is bash on Guix (Systems). <nckx>There's also /usr/bin/env by default, which is the standard for writing vaguely ???portable??? bangs. <anschlussy>ah, dang, my hacky fun was interrupted by the newer installer finishing the download, lmao <anschlussy>This is an official problem, now: I can't get the shell based process to work <anschlussy>it just drops me to a shell without me being able to know if anything's working <anschlussy>guess i can load up the image and burn a raw disk instead of using ventoy to see if it's that, but i doubt it <nckx>Let me scroll up and read the actual problem though ???? <anschlussy>seems like when the installer switches modes out of whatever ncurses thing is going on, it's just not verkink with my display; this particular computer doesn't have any other output (xeon processor), so it's through an rx r80 <nckx>???Whatever ncurses thing??? = kmscon. <nckx>But??? why would it not use that for the other TTYs. <nckx>But cool, great to know that your machine would *not* work with nomodeset, so there's literally no know-good fallback option that works across vendors. Truly excellent news. <anschlussy>for the record's sake, #linux told me to try ctrl-q (no work) and "echo ^V^[c" (or whatever the common terminal reset tech is, also no work), and it does indeed give me a terminal, as I can type reboot, etc <anschlussy>is there a way i can ssh in? i just need to complete the install <nckx>You need to set a root password first. <nckx>I guess you could do that blindy. <anschlussy>i did see some mention before the installer set it all up; oh what the hell, now flipping my display off and back on doesn't give me the output from before the installer froze up <nckx>Beats manually typing in ANSI reset codes, which is wild even by my standards. <nckx>(I cannot imagine that having worked.) <anschlussy>i think i have set the root password; what's the method to start up the ssh beast <nckx>It's running by default on port 22. <anschlussy>so let me think this through while i'm already doing 8 other things: it's running on port 22, but until i set a root password, it's not gonna accept login, so i couldn't have any time to mess with anyone else's installs? <nckx>Not unless they passwd root hunter2. <nckx>I hope it hasn't bitrot. <anschlussy>especially in the time it took me to copy it over :^) <nckx>Quick, switch to a VT to see if it's running ??? Hm. Not sure what to try. <nckx>I'm downloading an image myself but it's 5 minutes out. <anschlussy>no problem; you don't have amd hardware to test on, do you? <anschlussy>since i'm new to this, is there some method i can use to "guarantee" an sshd start? <nckx>Well, ???herd restart ssh-daemon??? would try to restart it if it was down, but if that fails it fails. <nckx>Since you're blind: I'll note that???IIRC!???the other VTs have a ???hit return to log in??? mechanism. <nckx>Whatever. So if you just typed ???passwd??? right away, it might've been swallowed. <pkill9>i don't even use nautilus but I don't like when desktop environment programs require any debugging cos they're supposed to be core things <anschlussy>nckx: I was able to ensure that after I typed passwd and entered the same word twice, that I had terminal control again, using the ICMP-ping-myself-and-check-wireshark method. <anschlussy>i should reclarify because i'm unsure if i properly expressed it: this box has video output, but it's just through the rx 480; i hope it wasn't interpreted earlier as it being some insane vnc thing <pkill9>now I'm hpapy that it's reported and looked into <nckx>Huh. Something to debug later, I guess. <nckx>Maybe the network came up too late (& we should wait). <anschlussy>i'm slightly worried that the sshd having a hard time might have been related to other things being busted <nckx>Since it happily restarted I'm not too worried. Nastly little timing bug, probably, nothing worse. But I guess. <anschlussy>timing bug for the ssh, but the fact that the installer hates changing out--i'll check real quick if the graphical path still works, though i need the shell mode <anschlussy>update: going the normal mode doesn't break anything, so far, in the installer, so just the other, necessary for my purposes mode. <anschlussy>Oh, I see, so when I run restart, it just starts it, lmao <nckx>I'm used to an unofficial installer that runs it by default, sorry. <gabber>what's the source/origin method to specify a local directory or file as a package source? <tricon>gabber: `(source (origin (method url-fetch) (uri "file://path")))' <tricon>also works with `git-checkout' as the source. <gabber>tricon: does that expect a tar file? or a directory (already extracted tar)? i get `guix build: error: fport_write: Connection reset by peer' <anschlussy>stevenroose: what was your workaround for not having mkfs.ext4? <tricon>gabber: i'm presently using it with a directory. haven't tried tar, though i would expect that to work as well. <anschlussy>makes me wanna fool around with btrfs again just to avoid manually invoking anything <nckx>[Resending some lost messages.] We'd have to run the getty in kmscon on all consoles. And hope that the hand-over is smooth, because IIUUC, once you lose video, it doesn't come back even when switching back to tty1? <nckx>An alternative (although we could do both) is to have that menu item actually spawn a shell on the current VT instead of switching to #3 as it does now. But none of the other VTs would work, including the manual, which is quite limiting. <gabber>huh. now the guix daemon is getting killed due to "Out of memory" (this causes the fport_write) <nckx>If the installer has /run/setuid-programs in PATH, one unholy work-around is manually locating mkfs.ext4 in /gnu/store and copying it there, then starting the installation. <nckx>Yes but also Gu{???}uix. <nckx>Somebody complained about locate not working, once, but who knows where they were trying to stick it. <anschlussy>maybe they didn't update the db first; perhaps they weren't in the installer? <gabber>how can i call an executable that resides in the source tree of a package? is (invoke) the right procedure? i always get error-code 127 <tricon>gabber: i use `invoke', yes. ensure you are in the correct dir. <gabber>do i check for the correct dir by displaying (getcwd) within an installation phase? <tricon>gabber: that's usually what i do, tbh. <gabber>this is weird. i check the path, i copy-pasted the file-name, i re-check (with --keep-failed) but i still get error-code 127 <tricon>gabber: what is the program you are attempting to invoke? <gabber>not sure if i'm allowed to say it out loud in here... some (giant) setup-program for a hardware-synthesis-suite <anschlussy>what's actually installed on a system if you don't choose things like xfce? <nckx>anschlussy: %base-packages and %base-services. <anschlussy>and if i want to modify that kind of thing to remove a bunch, i'd just modify it with the whole modify-inputs method? <apteryx>can Guix be installed in Docker and expected to run there (with daemon and all?) <tricon>gnucode: oh you know, just dreaming the dream. <nckx>anschlussy: modify-services (see manual) but yes. For packages you can use (rnrs lists)'s ???remove??? procedure: (remove sudo %base-packages). There might be a more user-friendly way. <civodul>apteryx: yes, but you need to do "sudo docker run --priveleged" or something along these lines <apteryx>OK! The user at the link above was trying to reproduce everything from a Dockerfile (e.g. at the time of running 'docker build') <civodul>apteryx: i think it can work, but they won't be able to run guix-daemon unless docker runs with --privileged, IIRC <zpiro>it's in preprint, and letters of intent matters as something to latch onto, if just a pragraph can be added. <zpiro>civodul: at pages 170 - 172 interest in baking reproducivility into the frameworks and software packages with open science activity widely supported. all else is physics regions of interest into the future. <stevenroose-phon>This is pretty annoying.. If you reconfigure and add both a new substitute server and new derivatives/packages, the new packages won't be downloaded from the new substitute server yet. <nckx>stevenroose-phon: Good point. <nckx>The obvious answer is to do that in two steps, but it's not great. <stevenroose-phon>Yeah s?? I just did a reinstall and replaced the config with my old one and suddenly I'm building the Linux kernel ???? Gave up and did a two-step indeed. <Phil51>FAIL: tests/guix-environment-container.sh <Phil51>+ guix shell --ad-hoc guile-bootstrap <Phil51>guix shell: error: ad-hoc: unrecognized option <Phil51>accepted connection from pid 998969, user foo.bar <Phil51>guix shell: error: package intelmetool@4.7 does not support armhf-linux <Phil51>It tells me to report these to bug-guix - but wanted to check first if anyone knew if this was a known issue? <nckx>So there is an error somewhere, but note that some tests test for errors, like the ad-hoc one, so the error is expected and not a failure. <nckx>Something else must've failed? <Phil51>============================================================================ <Phil51>Testsuite summary for GNU Guix UNKNOWN <Phil51>============================================================================ <nckx>???yeah that was going to happen. <nckx>Phil51: I'm going to let your client flush its flood-prevention buffers and then unquiet you if litharge still forgets to do so. <nckx>In future, please post multi-line (>3 or so) messages at paste.debian.net. <nckx>litharge is still buggy :< <nckx>Phil51: If you said anything after ???# SKIP:?? 29???, say it again. :) <Phil51>After that was just XFAIL: 4, FAIL: 2, XPASS: 0, ERROR: 0 <Phil51>Sorry for posting too much text :-) <nckx>Both the ad-hoc and intelmetool tests explicity test for failure, so failure is success, even if the error message is logged. <nckx>It's OK, it happens to everyone, ideally once. <nckx>Could you paste.d.n the entire guix-shell.log? <unmatched-paren>nckx: Maybe litharge should display a message somewhere along those lines if it detects a flood? <nckx>I've asked about that though. <nckx>PSA: there are no Guix bots besides the entirely passive goggles-bot. <unmatched-paren>nckx: what's goggles-bot supposed to do besides being entirely passive? <apteryx>podiki[m]: we're missing just one vouching message for you, unless I've lost track of it (you have ludo's and my backing) <Phil51>nckx - did my paste.d.n make it to the chat - my connection died as I was posting them <mitchell>Hello guix, I am trying to create an installation image for my pinebook pro which contains the deployment image and a script which knows how to format/encrypt the internal emmc and install it. Basically run `guix system reconfigure` but I don't want it to compile or download anything. <mitchell>It feels like it should be possible but i'm unsure if there is a mechanism in place for this kind of thing or if it'll just be something i have to hack together <nckx>Phil51: Nope. Was just wondering what happened to you. <Phil51>very bad timing losing my connection that's all :-) <Phil51>Not sure if it's relevent but I tried the make authenticate command to see if that was working <nckx>Phil51: Are you running ???make check??? in a ???guix shell -D guix???? <nckx>That is required but not clearly documented. Someone was going to submit a patch, but I haven't seen it. <nckx>???Someone??? not being a euphemism for me, for once. <Phil51>guix environment guix --pure --ad-hoc help2man git strace <nckx>Right. That's equivalent. It's in the manual, but easy to miss if you skip straight to that section. <nckx>Could you try ???guix shell --pure -D guix??? just in case? <nckx>Damn thing keeps passing here, sorry. <nckx>Full command: guix shell --pure -D guix -- make check TESTS="tests/guix-shell.sh" <Phil51>Same error here guix shell: error: package intelmetool@4.7 does not support armhf-linux <Phil51>The guix channel we use is slightly behind the head.- wouldn't expect that to make a huge difference? <Phil51>Apart from that I don't think my setup is anything out the ordinary <nckx>I don't know. I'm not aware of any recently fixed bugs. <Phil51>OK cool - good news is I don't seem to be doing anything fundamentally wrong :-) <Phil51>I can try on my laptop which I can pull in latest guix to remove that as a potential issue <nckx>Before I go: don't focus just on those above errors. They are expected! Look for others. <Phil51>Cheers for the help nckx - ok will do, I'll have a go on my laptop when I get home! <podiki[m]>apteryx: thanks, saw your email! Jonathan is my third, but it has been a while and I said I would email him when I got to this step. So I'll do that and send a signed email as well to the maintainers <awb9999>My desktop pc running guix with gnome goes to "suspend" after 30-60 minutes. In the desktop configuration I have disabled suspend. I tried to find gsettings to verify, but it seems that gsettings is not packaged on guix. Is there any other way how I can find out what is going on? <pkill9>i used that to change those settings <pkill9>these small bugs with gbome sadly seem commonplace, it's a sham e, but I guess that's the twrritory with running a niche distro as guix <pkill9>which is annnoying because the benefit of a desktop environment is it doing exactly what it implicitly claims to <pkill9>without error, but oh well, for the most part it works well <pkill9>even bluetooth works well,and that's always a pain point <pkill9>i managed to get my phone outputtting it's audio through my alptop speakers just by using gnome's bluetooth interface <pkill9>i.e. it Just Worked(tm) with no command line configuration etc <pkill9>so it just needa more ironing and resources <awb9999>so dconf I can use to query the settings? <awb9999>But to run guix on a desktop is very painful. <awb9999>Not only this little glitches that make it hard to run a machine. <awb9999>But also the packages change all the time. <awb9999>I was using a usb wifi driver that recently was removed from guix, <awb9999>and that has not yet been added to nonguix. <awb9999>So it is difficult to run a machine that once was running. <awb9999>It seems dconf and gsettings are made by gnome. <awb9999>But it seems that this tools are used also in other desktop managers <awb9999>or will I have to run them every time that I reinstall / upgrade the os? <awb9999>I have installed several services on my machine, <awb9999>and they sometimes loose the config when I do a guix system reconfigure. <awb9999>and with syncthing I have to authorize the machines it should connect to via web ui. <awb9999>and sometimes I have to add the authorizations again. <awb9999>it does not happen with every reconfigure. <unmatched-paren>awb9999: are you sure you don't need to add the authorizations in the (syncthing-configuration ...) somewhere? <awb9999>could be also guix home reconfigure. <attila_lendvai>emacs-magit should break the guix install of many other people... yet, i don't see anything among the issues. could my environment break it somehow? <attila_lendvai>it fails two tests (magit-git-error "clone of '/tmp/guix-build-emacs-magit-3.3.0.drv-0/magit-Qm34iP/remote' into submodule path '/tmp/guix-build-emacs-magit-3.3.0.drv-0/magit-Qm34iP/super/repo' failed (in /tmp/guix-build-emacs-magit-3.3.0.drv-0/magit-Qm34iP/super/)") <attila_lendvai>it's fishy though that it's not in the substitutes, and the home reconfigure tries to build it locally <attila_lendvai>hrm... i started a guix pull from my local guix repo checkout, and it may have dirtied that pull...? i try to pull again, luckily there's a new commit <attila_lendvai>ACTION pulls his guix checkout and tries to `./pre-inst-env guix build emacs-magit` there <unmatched-paren>((options->transformation '((without-tests . "emacs-magit"))) emacs-magit) <podiki[m]>indeed, can't say I've seen it specified otherwise (I've use /var on both foreign and domestic) <stevenroose>Where can I find the default directory structure that Guix puts in a new user's home directory? <stevenroose>I seem to have accidentally removed some PATH statement.. <stevenroose>I remember there is some PATH I think to all the bins in the store <stevenroose>sourcing ~/.guix-profile/etc/profile doesn't seem to give me access to all bins <stevenroose>Hmm, well this is about pinentry specifically maybe. I had pinentry in my config file, configured and booted with it. Invoking pinentry worked in terminal, but gpg was unable to find it. Then when I ran guix install pinentry, it suddenly worked, so it seems like there is some discrepancy between system-installed packages and user-installedo nes. <podiki[m]>can confirm emacs-magit is failing two tests, I guess with the git upgrade <jackhill>oh, looks like the Jami hash changed: sha256 hash mismatch for /gnu/store/aqai10zycvj8d91ndxzg2gcmqk0gs9p0-pjproject-jami-2.12-0.5e478bb-checkout. Does anyone have the old hash to compare: 1vzfpiwhd96a9ibk398z922a60j18xd7mblsmi6355r7ccj2aw7p ? <old>jackhill: Maybe apteryx <silicius>does anyone use emacs-guix? It hasn't worked for me in a long time apart from maybe searching for packages by name <tricon>i think so. i used to have issues with it. what is an error you're receiving? <apteryx>jackhill: works for me, I ran 'guix build pjproject-jami -S --check' <apteryx>ah, but I'm on an unpushed commit that updates jami <silicius>tricon: in shell, after typing guix it shows a buffer with 'Unknown # object: \"#~\"' in emacs-guix guile repl <apteryx>jackhill: nope, works as of guix 807bfe55dc382b88ab0fdade6886dfa2b15487ee (about a week old) <attila_lendvai>unmatched-paren, unfortunately i cannot make that work. it's a spcification list (string package names). i added a specification->package call, and changed some of my package list merging code to tolerate packages, but it still tries to run the emacs-magit tests. <attila_lendvai>unmatched-paren, i replaced "emacs-magit" with: ((options->transformation '((without-tests . "emacs-magit"))) (specification->package "emacs-magit")) <jackhill>apteryx: huh, I just pulled to commit c52cdd18d6f8dddd69dfb1f8b5214a0db6ebb562. The jami service is what pulled in the build <silicius>tricon: but looking at other stuff, I think that's my only problem with it now. And it's that error in shell that's breaking emacs-guix all the time <tricon>silicius: you're referring to tab-completion, yes? i actually am receiving the same thing. what i use is the Hydra-esque UI in Emacs: M-x guix <attila_lendvai>unmatched-paren, but then i had a (map (compose list specification->package+output) ...) which i had to augment with a (if (package? x) (values x #f) ...) <attila_lendvai>unmatched-paren, maybe something depends on emacs-magit? and the build is triggered by that dependency? <unmatched-paren>Building the following 21 packages would ensure 25 dependent packages are rebuilt: emacs-magit-org-todos-el@0.1.2 emacs-magit-annex@1.8.1 emacs-magit-gerrit@0.4 emacs-vdiff-magit@0.3.2-8.b100d12 emacs-consult-org-roam@0.1-0.9572c5b emacs-consult-bibtex@0.1 emacs-citar-org-roam@0.3 emacs-org-roam-ui@0-0.c75fc75 emacs-nix-mode@1.4.5-0.8fe2ccf emacs-magit-todos@1.5.3 emacs-treemacs-extra@3.0 emacs-orgit@1.8.0 emacs-straight-el@0-0.fed2153 emacs- <unmatched-paren>ement@0.4.1 emacs-git-email@0.2.0-0.b5ebade emacs-tokei@0.2 emacs-magit-svn@2.2.3 emacs-message-view-patch@0.2.0-2.40bc2e5 emacs-repo@0.2.0 emacs-forge@0.3.2 emacs-piem@0.3.0-0.0a571f4 <attila_lendvai>yeah, the build list has two emacs-magit entries: /gnu/store/zml9vaxd7q0bs622caxmp5q5gkh533nc-emacs-magit-3.3.0.drv <attila_lendvai> /gnu/store/kz210k13d9gd2vmv1nxvch3bhxz9n122-emacs-magit-3.3.0.drv <unmatched-paren>you'll need to add that options->transformation wrapper to all emacs-magit's dependents too <silicius>tricon: thanks for telling me it works, apart from that one thing. <silicius>Would be nice if the completion worked though. <attila_lendvai> /gnu/store/kz210k13d9gd2vmv1nxvch3bhxz9n122-emacs-magit-3.3.0.drv <attila_lendvai>(sorry about the noise above) it works now. removing emacs-nix from the package list did the trick. <unmatched-paren>attila_lendvai: you can use ((options->transformation '((without-tests . "emacs-magit"))) emacs-nix) <apteryx>jackhill: oh wow! erroneously updated in my commit 071d4bdc80cbf6b3a5c1232f0d704878431a864f. <apteryx>I've pushed 1b22c03cca that should fix this, thanks fro reporting! <stevenroose>Hmm, I seem to misunderstand the concept of a "package". There is this thing in my store called rust-src and I saw that installing the "rust-analyzer" package added it by saying it would download this thing: "rust-src-1.60.0 18.0MiB". But that doesn't seem to be a "package".. So what is it? <stevenroose>I spent some time before looking for packages named `rust-src` using `guix package -s` but without success. But it does exist and resides in my store. <nckx>Packages are hidden when they are not considered useful on their own. Can't speak on rust-src specifically. <unmatched-paren>a package is merely one kind of "file-like object", which is an umbrella term for "anything that has a ``gexp-compiler'' defined, and can thus be lowered into a derivation and ultimately a store path" <stevenroose>unmatched-paren: that's unfortunate.. if I consider it useful, can I make an MR markiing it unhidden? <stevenroose>Hmm you might be right, I was under the impression that it stored the documentation too <nckx>That does appear to be the case. <nckx>I was replying to (. If there are hidden gems in rust-src not available any other way, yes, it makes sense to expose them! Better throug a ???rust-documentation??? package or similar, though, that puts them in /share/doc, without all the source baggage. <stevenroose>Any rustafarians here that know of a rust-doc package? I can't seem to find one now that I'm back at square one after discovering rust-src :) <unmatched-paren>stevenroose: Heh, rustafarians. Isn't the term usually "rustaceans", though? :) <apteryx>what web server do you use for your own pastes? <nckx>stevenroose: ???is the stuff in (I can't believe I'm about to type the following string) /lib/rustlib/src/rust/src/doc/book/listings/ etc. actually documentation? <nckx>I can make little of its contents. <TopExpert>nckx: it's the book available at rust-lang.org <nckx>I mean, not in this sorry state. <nckx>We package SICP, though, so a built version would be fine. <TopExpert>I gave up on trying to build rust before in guix, are you working on this <nckx>I'm no masochist. This applies as much to Rust as to MarkDown. <unmatched-paren>TopExpert: but rust is already in guix; what do you mean by "build rust"? <TopExpert>unmatched-paren: well rust developers require the full platform <nckx>I fix Rust issues when they are in the way, nothing more. <TopExpert>the thing is, from my experimentation, that it needs to be build with llvm for wasm32 cross-compiler to work <TopExpert>gcc doesn't produce javascript, so it needs to use the llvm linker and archiver etc <unmatched-paren>but rust is always built with llvm, unless you use the experimental rust_gcc_codegen :) <TopExpert>it's built using the c compiler, I believe it has its own llvm for codegen <nckx>TopExpert: Even in Guix? <TopExpert>but the standard libraries are packaged with gnu ar, and those can't be used by wasm32 <TopExpert>something about "archive has no index, run llvm-ar to add one" <TopExpert>in my code links above I tried creating a new config.toml for the rust build from scratch, adding all the features etc, and even nightly, but as it takes hours to build I had to give up trying <TopExpert>unmatched-paren: yeah, but it needs llvm-ranlib <TopExpert>nckx: are you exploring the rust source code <nckx>TopExpert: No, I was just reading the Guix rust-x.y packages. <TopExpert>booted up 48 cores and thought let's do this, after 10 hors or so I discover I'm not really a masochist either <nckx>And the string ???Use the system-provided LLVM.??? And just in general, those kind of NIH shenanigans are frowned upon in Guix. <nckx>ACTION looks at their cores. They are few, and smol. <stevenroose>nckx: it actually looks like a partial version of one, yeah <stevenroose>It's part of the rust-book it seems, but only one chapter <nckx>Hah, I had the same thought, but that's because I thought I was looking at test suite data. Then I was not sure sure. <stevenroose>Looking for all html files in there, the stdlib docs are not part of it <TopExpert>the rust packages for guix are for producing system packages imo., one should use crates.io when coding in rust <nckx>Good luck with that for now. <TopExpert>like dependencies for OS programs, that don't need to be updated always <stevenroose>TopExpert: the default setup is to use rustup which is not available on Guix. But honestly most things are available in the guix repos I think. Except for docs, at least for what I need. <TopExpert>would my build have worked, using llvm, we would have had the full rust platform <unmatched-paren>it does on nix, but that's because they special-cased nixos in the rustup source code <TopExpert>rustup just installs "supported" toolchains, and they are all built from that source <lilyp>"one should not use distro tooling but our own package manager that can't even handle two thirds of the common use cases" is what got us in this mess to start with <stevenroose>yeah I agree we can do without rustup especially since guix can so easily manage different version of rust as well <TopExpert>I'd really want to use Guix but as you understand, not having access to the major programming language is quite a blocker <stevenroose>some small utility linking between different versions should be enough <tricon>stevenroose: call it "Rustoleum". <lilyp>major sounds like a very ungooglable name <lechner>Hi, is there a guide on how to use podman or docker in Guix for a container with a different OS? <lechner>With podman I get Error: open /etc/containers/policy.json: no such file or directory <nckx>I can never reproduce these emacs-guix problems. <TopExpert>lechner: sounds like a security feature you can disable, the policy.json is for image signatures <TopExpert>lechner: ah it's for pulling images from redhat.com etc, so you have a policy.json specifying public keys, you should really spend time learning these tools before using them <tricon>nckx: you are One with the Guix. <nckx>I'm in fact borderline incompetent at Emacs, is the funny part. <nckx>My average emacs session is a fuzz test. <nckx>But emacs-guix has proven immune to my powers. <tricon>"My average emacs session is a fuzz test." ???? <nckx>Speakinowich: how do I swap between an existing vertical/horizontal 2-window split? That is, make the result of C-x 2 look like I pressed C-x 3 and vice versa. A trivial operation, but Emacs' self-documenting features are toying with me. <nckx>That some fancy evilthing. <nckx>tricon: So it's not built in at all? <nckx>I feel mildly better about myself. <podiki[m]>as a heavy emacs user I also feel better that I always did that manually <nckx>s/A trivial operation/Advanced voodoo for true hackers/ apparently. Weird. <mekeor[m]>hello. does guix system support full-disk encryption via cryptsetup, including encrypting the swap-partition, while still keeping suspend-to-disk possible? <mekeor[m]>and second question: (how) can i invoke the graphical installation of guix system from guix system (i.e. not from the installation image)? <apteryx>fun, qtcreator seems relatively easy to package <nckx>mekeor[m]: (1) No. (2) This is extremely unsupported and very likely to fail on some assumption, or worse, but: echo -e ',use (guix)\n,build ((@ (gnu installer) installer-program))' | guix repl <nckx>Here, right out of the gate, it tries to write to some /proc file that doesn't exist and dies. <nckx>Use the installer image. <jackhill>now I'm waiting on CI to build libjami. I see it's scheduled, but curiously most of the works are idle <nckx>I still don't trust it 100% but that's my problem, not yours. <nckx>(That is not a comment on the quality of the commit; just on the complexity of $everything.) <mekeor[m]>nckx: thanks. but manual installation (i.e. non-graphical) should work from guix system, right? <nckx>Oh, they're here. jackhill: Have you reliably hibernated to encrypted swap? <nckx>mekeor[m]: I use it regularly. Consider it reliable. <nckx>Just skip live-image-only steps live cow-store. <mekeor[m]>reading some arch-linux-wiki-articles on hibernation with encrypted swaps, i get the impression that an encrypted swap-file is more likely to work than a -partition <apteryx>jackhill: odd, that it takes so much time <nckx>mekeor[m]: I can't immediately see why that would be true with FDE, but it's possible.