<Telior>is there like a guix way of installing quicklisp, or just following the regular instructions on the page? <jboy>rekado_: I saw that you initially added python-igraph and python-leidenalg. The packages don't currently work because two tests for python-igraph don't pass. <leoprikler>that said, given its function as a package manager, you'll probably want to build a build system around it (perhaps?) <Telior>packaging it would add it to the other guix packages, right? the ones you just "guix package -i " install? <leoprikler>it appears you can build it using asdf-build-system, but you'll have to work out the specifics, since the site doesn't tell you the magic behind that <Telior>k, I'll check out asdf and see if I can build quicklisp from there, ty! :) <wheeler>What's the best way of iterating when packaging? I'll try guix package --install-from-file=best-package-ever.scm ... wait ... view logs with vim, etc. <wheeler>Is there a way to have the log output to stdout? For now I've been viewing the resulting /var/log/guix/drvs/...-bpe-0.7.3.drv.bz2 with vim; but I'd love to find a more convenient way to see where my build is failing <wheeler>What do guix packaging experts normally do? <lfam>Use `guix build` instead of `guix package` <lfam>You can also do `guix package --verbosity=9` but it's not productive to actually install the package in most cases <wheeler>lfam: ah, that's much better (`guix build -f my_package.scm`); thanks! <dalas>i got two queries as a novice user of guix <dalas>anybody know how to configure a desktop environmen specifically LXQT? <Nios34>Is ci.guix.gnu.org or ci.guix.info working well? <Nios34>These days, Guix's service seems very unreliable. :-( <bandali>(besides the current ci.guix downtime) <bandali>only issues.guix has been flaky for a while <bandali>but that's not a big deal for me since i use emacs-debbugs anyway <commanderkeen>what is the proper way to get a window manager working with guix? I'd like to get stumpwm running <vits95>Hi. I'm going to install Guix soon to see it in its "native environment". Hope it will be funny! <commanderkeen>never mind. i got it to work. it appeared in gdm after restart <vits95>BTW: LVM support isn't there yet? <vits95>i'd unpluged mouse. And have no idea how to reach the link wr'd posted with keyboard with qutebrowser... <bandali>indeed, no lvm support in mapped-devices in config.scm yet <vits95>I was need to leave some place unpartitioned, then! <str1ngs>vits95: does qutebrowser not have link hints? <str1ngs>vits95: f show show link hits I think <str1ngs>ironically I'm just working on nomad's link hints :) <vits95>strings it should, but not this one. <vits95>also i haven't see the hints over the download button in the bottom-right corner in this irc. <str1ngs>do you want to install guix system aka the ISO? <vits95>yes. just i'm waiting for my 2 cores oldie CPU make a squashfs from my /home. Then there is a transfer to another disk. Just i has LVM ;) <vits95>cp -r makes fun of Cyrillic letters today. fun for the whole family! <Nios34>I run out all of my space when install guix. Can i mount /gnu by my self? <str1ngs>vits95 cp -a might be better it preserves more :) <str1ngs>Nios34: do you run out of space while installing? <str1ngs>if your new partition is /mnt you can enable cow with . herd start cow-store /mnt <commanderkeen>so leinigen (lein) isnt available in the repositories. so im guessing i'll have to make a package for it? <str1ngs>Nios34 maybe because you did not enable cow before installing. the ramfs is now full <str1ngs>vits95: also rsync is even better for many files or large data transfers. plus it's you can stop and resume it <NieDzejkob>commanderkeen: there's a pending patch on guix-patches IIRC, check the archives <NieDzejkob>I wonder if guix system init could warn when the store is in a tmpfs <NieDzejkob>so, I think I'm at the point when I'd like to help review some patches. How would you (as in the commiters) feel about granting me commit access? (for reference, I'm Jakub Kądziołka on the ML and git) <vits95>strings: i'd joyed man of rsync. But somehow it isn't works smoothly with my FAT32 thumb-drive. i'll just pack all in squashfs for now. <vits95>"rsync: chown FILE failed: Operation not permitted (1)" ;) <vits95>I was forget about this touchpad. So used to mouse! Will see the link above... <str1ngs>vits95 fat32 does not have permissions <str1ngs>it's better to use ext if you need permissions or worst case ntfs <vits95>indeed: "File too large!". This machine is cleanly trolls me. I'll use gnome-disks to write iso of my /home to this thumbdrive! <str1ngs>if you don't need windows to use the thumb driver just format it to ext4 :P <vits95>;) My rage makes me dd if=vits.sqfs of=/dev/sdb status=progress ... muhaha! <alextee[m]>what does this mean when using git repositories in apckages? <alextee[m]>what does this mean when using it repositories in packages? <alextee[m]>fatal: dumb http transport does not support shallow capabilities <alextee[m]>probably means it can't find the version so it has to download the whole repo <tuete>Hej. I am using FDE with separated boot. It only works, when I copy bzImage and initrd.cpio.gz to /boot manually. I'd like to fix that asap. How can I search and deploy items from /gnu/store in guile scheme? <tuete>Somehow like I would do it in bash: "find /gnu/store/ -wholename "*initrd.cpio.gz" -exec echo cp {} /boot/initrd.cpio.gz \;" <tuete>It should behave the same way on the installation image instance and the later system installation -> /mnt/boot/ vs /boot <Nios34>Guile task don't use CPU, Is it normal? <mbakke>NieDzejkob: I'd be happy to vouch for you, as per the new commit access guidelines. :-) <g_bor[m]>I have a service that can now start postfix! <g_bor[m]>It is very much a work in progress, I intend to push it to wip-postfix soon. *janneke added --repl to u-boot spec for pinebook pro...now what? <janneke>ah, --system is missing from command line *janneke gets ready for next try... <jlicht>janneke: glad to be of service :P <janneke>*woot* /me booted guix on pinebook pro (still using original, pine64 provided kernel) <g_bor[m]>I've just repushed fixing some warnings and making start and stop reliable. <janneke>g_bor[m]: i'm really looking forward to checking this out! *janneke gave up running their mta a long time ago <g_bor[m]>:-) at the guix days we discussed that having this would be nice. <g_bor[m]>It seems the postfix is great in security. <g_bor[m]>Almost everyone running a mail server used postfix there, some even having their last machines not using Guix System for this. <janneke>g_bor[m]: yes, i was quite a fan of postfix on debian <Nios34>How to skip the make check or only install pre-build? <g_bor[m]>Currently it is very limited, as in you can't configure anything yet, but getting this up and running was the hard part. <g_bor[m]>Adding single config settings from now on should not be a problem. <Nios34>Can I just use "-K", and kill the testing task? <Nios34>Some packages like mariadb take such a look time to do check:-( <g_bor[m]>Nios34: you add tests? #f to the arugments of your build system. <g_bor[m]>Nios34: one of the most problematic is git. <Nios34>I want to skip the check. I hope I can add -K option, and kill the check task, then move forward. <g_bor[m]>The problem is that we can't split the check, as then unchecked packages would get added to the store, breaking some guarantees. <laloge_h>Hi! Is there a way to fetch a package description, i.e. the Guile code to build the package? I couldn't find this anywhere. <g_bor[m]>Nios34: you can try modifying the package, so that it does not check, but everything depending on it would have to be rebuilt. <g_bor[m]>laloge_h: you can find the derivation that is responsible to build the package. <g_bor[m]>The build logs of the package are in the directory indexed by the derivation. <Nios34>I have no idea about modifying the package while running "guix system init" <g_bor[m]>Look for something in the store like package.drv <g_bor[m]>laloge_h: but you usually want the package receipe, which can be found by guix edit package <laloge_h>g_bor: Thanks, that exactly what I needed <g_bor[m]>What I would do is init a small system, then set up a git checkout there, edit the packages and use pre-inst-env to reconfigure <Nios34>hmm, how about just go to the workspace and modify the makefile <dal>good Morning, anybody knows why Guix warns about "fail installing locale" ? <laloge_h>dal: Install glibc-utf8-locales and set your env variable $GUIX_LOCPATH to $GUIX_PROFILE/lib/locale <dal>yes fresh, laloge_h thanks I will try that <laloge_h>(With GUIX_PROFILE set to your profile directory, probably $HOME/guix-profile) <dal>btw, I'm mostly an arch user, is there a way to get packages not in the package list just like the AUR? <Nios34>I just kill the check task, and it moved forward <bricewge>dal: Yes we have channels, but it's decentralized unlike AUR. <dal>I heard of them but the person I heard it from was not clear enough, but you clear me up <pkill9>or it could be a single person's channel <dal>oh, so it's like i can host my own channel with packages and people could pull them from my channel? <bricewge>And you can even publish binaries of your packages as substitutes. <dal>nice I will read more on that, and I have seen one that says that some of the channels cannot be promoted on official Guix discussion platforms right? <bricewge>Yes we ought not to talk about proprietary software on the official Guix communication channels (IRC, Mailling List). <apteryx>fftw test suite really takes a long time to run <laloge_h>How can I know the sha256 of a Git repository, when using git-fetch? <laloge_h>Or alternatively, how can I make `guix download` download from git? <shtwzrd>bricewge: does that include talk about the mainline kernel? I feel like I've seen several unofficial channels with mainline kernel package definitions and it seems like kind of a waste with the duplicated effort. Are any 'official unofficial' channels out there we should try to support? <wingo>i think you have to do a "git checkout" then "git hash -xr ." in the checkout <shtwzrd>:laloge h a lot of git hosts, including github and gitlab, have endpoints you can call to download the repo as an archive <laloge_h>Yes, but the one I try doesn't for non-release version <shtwzrd>Oh, it doesn't take a commit ref as a query parameter or anything? :/ <bricewge>shtwzrd: That I don't know, you should ask a more GNU minded person. <bricewge>Teaching `guix download` about git repository would be a usefull feature! <shtwzrd>Ah, I see. I don't know what the 'correct' way to get the hash of the git repo. The 'wrong' way (as in, what I do) is to leave the sha256 empty, invoke the package definition, and let guix tell me what it thinks it is. <dal>oh btw is there a way to get a faster start up in guix? <kahiru>hey, anyone managed to run guix on odroid c2? <bricewge>kahiru: Check the commit 2bb915e679b8a9e071f15b4caa3274fe9c6396c1 <bricewge>It's u-boot package got removed from Guix because « [it] requires a binary blob to boot correctly ». <shtwzrd>janneke: congrats with the pinebook. :) I'm trying to do the same. Want to do the manjaro kernel and bootloader too <janneke>shtwzrd: ty, i'm green to arm and got a lot of help; but learning now <str1ngs>janneke: oh I ordered a pinebook pro too <joshuaBPMan>Anyone know what the status of guix and wordpress packaging? <shtwzrd>janneke: i'm green as well, been trying to psyche myself up to go ahead and boot guix with this thing :) Do you remember anything tricky you needed to do? I'm mostly concerned about the bootloader <dal>is the MySQL package the cli only or I'm doing something wrong? <dal>I cant find it in anywhere other than running mysql , and gives me an error <dal>(dont remember the error, but is a long the lines of not finding a server) <leoprikler>dal: I'm pretty sure you want a service for mysql the server <NieDzejkob>alextee[m]: I'm reading the backlog, so I don't know whether you got an answer, but the issue is that git has two possible ways of sharing a repository online. The more common one is that the server knows about git internals. It seems that you stumbled upon a URL that's just a .git folder put on an http server. The latter doesn't support shallow clones, which is what guix uses. Shallow clones create a repository without the commit history. <joshuaBPMan>Hey guix, my tor service is not starting...how can I fix this? <joshuaBPMan>where can i find more messages about what the problem is? <leoprikler>probably /var/log/messages, but that depends on how tor actually does logging <leoprikler>you can also try manually starting it after the fact, since network services have a tendency to not be ready when they should be started <joshuaBPMan>leoprikler: all I see in var log is "Service tor could not be started". <Kimapr>hello everyone, i have a problem in guixsd <Kimapr>some apps, like cargo or minetest can't do https <Kimapr>weird thing is that wget and curl work normally <Kimapr> [60] SSL peer certificate or SSH remote key was not OK (server certificate verification failed. CAfile: none CRLfile: none) *apteryx fftw 3.3.8 on core-updates still running its benchmark test, after using almost 500 CPU minutes... meh. <apteryx>with no new output for at least an hour or more now. Seems like it's hung. <bavier>apteryx: my tor is having trouble starting too <str1ngs>Kimapr do you have nss-certs installed? <apteryx>bavier: I think you meant to reply to joshuaBPMan :-) <janneke>shtwzrd: ah, manjaro is a gnu/linux distro; /me hasn't kept up apparently <bavier>joshuaBPMan: it seems to hang at the bootstrap <g_bor[m]>An mail just fell in stating the ci website is down. <bavier>joshuaBPMan: just run from cli as 'tor', the service I think will timeout <g_bor[m]>I am on mobile currently, so I can't have a look <joshuaBPMan>bavier: hmmm. I think my issue might have to do with "Kernel reports time error." I'm still getting those messages. <Kimapr>installing nss-certs didn't fix the problem <g_bor[m]>Kimapr: did you install it as a system package? <Kimapr> ;; Add a bunch of window managers; we can choose one at <Kimapr>snippet from my system config file <g_bor[m]>Ok. That should be enough. I guess that these packages somehow do not pick up the certificates. <str1ngs>Kimapr: does echo $SSL_CERT_FILE produce an output? <str1ngs>Kimapr: can you give an example of a command that produces this error? <Kimapr>d@DComp ~/guttt$ cargo build --release <Kimapr>trying to build my rust project fails when it tries to fetch dependencies <str1ngs>kk I thought you mentioned that. wanted to make sure <Kimapr><!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN"> <Kimapr><title>301 Moved Permanently</title> <Kimapr><address>Apache/2.4.7 Server at gnu.org Port 443</address> <str1ngs>regards to cargo this is strange. libgcc_s.so.1 => not found <Kimapr>d@DComp ~/guttt$ cargo build --release <Kimapr> [60] SSL peer certificate or SSH remote key was not OK (server certificate verification failed. CAfile: none CRLfile: none) <Kimapr>have no problems with libs not being found <Kimapr>i installed cargo with this command: <apteryx>what is the expected behavior of services upon reconfiguring? Are they supposed to be reloaded? I ran 'guix deploy', and can't see the usual output. <Kimapr>nya, it's my generic ultimate-tic-tac-toe project <str1ngs>Kimapr: I think my lib issue was unrelated. I had a ~/.cargo *apteryx self answers from 'Invoke guix system': "The command starts system services specified in FILE that are not currently running; if a service is currently running this command will arrange for it to be upgraded the next time it is stopped". <apteryx>so, 'herd restart service-of-interest' is required to upgrade a service. <NieDzejkob>Kimapr: it's a known problem, there's a patch that fixes it, it landed on core-updates but not yet on master <Kimapr>is there a way to switch to core-updates or something? <NieDzejkob>you need to create a channels.scm file and put core-updates as the branch. Either in /etc/guix or ~/.config/guix, IIRC. I wouldn't do this, though, core-updates has substitutes much more rarely <NieDzejkob>It would be less computationally intensive to checkout the guix git repo and apply the patch locally <janneke>shtwzrd: i intend to do a writeup some time soon <janneke>shtwzrd: the trick i used was to not partition the microsd or install u-boot myself <str1ngs>NieDzejkob: Kimapr yep seems cargo is not using the right CA file <janneke>shtwzrd: i put stock debian image onto the micro-sd and then did guix system init onto /dev/mmcblk0p2 <scaredysquirrel>how might I enable a serial console with a root login that if you logout it will just relogin? <str1ngs>Kimapr: I would clone guix.git as NieDzejkob mentioned. <scaredysquirrel>with qemu on the host machine how do I enable a reconnectable serial console? <NieDzejkob>Kimapr: clone the git repo, the url should be in info '(guix)Contributing'. Download the mbox file from the link above, and pipe it into `git am` when you're cd 'd into the git repo <NieDzejkob>then put file:///path/to/git/repo as your channel path <NieDzejkob>and then guix pull will get the package definitions from your local checkout. You'll need to use git pull --rebase && guix pull in the git repo checkout to update the package definitions later on <sirgazil>sneek: Later tell civodul I see the same headings problems in guix.gnu.org using other computers. <Kimapr>> you need to create a channels.scm file and put core-updates as the branch <wheeler>Building a package using guix build -f mypackage.scm and trivial-build-system. The build keeps dying on (invoke "tar" "xvf" (assoc-ref %build-inputs "source")) and I have `(#modules ((guix build utils))... in (arguments). is there something else I'm missing to get (invoke "tar" "xvf" ...) to work? <kirisime>Which group should my user belong to on guix system in order to use KVM? <bavier>wheeler: with trivial-build-system, you need to set up PATH so that it can find both "tar" and the appropriate decompression tool <janneke>anyone running X without a display manager; using startx or so as a user? <mbakke>kirisime: check the permissions on "/dev/kvm" <Kimapr>d@DComp ~/guix$ cat /home/d/Downloads/bug_38873_message_5.mbox | git am <Kimapr>Applying: gnu: curl: Make libcurl respect SSL_CERT_{DIR,FILE} <Kimapr>.git/rebase-apply/patch:44: trailing whitespace. <Kimapr>.git/rebase-apply/patch:54: trailing whitespace. <Kimapr>.git/rebase-apply/patch:64: trailing whitespace. <Kimapr>.git/rebase-apply/patch:94: trailing whitespace. <Kimapr>warning: 4 lines add whitespace errors. <Kimapr>is that how it's supposed to work? <kirisime>Uh, can I expect to not break things if I just add my user to the group without going through the whole reconfiguring thing? <Kimapr>> then put file:///path/to/git/repo as your channel path <Kimapr>what exactly are those channels for? <NieDzejkob>you don't need "nonguix", I just forgot to remove it <shtwzrd>janneke: aha, clever. I hadn't realized the existing bootloader could boot guix, I thought the path to linux would be in /gnu/store and the loader wouldn't find it <NieDzejkob>Kimapr: channels are where guix gets the package definitions from, setting this to file:///home/d/guix will use the patch when you do `guix pull` <Kimapr>but what if i don't have a channles.scm file? <mbakke>kirisime: you can modify the user database, but the changes will be reverted on the next reboot or reconfigure <shtwzrd>janneke: what did you do in order to get xorg on aarch64? It seemed like if I pulled in any services that reference %default-xorg-modules, it would try to build xf86-video-intel and fail, because that package can't build on arm. <Kimapr>> git pull --rebase && guix pull <kirisime>mbakke: Thanks again, I'll see if my things work with this. <janneke>shtwzrd: ah, but i am currently using debian's linux Image etc; i copied the "append" line from the guix-generated extlinux.conf to the debian-distributed extlinux.conf <shtwzrd>janneke: ohh! nice :) That sounds like a really good way to handle it for now honestly, with all the custom patches the machine still needs that aren't mainlined at the minute. <janneke>shtwzrd: yes, very ugly but nice as a first step -- better than debian + guix :-) <NieDzejkob>I've asked earlier today: <NieDzejkob> so, I think I'm at the point when I'd like to help review some patches. How would you (as in the commiters) feel about granting me commit access? (for reference, I'm Jakub Kądziołka on the ML and git) </NieDzejkob> <NieDzejkob>so far mbakke (and possibly g_bor, wording unclear :D) offered to vouch for me (thanks a lot!). I'm reposting my message, since timezones are a thing and a different set of people is online right now... ***k_ is now known as kkebreau
<bandali>rekado, can we get bugs.gnu.org/39375 by leoprikler merged? <raghav-gururajan>kirisime I think you can disable 'osinfo' via meson build options. There is an outstanding patch for gnome-boxes in guix-patches maillist. It is not fully completed though. <kirisime>raghav-gururajan: I worked around it by packaging the osinfo-db-tools and importing the provided database. It lets me download ISOs with a single click from boxes. <kirisime>raghav-gururajan: If you're more familiar with this thing, it seems to create faulty VM definitions if you're not in the kvm group by always using host passthrough for cpu. <kirisime>raghav-gururajan: And currently, I'm trying to figure out which part of GUIX' qemu stack has decided to not support smartcards and if it's practical for me to try to add the support in. <bdju>I'm noticing the rakudo package is nearly a year behind now. Not as bad as nix shipping one from 2017, but if anyone has time it would be cool if that got updated. <mbakke>NieDzejkob, g_bor : I emailed guix-maintainers@ wrt commit access. <shtwzrd>is there any way to take advantage of `package-input-rewriting/spec` for a dependency that comes in via a service? <bavier>shtwzrd: many services allow you to override the package that is used, so you can pass a rewritten package to the service <kkebreau>I've been trying to rebase my GNOME 3.34 changes on core-updates, but Rust-related build failures have been an issue... <Kimapr>d@DComp ~/guttt$ cargo build --release <Kimapr>very big thanks you saved my workflow <sneek>civodul, you have 2 messages. <sneek>civodul, sirgazil says: I see the same headings problems in guix.gnu.org using other computers. <civodul>sirgazil: actually could you show me what's wrong because it seems to work as intended for me? <civodul>sirgazil: oh i see, the download page is broken for me as well <civodul>i'm interested in ideas on how to get the same effect back <sirgazil>civodul: What did you wanted to do, though, maybe I can help... <civodul>i think the problem before is that <h1> headings would look smaller than <h2> headings in blog posts <apteryx>when using /etc/guix/machines.scm, is it normal that builds happen *only* in the machines defined therein? I seem to recall that the behavior used to be that 1) build machines would be used 2) the local machine would be used as well. <kirisime>I started writing a package for gnome-boxes. Along the way I've written packages for osinfo-db-tools, libcacard and qemu that should support smartcards. Now it looks like my adventures continue to opendnssec. <janneke>shtwzrd: ah, yes; hitting the xf86-video-intel problem now too <pkill9>where is the prototype recursive npm importer? <pkill9>someone said they made one that works <NieDzejkob>ah, fsck, git send-email creates a separate bug for each patch in the patchstack <pkill9>the email archive search doesn't seem to wrok <apteryx>NieDzejkob: there's an explanation in the info manual about how to handle patches with git send-email <bavier>NieDzejkob: I hope your file-system is ok :) <NieDzejkob>apteryx: Thanks. How do I send the first email to a different address with git send-email? <bandali>git send-email --to="guix-patches@gnu.org" <bandali>then wait until you get a bug number from debbugs <bandali>then pass that to --to in the next emails <bandali>git send-email --to="39333@debbugs.gnu.org" --in-reply-to="<20200128234911.8601-1-mab@gnu.org>" <NieDzejkob>huh, but doesn't git send-email send an entire patchstack at once? <bandali>(also a good idea to supply --in-reply-to as well) <pkill9>you can also specify message-id to reply to <bandali>and send them one by one with git send-email <bavier>what!? no terminalphase package yet? <jbnote>NieDzejkob: yes, emacs is from guix also (i'm mainly on GuixSD, except at work). I think I nailed it: for the emacs-forge package, for instance, the autoloads.el file is put into .guix-profile/share/emacs/site-lisp and the main files kept under .guix-profile/share/emacs/site-lisp/guix.d/forge-0.1.0-2.63cbf81. For the gnuplot mode, the autoloads file is kept under .guix-profile/share/emacs/site-lisp/guix.d/gnuplot-0.7.0, which is, <jbnote>AFAICT, a bug. Now on to understanding how to fix it :) <jbnote>NieDzejkob: the code in emacs-xyz.el for gnuplot uses elisp-dir when calling emacs-generate-autoloads while it should probably use lisp-dir. I'm going to test this and send a patch. It's very hard to understand what Scheme code does in details, but it's rather easy to understand what to change :) <pjz>I just installed guix in a vm to play with. I selected i3-wm as that's what I use. Annnd... I can't do anything because there's no terminal installed. <pjz>I see that i3 is from the 'current-system' profile; if I install, say, xterm in my local profile, will it be able to run it? <pjz>also: there should maybe be a default basic system terminal? <drakonis>pjz: install it with one of the system ttys <vagrantc>so i have one question left before submitting reprotest ... which gnu/packages/*.scm should it land in. <pjz>right, that's what I did. And it seems to have worked. <vagrantc>i almost want to rename gnu/packages/diffoscope.scm to gnu/packages/reproducible-builds.scm and put it there ... as it depends on diffoscope and there were cyclical issues where diffoscope was before <vagrantc>diffoscope used to be in package-management before getting moved to it's own module ... <vagrantc>i also haven't tested it on guix system yet, only on Debian running my .deb packaged version of guix :) <vagrantc>also did some reproducible builds tests: 2,463 store items were analyzed: - 2,016 (81.9%) were identical - 37 (1.5%) differed - 410 (16.6%) were inconclusive building bare-bones.tmpl, diffoscope, guix pull, etc. all without substitutes ... including reprepro, which won't successfully substitute yet <vagrantc>eventually plan to set up a weekly job or so to attempt to rebuild everything *vagrantc wonders how many builds differ due to embedded username <vagrantc>that's one thing not normalized by the builds in guix-daemon, no? <vagrantc>considering diffoscope was pulled in, that's a pretty good swatch of the archive though. :) *vagrantc also wonders on any given day running guix challenge will give different results due to guix gc or not-yet-built things <vagrantc>wonder how crazy it would be to never guix gc for the substitute servers ... <janneke>vagrantc: i just booted guix on my pbp -- but i am really cheating ***capnick is now known as cap
