IRC channel logs
2023-01-30.log
back to list of logs
<seninha>everything working now, it was indeed a silly syntax thing. <nckx>terrified-user: OK, simply ‘guix install grub-efi’ (no pull needed, just that). <nckx>terrified-user: Oh, that's back at C-M-F3. <nckx>We're done with the installer UI now. <nckx>Yeah, it'll download some weird stuff (Ruby!). <nckx>But it won't build anything from source. <lechner>what is written in ruby here, please? <nckx>OK, now log out with ‘exit’, and hit Return to log back in. <nckx>Your history is still there, so you should be able to arrow up until you see the previous long grub-install line. <nckx>Does ls /mnt/boot/grub/x86_64-efi now list many *.mod files? <nckx>Let's calmly ‘umount /mnt/boot/efi’ and ‘umount /mnt’ [no n!]. <nckx>If that succeeded, you can ‘reboot’. <Nathan-web>Does anyone know how I might add a cups filter to my system? I don't think it is included in any packages. <nckx>That's it's umount, not unmount. <nckx>Reboot away. I assume/hope your firmware will go back to trying the internal drive first. <nckx>lechner: I don't know, but I bet it's one of the 7590 packages reported by ‘guix refresh -l ruby@2’, happy searching 😉 (and it can be a profile hook not directly related to GRUB). <nckx>You're very welcome, sorry we had to make that anticlimactic detour through networking. <nckx>Yeah, I still wonder what actually happened. <nckx>If you reconfigure again and this happens again (I don't expect it to! But if it does), let us know, but at least you'll be able to fix it yourself. <seninha>i want to try a GNUstep-based desktop on my Guix system, but it seems there are not much gnustep applications packaged yet, is there any repository (or "channel"? is it the name?) for GNUstep? <relieved-user>seninha: I don't know of any channels for GNUstep apps, but I guess missing apps should be contributed to the standard Guix channel, or are they not free software? <seninha>they are GNU projects, so 100% free as in freedom <seninha>i found a few packages on `gnu/packages/gnustep.scm`, but there a lot more GNUstep applications missing <seninha>i'll learn how to write package config and contribute them <seninha>it's weird, my BSD system has the entire GNUstep packaged in their repos, but the GNU system lacks one of its desktop environments. <seninha>(probably because GNOME has won the battle as the default GNU DE) <relieved-user>Maybe, but also because not all GNU people use Guix and Guix doesn't have enough human resouces to package everything. <old>Question. On my system I have Guix version A, which has guile-3.0-lastest with version 3.0.9 in it. <old> I then enter a pure environment like so: guix shell --pure guix. Now I have a version B of Guix, which has guile-3.0-latest with version 3.0.8 <old>So is it possible to enter a pure environment that has the same Guix version in it? <old>Or should I make it unpure and remove the Guix from the packages list <apteryx>you want both Guix and Guile 3.0.9 in the same profile? <apteryx>seems guix already depends on guile@3.0.9, on master <old>Yes I want a Guix that has the 3.0.9 definition <old>eeh I'm confuse. I think I've got something wrong in my profiles <old>Right so I don't have the same version outside and inside the shell for Guix <old>I guess it make sens? Guix can not have its own definition? <old>Outside shell: guix --version => guix (GNU Guix) 60ef511147a33afc317aeeb0f2c8650c6699fdb1 <old>Inside shel: guix shell --pure guix -- guix --version => guix (GNU Guix) 1.4.0-3.d5fece6 <apteryx>old: that's the old chicken and egg guix problem :-) <apteryx>I guess there hasn't been a new 'make update-guix-package' since guile switched to 3.0.9 <apteryx>actually that version is supposed ot be using guile 3.0.9 <apteryx>it's a bit too late for me to really think <lilyp>old: inside the spawned guix, use a time machine to get back to your current commit <bumble[m]>hey guix I think I want to import and install a custom file-defined guix package inside my home configuration. Is there a way to do this? Any examples or documenttion? <bumble[m]>I'm not experienced w/ guix but think I am looking for something like (import-package-from-file "./mycustompackage"), but not sure <iyzsong>bumble[m]: 3 ways, 1: inline the package definiton in your home configuration file, 2: (define my-package (load "my-package.scm")) in home.scm, and my-package.scm should ends with the package object. 3: use "(use-modules (my package))" in home.scm, then package should be a proper guile module file (with define-module, and can be find in guile's load-path) <ArneBab>Trying to run core-updates: Bail out! GLib-GIO:ERROR:../glib-2.72.3/gio/tests/socket.c:1599:test_reuse_tcp: assertion failed (error == (g-io-error-quark, 33)): error is NULL <ArneBab>170/267 glib:gio / socket FAIL 5.23s killed by signal 6 SIGABRT <ArneBab>/tmp/guix-build-glib-2.72.3.drv-0/build/meson-logs/testlog.txt <iyzsong>how to run mumi? ./pre-inst-env mumi fetch report error: No such file on directory /.../mumi/data/spool/index.db.realtime <sloped>iyzsong: how do you try to run mumi? <sloped>iyzsong, i guess the easiest way is to use the mumi service? <iyzsong>sloped: i cloned its source, and do its bootstrap.sh; configure, make, then tried to run it uninstalled <civodul>iyzsong: hi! there's a mumi service in maintenance.git <sloped>would be nice to have something like a guix-shell for services :D -- like "guix system shell mumi-service-type" or so :D <civodul>"guix system vm" is not far from that <iyzsong>i want to improve my workflow with some cli tools for mumi/qa, since emacs/debbugs is kind of slow, it list all patches.. <civodul>but yeah, it'd be nice if you could add services from the CLI <iyzsong>mumi service fails to up for me, well have to go for now.. <Kabouik>I'm losing 2GB of disk space on my btrfs system every time I do a new `guix install` or `guix remove` transactions (working with just emacs packages right now, so they are relatively light packages). Is there something I can do to limit this? I assume my guix history is huge and every new item is taking a lot of space? <Kabouik>I don't have much free space left so I'm feeling I shouldn't postpone too much before finding a solution to this. :< <Kabouik>I just went from 153GB free to 147 after just installing an emacs package,then another, then removing them. This leaves me with no extra package compared to before, but 6GB burnt. :> <iyzsong[m]>Kabouik: you can run "guix gc" to free space, after delete old profile generations <Kabouik>Trying that, thanks. But was I assuming right the cause of such big storage losses when using guix package/remove? <iyzsong[m]>usually not, does it download many packages like glibc, gtk? <Kabouik>No, I was just playing with emacs packages I wanted to try, which didn't have such dependencies <Kabouik>I removed 2m of old generations (guix package --delete-generations=2m && sudo guix system delete-generations 2m && guix gc), now I think it will take ages to delete /gnu/store/trash. I hope I didn't go too far. :< <Kabouik>Just so I understand it right, delete-generations actually prunes, right? So in the case above, it removes 2 months of generations starting from the oldest, correct? <iyzsong[m]>not sure, i'm on phone, you can look up the help or manaul <civodul>i usually do what you did, except i run something like "guix gc -F20G" to free up to 20G (rather than as much as possible) <civodul>also, you can combine those in one command: "guix gc -F20G -d2m" <fiesh>what's the advantage of that compared to freeing up as much as possible? <Kabouik>Oh, nice. I didn't those could be combined in a single command. So, leaving alone -F (if I want to free as much as possible), I could use something like `guix gc -d2m`? I guess that doesn't cover guix system generations too? <Kabouik>fiesh one advantage I see is it makes it easier to keep a relatively stable storage dedicated to generations if you run the same command with the same -FamountG regularly, as it will not depend on which packages have been installed between two gc <civodul>Kabouik: "guix gc -d2m" doesn't cover system generations, right <sloped>fiesh: maybe if you don't want the command to run very long because you don't have time right now <Kabouik>That too (I expect the deletation of my /gnu/store/trash for 190235 MB to take a few hours) <fiesh>sloped: ah hmm I guess... seems somewhat like a niche use case to me though... for me it's always super fast, but that maybe because it's a new nvme device <Kabouik>I'm on a relatively recent nvme SSD too and it's super long, but that may be related to btrfs too, I don't know. <sloped>fiesh: hm, or maybe you run it quite frequently? <fiesh>I suppos that too, yes, valid point <efraim>oh man, master->core-updates is quite hands on. nearly 7000 commits <fiesh>I just `guix gc`d 11,956.8164 MiBs in 35.349s on btrfs <fiesh>but it's only a laptop and thus consumer hardware -- it sounds fishy a 200gb gc should take _hours_? <efraim>is there a way to add a 'fixup' commit into a 'merge' commit? <efraim>ok, i found 'git rebase -i --rebase-merges' and hopefully that doesn't ruin all my work <efraim>it seems to have worked. now to test it a bit and see about merging some package updates <efraim>also: 740 files changed, 606278 insertions(+), 301527 deletions(-) <futurile>Q: Is there a way to 'skip' an optional dependency when packaging a rust crate? <Kabouik>fiesh I'm still gc-ing one hour later and I have no idea how far I am in the process. :< But admittedly, the amount of files to delete is much bigger in my case (~190GB) <Kabouik>It's also a laptop in my case, but still has a nvme SSD and i7 cpu. It's probably throttling though because the thing is stabilizing at 85-90°C. <tschilptschilp23>Mhm, depot_tools README tells me, that it needs python3.8 -- I'm on guix' 3.9, any ideas how I would get around this? <FlaminWalrus>What's up with LLVM not having substitutes on i686? My rig doesn't have the RAM to build it...(sorry if this is a tired/dumb question) <tschilptschilp23>OK, I guess I will have to start up docker. It does not seem to be python3.8 vs python3.9, also if add python2 to my guix shell, and prepend GCLIENT_PY3=0 to the fetch command, it will not find pyton2.7, which we clearly have. <jlicht>efraim: not sure it's relevant right now, but perhaps `git rerere' helps you when doing big merges/rebase chains. It stores your "resolved conflicts" locally, which makes it so you often only have to fix one issue once :) <Kabouik>I think I'll never understand how guix gc works with btrfs. guix gc told me it deleted 190'000MB, then hard linked saves 29'000MB, then said t freed 134'000MB. At this point already I'm confused by how these numbers don't atch. And when I check usage with btrfs, I see that only 48GB have been freed. <jonsger>ACTION added a session about Guix in air gapped environments to the Guix days programme... <florhizome[m]><Kabouik> "fiesh I'm still gc-ing one..." <- It would be great to have some configuration Options for guix to occupy less Disk space 🤔 <Kabouik> In the end it took somewhere between 1 and 2 hours to gc 2-month worth of generations in my case (190GB), I was away for lunch when it ended so I'm not sure. But what bothers me is how the numbers at the end don't add up. <florhizome[m]>maybe deleting substitutes from local builds automatically when they are not installed and updated substitutes after profile upgrades. <nckx>Kabouik: Is it not plausible that 56GiB were deduplicated? <nckx>Kabouik: For the btrfs mismatch, uhm, *waves hands* compression? <Kabouik>My knowledge of how Guix works under the hood is too limited to tell, but if you think it's plausible, I'm convinced it is. <florhizome[m]>I think a distro shouldn't rely on users developping their own solutions to cope with it's memory usage if it knows that's a Problem. <nckx>Oh, it was just a question. I read those numbers as: I deleted files that, if you add up all their sizes, add up to 190G; this freed 134 of actual space (because 56 of that 190G was shared internally); in what remains, 29G are still saved by deduplication. <nckx>It's possible that those numbers are wrong, but at the same time they don't sound implausible, so I don't think we can say. Now, the btrfs thing… not sure. That's impressive compression if that's what it is. <apteryx>Kabouik: do you use btrfs compression? <florhizome[m]>Well actually i am Here because i have a Problem with gnome screensaver or however it is called in gnome wayland <gabber`>florhizome[m]: is there anything of interest in any of the relevant logs? <florhizome[m]>And i don't even know where to look for logs or which package it is in <gabber`>can you switch console (Ctrl+Alt+F1) and log in there? <florhizome[m]>I meant this feature earlier. but i guess it only makes sense with Automatic system updates <florhizome[m]>gabber` i looked it /var/log/messages and i can't find anything there <gabber`>huh. no mention of pam or screen locker or anything? <gabber`>you could try (caution!) to kill the relevant process(es) -- find suspicious names with `top` or `ps aux`, kill them, go back to the graphical-login-console (probably Ctrl+Alt+F7) and check whether you're still locked out <gabber`>i have no experience with Gnome et al. so you might want to check for gnome specific logs? maybe ~/.cache/gdm/session.log ? <florhizome[m]>so i just reconfigured to remove a custom service that has been spamming the crap out of me <apteryx>jonsger: hm, some files apear to be missing though :-/. where's bin/hello for example. <apteryx>ah! that's because find-files #:directory? returns only directories? <apteryx>I thought it'd return files + directories <gabber`>florhizome[m]: LPT you can use `less` to filter for specific lines -- but you can also reverse that filter to exclude specific lines (& and ! are the key buttons to press) <apteryx>the issue was that symlink from a profiles were not listed by find-files until I used #:stat stat (the default is lstat, which doesn't follow symlinks) <elais[m]>Where might I post to ask for features turned on in the aarch64 build of Linux libre? Right now netfilter isn't enabled that kernel which breaks the wire guard and nftables services. <elais[m]>Is this a discussion for guix develop or a bug report? <Kabouik>nckx: Actually the deleted files add up to 190GB, gc says it freed 134GB of actual space (because 56 of those 190G were shared internally; thanks for the clarification, I didn't know that. I think guix gc should take that into account when giving numbers); in what remains, 29GB are still saved by deduplication (presumably)… But again, `btrfs fi usage /` only shows about 45GB freed. Which is a disappointment when I was expecting 190GB. :< <Kabouik>apteryx: yes, I use zstd compression. <apteryx>Kabouik: then that's the reason; the guix daemon reports the uncompressed size of the garbaged files, I think <apteryx>that's how I regularly manage to gc twice the amount of the physical size of my partition <nckx>That compression ratio sounds what I'd expect from zstd. In order to give btrfs ‘compression’ stats, Guix would have to explictly probe for that information rather than rely on the standard interfaces, which I find overkill for a brief informational message. <Kabouik>To be honest I still don't understand how Guix uses so much space on my 1TB device. This has been a running topic since I'm on Guix, we discussed it here several times, and I'm just still foncused by btrfs too (which admittedly seems to significantly complicate calculating free space, even if I'm not using multiple drives in RAID). <apteryx>yep, it can be confusing. just trust the non-allocated space as the true available space for btrfs, and run a 'btrfs balance' regularly to regain non-allocated space following file deletions (such as with 'guix gc') <elais[m]><apteryx> "elais: sounds like an actual bug" <- It's tricky to say bug because technically nothing is broken but I suppose it is one considering a couple of services are effected <elais[m]>Like configuring a kernel isnt an exact science <Kabouik>Thunar's right click/Properties reports that my /home is 255GB but uses 200GB of my 950GB drive. I don't get that mismatch. Now df -h /home says my home uses 389GB (what?). And btrfs fi usage /home says that /home uses 387GB (close enough). I assume Thunar is just wrong. But still, since I seem to have 200GB unallocated on /, that means my system itself is using 550GB, and I don't know how. <florhizome[m]>Seems like it could be a Problem with the blur-my-shell plugin <nckx>elais[m]: That's true, but in this case it's an actual bug 😉 The aarch64 kernel configuration in Guix *is* weird and contradictory in many places. Many choices aren't deliberate and should be improved. <nckx>I.e., don't assume that every option was deliberately (un)set by a human for a reason. <florhizome[m]>yikes, seems like gnome session freezes every time i disable an extension <elais[m]>That's fair. But at the same time I'm happy with how much progress has been made. I remember a year and a half ago angrily trying to boot guix on my rockpro64. Now it's as simple as preparing a new SD card. Granted shutdown doesn't work but you can't win them all <gabber`>elais[m]: does `halt` not work? i think that worked with mine <imstuck>hi, i'm having issues installing openssl-1.1.1l, it fails on the 'check' phase, specifically, `test/recipes/30-test_afalg.t`. <imstuck>It's my first time playing around with guix, I'm using the package manager on gentoo. I just installed it, and then `guix pull`ed. I understand I can probably write a patch that deletes this file - but how? <elais[m]>I've not done halt I've only tried issuing the shutdown command <elais[m]>Also rebooting on mmc doesn't work, I have to physically reset it each time. <apteryx>imstuck: hi! so your 'guix pull' completed successfully? what does 'guix describe' report? <imstuck>apteryx nope, it failed on openssl-1.1.1l installation <apteryx>OK; so what are you currently on? ('guix describe' output) <imstuck>`guix describe` returns: Perhaps this `guix' command was not obtained with `guix pull'? Its version string is . <imstuck>the gentoo package version (of guix) is 1.4.0, if that helps <apteryx>seems another user reported such openssl build failure: #60821 <apteryx>I'm guessing you can workaround it by authorizing substitutes from the build farm <imstuck>do you reckon that if I bootstrap my `guix pull` with a substitute, i can rebuild then the system and successfully compile openssl locally? <apteryx>see: info '(guix) Official Substitute Servers' <apteryx>I'm guessing there may still be time bombs in openssl test suite, where certs expire and fail the test <apteryx>so by using an old build substitutes, you avoid it <apteryx>I don't have time to investigate but 'guix build openssl --no-grafts --check' should fail for anyone too, if that problem is reproducible <f3n1x>ahós guixers! I'm aiming at enjoying 'pdfgrep'ping in Emacs vterm in GNU Guix, but i cannot, for now. Uh... it looks like i'm missing some package in my current generation ? Searching for 'pfgrep' guix says that there is 'pdfgrep' and there is 'emacs-pdfgrep' ..ah?! MAy i install 'pdfgrep' ? May i install 'emacs-pdfgrep' ? ... Or both ? Thanks, thanks, thanks <seninha>hi, how can i modifye a group from the %base-groups in /etc/config.scm? <gabber`>f3n1x: my guess (from shortly looking up `guix edit emacs-pdfgrep`) is that emacs-pdfgrep is the only package you need installed - since that package has `pdfgrep` as input. <seninha>i need the group "users" to have gid 1000 <seninha>(for me to have access to my guix user's home directory when mounted via sshfs) <gabber`>i think Manual section 12.6 has some relevant information for you <seninha>i'm reading 12.6, but it only says that %base-groups is a list of user groups, not how i can modify it <seninha>(maybe i should throw some scheme scripting at it?) <gabber`>seninha: yes, you should :) AFAICT there's no super-handy modify-groups function available, but i guess you can filter the %base-groups list and add a group according to your liking to it? <seninha>btw, the case is: in unix, each user has a main group. In most GNU/Linux, the standard is for all users (say "alice", or "bob") to use the generic "users" group as main group ("alice:users", "bob:users"). On BSD, each user has a proper main group with the same uid ("alice:alice", "bob:bob") <seninha>i'm mounting the home directory of my guix system into my bsd system, and, for my BSD "seninha" user (uid=1000) to access the Guix "seninha" user, both the uid and gid of the /home/seninha on Guix has to be 1000 <seninha>i configured "seninha" on Guix to use uid 1000 (just like in my BSD), so i can mount his home directory at my bsd with uid=1000, but i need also the gid to be 1000; otherwise there'll be a bunch of new files with unmatching groups in it <imstuck>apteryx if you're still here, tried both official (bordeaux and ci) substitute servers (guix pull --substitute-url=...), but it doesn't seem like those servers have openssl available, as I'm still compiling that package and still failing the test <seninha>is there some way i can automate the creation of a group to match a user? (emulating BSD/Arch/Slackware/etc behavior) <gabber`>seninha: you might want to go the (far easier) way of creating a specific `seninha` group with gid 1000 <cbaines>imstuck, can you share the derivation/output for this missing openssl? <seninha>but since a few GNU/Linux distros (mainly non-Debian based ones, like Arch and Slackware) use the BSD style of automatically creating a group for each user, it should be nice to also use it on Guix. <imstuck>cbaines sorry, i'm completely new to guix, derivation refers to the hash? bfirgq65ndhf63nn4q6vlkbha9zd931q-openssl-1.1.1l.drv <cbaines>imstuck, yep, that's the .drv (derivation) thing <cbaines>imstuck, if you do guix build /gnu/store/plr00nij45964cyy7sfcg5rcsi8hks0h-openssl-1.1.1l <imstuck>it fails on test/recipes/30-test_afalg.t (after compile phase). <imstuck>background: running amd64, guix on Gentoo Linux, freshly installed 1.4.0, just trying to `guix pull` <cbaines>running guix build on a specific output shouldn't build anything, it'll just attempt to substitute it <cbaines>so I'm not sure where this failure is coming from? <imstuck>sorry, I'm rushing into things then - I previously ran guix build openssl@1.1.1l and I assumed it would be the same, I'll be more attentive <imstuck>;;; note: source file /usr/share/guile/site/3.0/gnutls.scm <imstuck>;;; newer than compiled /usr/lib64/guile/3.0/site-ccache/gnutls.go <imstuck>;;; note: source file /usr/share/guile/site/3.0/gnutls.scm <imstuck>;;; newer than compiled /usr/lib64/guile/3.0/site-ccache/gnutls.go <imstuck>substitute: ;;; note: source file /usr/share/guile/site/3.0/gnutls.scm <imstuck>substitute: ;;; newer than compiled /usr/lib64/guile/3.0/site-ccache/gnutls.go <imstuck>substitute: ;;; note: source file /usr/share/guile/site/3.0/gnutls.scm <lechner>imstuck / please use a paste service such paste.debian.net <cbaines>imstuck, I think the last line is relevant here <cbaines>substitute: guix substitute: warning: ACL for archive imports seems to be uninitialized, substitutes may be unavailable <imstuck>cbaines do you happen to have some documentation on that? couldn't find anything online other than dead threads mentioning the warning with no resolution <imstuck>cbaines you're a lifesaver, thanks so much :) <cbaines>imstuck, are you getting substitutes now? <imstuck>yes I am, openssl installed successfully <cbaines>if for some reason you don't want substitutes, I think there's an option to pass to the guix-daemon to disable that <apteryx>imstuck glad the substitutes allowed you to bypass the openssl bug <f3n1x>thanks, thanks, thanks gabber` <civodul>Htop170: hi! i don't know; someone suggested in that thread that some dependencies might be missing, which could mean extra work <lechner>Hi, the latest comment to #60899, which also appeared on guix-devel, may reflect a growing impression that patch acceptance has slowed. (i don't know if that's actually true.) would it be helpful for someone with authority to explain cbaines's efforts to implement automatic patch approval, if that's what everyone is waiting for? <cbaines>lechner, what specifically would be helpful to explain about trying to automate aspects of patch review? <lechner>cbaines / it may help to eliminate useless "ping" messages (which I think should be disallowed) by restoring confidence that relief is coming <Htop170>civodul: Hi! Thanks for your answer! <lechner>cbaines / maybe useless was too strong a word. let's change that to "desparate" <civodul>i'm in the mood for some riscv substitutions <unmatched-paren>ACTION :e ~/code/guix-artwork/website/posts/dissecting-guix-2-store-monad.md <unmatched-paren>the only thing harder than reading about monads is writing about them... <cbaines>lechner, I don't really see a relation between bumping/pinging specific patches and automated aspects of patch review <cbaines>lechner, I don't really mind people sending reminders when there's been no feedback in some reasonable time <cbaines>lechner, right, I don't think maintainers/committers need to "let" more volunteers participate in the patch review process, so I'd disagree with the premise of that message <lechner>cbaines / since you don't mind an occasional reminder, i also have not heard from you for my own #58658 <cbaines>I think there's more that can be done to facilitate people without commit access to review patches, but I think it's wrong to say that the project maintainers aren't letting voluenteers participate <lechner>cbaines / i showed that message more to support my claim that there is growing sense that patch review has slowed <Guest46>Is it possible to delete the swap partition in the Guix System installer? I just don't understand how <gnucode>I am having a weird time trying to run a bash command and for some reason the fish shell is giving me errors. Is there a way to open a terminal and say, only load bash. Or unload this fish stuff? <gnucode>hmmm, 'guix shell -C coreutils rsync openssh-sans-x' almost worked. <gnucode>ssh: connect to host SOME.IP.ADDRESS port 22: Network is unreachable <gnucode>but ssh user@SOME.ID.ADDRESS works outside in a normal termial. <Guest46>oh yea because of container (-C flag) <lechner>gnucode / that may not be related to your choice of shell <gnucode>Guest46: so I need to enable networking. ok. <gnucode>that still didn't work. I think it has to do with my pinephone's default shell being fish. <gnucode>I gotta get ready for work now unfortunately... <elais[m]>Looks like somebody already added the aarch64 kernel config enabling all the wireguard and firewall goodies <gnucode>lechner: and Guest46 thanks. I will check it out! <elais[m]>Maybe now I can finally get wireguard working <gnucode>the #fish people just asked me if I have rsync installed on my pinephone...I didn't realize I needed it installed on both devices. hahaha! <Guest46>ah make sense, i wondered why fish responds since you used bash <lechner>Hi, is there a way to run the guile-build-system manually, especially inside guix shell --development? <apteryx>but a Makefile with a pattern rule should get you close <apteryx>something like %.go: %.scm\n\tguild compile $@ <lechner>apteryx / thanks! i am just trying to run a Guile program that ships custom modules before pushing the changes to my repo. I guess it's GUILE_LOAD_PATH for me <Htop170>I am in the process of writing a guix package for languagetool, after trying to build it for the first time, I got an exception telling me that `Throw to key `no-such-input' with args `("org.owasp" "dependency-check-maven")'.` After giving a quick glance at java.scm, I saw that some package using maven-build-system exclude some args looking like <Htop170>the one I got with my exception. Do you think it is the thing to do here? When do you know when you have to exclude something ? <ArneBab>The branch core-updates does not build for me. Bail out! GLib-GIO:ERROR:../glib-2.72.3/gio/tests/socket.c:1599:test_reuse_tcp: assertion failed (error == (g-io-error-quark, 33)): error is NULL <old>is there mailctl avaiable in a package? <apteryx>lechner: what is not working? what kind of output are you getting? <lechner>apteryx / Guile recompiles the installed modules <apteryx>the GUILE_LOAD_COMPILED_PATH must not be set correctly? <lechner>apteryx / maybe that's it. i have not worked on that for a little while. it's actually quite hard to package Guile modules, which PAM modules would be, in Guix. I think it works only via the system profile <civodul>so, time to send an announcement to meet at "Au Bon Vieux Temps" 