<roptat>so what does neofetch tell you, and what is the output of "guix package -I | wc -l"? *roptat waiting for the TP robot to reject my changes ^^' <roptat>225 as guix-user or guix-system? <Dzhigit>my version of neofetch does not specify <roptat>so what does guix package -p "/run/current-system/profile" -I | wc -l tell you? <Dzhigit>1. I don't know if I should have specified, but while I do have the Guix distribution, I am currently running this on a dpkg-based system where I installed the Guix package manager 2. I have installed neofetch with Guix, not dpkg <roptat>so if it doesn't specify, maybe it just added package installed with dpkg and installed with guix? <roptat>or it does something, but I need to know your version of neofetch then <Bumblehorse>How do get an .xinitrc equivalent when using GDM? .xinitrc and .xprofile both don't work. I resorted to having i3 execute the commands, but now I need to set the userresources variable (like in .xinitrc) for urxvt to read .Xresources. <Dzhigit>guix package --list-installed shows neofetch <Dzhigit>@raingloom LibreJS is blocking 131 scripts from that page. that is the most I have ever seen. <raingloom>Dzhigit: i opened it in Netsurf, so i'm not worried about JS. UwU <roptat>Dzhigit, what does "neofetch --version" tell you? <roptat>raingloom, "all right reserved" I'm afraid, though they probably haven't thought about licensing is all <roptat>are you in the US? You could probably call them and ask politely if there is any license? <dannym>How do I package a guile script in automake so it can actually be run like a normal program (i.e. so a script shows up in $out/bin which I can run and it will work)? <roptat>raingloom, ok, I can try and call them tomorrow morning if I remember to do so :) <Dzhigit>well I feel rather now, even though I installed neofetch through Guix, and which neofetch returned Guix version of neofetch, neofetch was using the dpkg version <Dzhigit>using the guix version now shows 6 guix user packages <Dzhigit>but that means it is just using guix package --list-installed <Dzhigit>it's not showing any of the dependencies like perl <maav>roptat, raingloom: i use autoconf to provide the load-path and the guile path to the topscript and bin_SCRIPTS in am for the resulting file. <maav>sorry, dannym, not raingloom :( <roptat>arg, I can never get it right :/ <roptat>the TP robot is being mean to me again <maav>dannym: well, configure is a shell script, but hopefully you don't have to write if you use autoconf :) <maav>roptat: sometimes it gets several tries, hope you get it soon :) <maav>you have to add the main.scm.in to AC_CONFIG_FILES, and use there @somedir@ to path you're installing the rest of the modules, and you can use AC_SUBSTITUTE to replace that or any path you need there <maav>or you can turn to sed and do that in (auto)make... but i think the configured values should be already provided by configure, as a matter of taste, and leave make only to generate things, not substitute them <maav>AC_SUBST, sorry, too verbose for autoconf ;) <maav>roptat, i've just seen your example, just like that but also you can provide any load path there too if you depend on an external library <maav>dannym: you're welcome, happy to help :) <maav>roptat: e.g. i'd substitute there too gitolite and gittile /var/lib paths, as a normal user usually couldn't write there <roptat>well it's a daemon, it's not meant to be run by a normal user <roptat>although you can still configure that path at runtime <roptat>but you're right, it should be @localstatedir@ or something like that <joshuaBP`>setting up dkimsigning with opensmtpd...hopefully it works. <maav>roptat: yup, some users run daemons for tests too, or have stringent policies on their machines about paths usable by certain applications... <dannym>Also, I get a lot of "In procedure load-thunk-from-memory: incompatible bytecode" for guix modules <maav>dannym: i had them while doing some tests but not later when i pull'ed my branch... so i couldn't check anything more, i hope i can borrow a computer this weekend too and run make check-system again... <roptat>dannym, I think that's because guix itself is built with guile 3.0.2, but your app and dependencies are most likely built with guile 3.0.4 <maav>roptat, thanks! that could be the key for the issues i had with check-system too, at least now i have something to look at :) <maav>i'm leaving now, have a good night/day :-) ***catonano_ is now known as catonano
<ryanprior>Much better documentation than we've had before. If you want to realistically ship javascript apps without the npm madness, esbuild is a viable route. <vits-test>Order 3 esbuild in the following hour, and U'll get 3 git clones for a price of one! Only today! <seeg123>hello, silly question but does guix provide firefox? <xelxebar>seeg123: Do you have guix installed? guix search <regex> lets you search for packages :) <seeg123>yes, but i think icecat is the only option <vits-test>seeg123: ues keyboard driven browsers. There are Nomad, it is Guile (customizable, Emacs-like) + WebKit. Also qutebrowser, Python + QTWebengine (not alpha :). <lfam>seeg123: Guix doesn't provide a firefox, but rather icecat as a variant <vits-test>seeg123: Don't forget to do `guix build --dry-run BROWSER`, to check if binary substitutes available. Else Ur machine may attempt build it from sources. <vits-test>`guix weather X` not always work, as U're asking for, say, "nomad" -- which is Guile (not compiled lang) front-end for WebKit. And "qutebrowser" is Python f. end for QTWebengine. `build --dry-run` sais most of the times if things would be downloaded, or built. <seeg123>i'm testing guix sd, is there a way to specify "binary first" globally? <vits-test>seeg123: U may need to install fonts, and run fc-cache -rv. <vits-test>2.6.3 of manual recommend the essential font packages. <seeg123>btw, do you have any recommendations for HiDPI? the only one I know is xrandr --scale but it doesn't work too good for all apps <vits-test>str1ngs: ^^ Hello; any recommendations for HiDPI? *vits-test is hobo with ~ 1440 x 900 <seeg123>or maybe i'll just reconfigure system to use a lower resolution <seeg123>ah, another thing: is haskell stack packaged for guix? <divoplade>Hello, where do I find a program named mkfs.vfat? <sneek>zimoun?, pretty sure was seen in #guix one day and 11 hours ago, saying: ‘guix build’ already does some work. Is it not enough?. <jonsger>I'm in a chrooted Guix system and want to change from the root user to foo user, but it says: "su: cannot make/remove an entry for the specified session" what am I missing? <jonsger>vits-test: thx, but still the same error message *zzappie had same reaction <jonsger>leoprikler: the command does work, the daemon as well with `--dissable-chroot` but I couldn't get shepherd into the party <str1ngs>leoprikler: it does work from namespace <nckx>As well (=as poorly) as most. Systems assume services are available (here: PAM). <nckx>Not really a Guix thing. <leoprikler>try having a look at /var/guix/profiles/system to see what guix itself sets up <jonsger>leoprikler: is there something like `jq` to get a readable output out of this online guile thing? <jonsger>leoprikler: it's guile code written in one line <leoprikler>guile -c "(let loop ((l (read (current-input-port)))) (unless (eof-object? l) ((@ (ice-9 pretty-print) pretty-print) l)))" <leoprikler>try it using echo "(let loop ((l (read (current-input-port)))) (unless (eof-object? l) ((@ (ice-9 pretty-print) pretty-print) l)))" | guile -c "(let loop ((l (read (current-input-port)))) (unless (eof-object? l) ((@ (ice-9 pretty-print) pretty-print) l)))" <jonsger>oke, it's wasted time. I don't think I can repair my system in chroot -> new installation ftw... <mfg>jonsger: isn't that one of the advantages of Guix as a system? That it's (more or less) trivially possible to just reinstall/reconfigure, given the data that was on it is backed up? <mfg>Okay, i'm also assuming a correct operating-system definition... <vits-test>jonsger: Why U cannot choose one of the previous generations? What happen? <jonsger>mfg, vits-test: something with EFI/GRUB got completetly broken so It doesn't go until grub menu <mfg>Bah, that sounds bad... *vits-test imagines.. GNU/GuixBSD, a fork of NetBSD w/o GRUB. <maav>jonsger: i connected too late so i had to check the log, do you get at least the rescue console from grub? <maav>might it be that the links inside grub.cfg break somehow with the chroot? <maav>or the image generated, as it contains something like "set root=$prefix" that sets the path for grub.cfg <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, nckx says: I'll look into sergei. Dmitri has a busted fan. I'll open him up & invoice Guix if needed. 😉 <civodul>i was surprised to see libsoup depends on php, but apparently it's for its tests <maav>question: in that case it should be a native-input? would that remove the reference from the out? <civodul>and there's no reference in the output, per "guix size libsoup | grep php" <civodul>it's just that until substitutes are available, php is on the way :-) <maav>yeah, i should have asked that to guix size instead here 0:-) <brendyyn>being a native input should make no difference. i think if you are not cross compiling then its indistinguishable from regular inputs? <maav>regarding ci, are these jobs from October 20th stalled? or is it normal? <sneek>g_bor[m]?, pretty sure was seen in #guix 3 hours and 20 minutes ago, saying: sneek: seen zimoun?. <maav>zimoun: thanks for the link, i have to learn a lot then :-) <zimoun>it is a really interesting subproject (and experimental yet) mainly targetting bio-informatics but could be used for other stuff. On the Nix land, the experiment is FunFlow https://github.com/tweag/funflow <nckx>Sigh. Guix. Why. Sometimes. <nckx>civodul: My Guix System SSH-to-ttyACM gateway next to the Russians is also broken. Boots to a GRUB rescue shell after a simple reconfigure. <nckx>If there's another local lockdown before next week this can take a mightly long while. <nckx>How important are these two for 1.2? <maav>nckx: what is the problem? maybe there is already a patch for it :) <maav>if you can access to the rescue shell, there is hope at least... <nckx>maav: It's a meta-problem. For example, you reconfigure with a new printer driver and GRUB just breaks because Friday. As much as I like Guix & The System feat. the Freedoms, I can only rely on it 99%. <nckx>Mostly minor issues but debugging & fixing this one will involve me physically kneeling in front of a monitor for at least an hour and it's all so disproportional. <maav>nckx: you cannot rely 100% on anything but that in real life we are all mortal beings <maav>i'm sorry that you have a problem, really *nckx in a mood, but because I feel responsible for these machines, not angry 🙂 <maav>i hope is less than one hour in front of the monitor :( <nckx>maav: Sure. I don't want Guix to be reliable because it's boring, trust me 🙂 <nckx>I love it but love is a rollercoaster. <maav>usually, when you have stage1 loaded, it's just a matter of loading manually the system, but when stage1 doesn't boot you usually, and sadly, just need to boot with something else <nckx>I'm miles away from the thing. All I know is it's apparently hanging at something like ‘no driver for file system’, with the pre-normal-mode GRUB prompt under that. I'm pretty proficient at GRUB but I can't debug that over the phone. 😛 I doubt it's fixable. The efficient solution is to boot a live medium & reinstall GRUB with the right drivers, but then I'll never know what went wrong... <nckx>Also, hullo!, since I don't think I've seen you around (much). <maav>yup, i'm quite noob in the irc world too :) <maav>i hope it's easy and fixable at least <nckx>Thanks. I'm sure it'll be easy, just tedious. But if this happens to a ‘new user’ (or just a ‘regular user’, whatever that means), it's probably game over to win them over. <maav>sure, it's pretty daunting finding the grub console when you are not used to that <nckx>Heh, it's still pretty daunting if you are 😉 ‘Hello, screw your morning plans, play with me now.’ <nckx>You hit /gnu/store/<TAB> out of muscle habit? You lose, collect no ^C, go immediately to hard reset. <maav>well, i translate grub too, so for me is just "check me! check me!" <jonsger>maav: nope, no sign from an existing OS on the hard disk. So the BIOS doesn't even see that there is a system on this particular disk, yet the disk looks good (when mounted) <nckx>maav: Is our GRUB localised yet? Was that you posting about it on the ML? <maav>jonsger: do you have gpt or msdos table? <maav>nckx: grub... it's a hard story, it should have been localised, if there wasn't a bug in grub-install and /boot/grub/locale was populated, but that was not the case since grub 2.02 <nckx>jonsger: Above you said EFI, I assume this is x86_64 so that would be UEFI, which implies GPT. But then ‘BIOS doesn't even see that there is a system on this particular disk’ makes me unsure about the actual situation. <maav>so i have some patches, together with fixes for btrfs systems with the store in a subvolume (not /gnu/store) waiting for some review as #44196 :) <jonsger>nckx: sorry about the word BIOS, I mean the software running on the mainboard. Don't know whats the correct word for that <nckx>Welp, the broken server is my only btrfs (although it doesn't use subvolumes) I could use to test them 😛 <nckx>I guess I could test in a VM. I just find it tedious for some reason. <nckx>jonsger: ‘(Boot) firmware’, but many people use BIOS as an abstract synonym. It's hard not to be confused by these things. <maav>jonsger: then you should have an fat partition marked for efi usage, does it have anything inside? <jonsger>yeah, so the boot firmware doesn't recognize my hard disk with Guix system on top as bootable system anymore. It worked for three or four month, then I did two days ago a boot firmware update and afterwards it broken. Installing the original boot firmware didn't helped. So I have no clue what broke... <maav>(the one that should be mounted in /boot/efi) <nckx>Assuming your EFI system partition exists, has GRUB on it, and isn't damaged: if I'd have to guess (and it seems we do, since most firmware is undebuggable crap) it sounds like you need to register Guix's GRUB in the UEFI firmware's nvram using efibootmgr. I've had to do that on multiple machines that ‘just broke’. Unfortunately I don't remember exactly how... <jonsger>nckx: can I do that via live system which is not a Guix Live system? <jonsger>it seems so, so I will give it a shot :) <maav>yup, the binary should be in that partition, that's the only information needed by efi... perhaps the firmware installation activated something "secure boot"-related? i'd check that too <jonsger>maav: didn't find any "secure boot" related stuff, and my ISO USB sticks booting just fine <maav>jonsger: great, just to be on the safe side, propietary firmware companies shouldn't be trusted in any case <maav>(the best would be not use their software at all) <jonsger>maav: I tried that for a year (looks at POWER9) but came back to the bad side of mainboards. So I have to live with that... <maav>i haven't had the courage yet to libreboot mine---it isn't implemented yet, and probably would need hardware work... but who knows what could my bored self make some day :) <maav>i even don't know really, it's a laptop a friend gave me... <maav>and then i moved it to check the model and removed the cable... hehehe <maav>even worse, lspci says something different, hehehe <maav>dmidecode says the same as the label, but it is none of the currently supported <maav>i'll enter this rabbit hole later :) <nckx>*disappointed carnivorous rabbit noises from hole* <maav>don't be, alice and i will come sooner or later :) <maav>nckx: you said that your current server was on btrfs but didn't use subvolumes, isn't it? <nckx>Mainly btrfs on top of mdraid 5 or 6 (I forget). <maav>never tried raid 5 nor 6 <nckx>I've never used 1. 5/6 is enough to guarantee reasonable uptime for a (beefy) home server. Only drawback is the speed, but it's not a problem in practice. <nckx>Never use btrfs's ‘raid’ 5/6 if you do. <nckx>It eats data and small household animals (probably). <mfg>is btrfs that bad or is it just it's raid5/6 mode? <maav>i meant with mdraid, i tested for the first time the btrfs subvolumes and came with those patches <nckx>mfg: Btrfs's original implementation, it was insanely bad. <mfg>so at least it got better over time? <mfg>i wanted to try btrfs with guix but haven't had the time to set it up... <nckx>No, IIRC it was so broken it was scrapped and replaced, but I don't know if the replacement is ready. One rather loses interest after such a mistake 🙂 <mfg>i have used FreeBSD and ZFS for a long time, but sadly ZFS has a weird license... <nckx>~ bcachefs all the way ~ <abcdw>mfg: afaik, brtfs is a default in Fedora now. <mfg>abcdw: lel didn't RHEL abandon it? <mfg>just reading abotu bcachefs :D <mfg>Let me search ... one moment <maav>mfg: rhel yes, fedora stays <nckx>Just the word ‘abandoned’ sounds like it came from a clickbaity article. Who talks like that. <mfg>nckx: possible :D I can't remember where i originally read that... <wehlutyk>I'm having a couple issues upgrading opencv to 3.4.12 <nckx>mfg: It was years ago, too. IIRC the ‘btrfs guy’ left the team, nobody took over, the end. This was then spun into nonsense articles. RHEL isn't a distro, it's a ‘enterprise product’. Support == $$$$. Btrfs was a tech preview, nothing more. RHEL supports ext and XFS. That's it. Yet no clickbait that RHEL ‘abandoned JFS’. <wehlutyk>if anyone wants to have a look, let me know! <jonsger>nckx: your efibootmgr tip created fruits :) <mfg>thanks for the ckarification nckx :-) <seeg123>i incrementally modify my /etc/config.scm and run guix system reconfigure but it just takes longer and longer to finish even though i added only 1 package, is this normal or am i doing something wrong? <seeg123>str1ngs i use awesomewm, i think it's x11 only <seeg123>in fact i see it redownloads everything when doing guix system reconfigure <mfg>even adding only 1 package can mean that you have to add a lot of packages due to dependency chains... <mfg>i'm not sure about the redownloading though, normally guix doesn't redownload items already in the store... <leoprikler>only if you `guix gc` in between and the dependencies can thus not be correctly resolved <maav>because you're then updating the system too, not just installing that package <maav>nonetheless, it has to take a bit longer each generation, as the old entries for the bootloader have to be generated and this is O(n) right now <maav>but should be negligible compared to the system generation itself <roptat>or you're using the guix installed in the system profile, in which case you're actually downgrading every time you reconfigure <civodul>nckx: i think that's half of our aarch64 build power or something? :-) <civodul>but anyway, do what you can, i understand it's even more complicated these days... <civodul>we can make the release even with fewer build machines <seeg123>it warns about doing guix pull, but i did and it still warns me <zimoun>does Guile apply some deforestation/fusion techniques? Well if it makes sense for Scheme context. <roptat>seeg123, can you check ~/.config/guix/bin is first in your $PATH? <roptat>also that the output of "type guix" is ~/.config/guix/bin/guix and not /run/current-system/...? <seeg123>i have only ~/.config/guix/current actually which is a symlink <seeg123>type guix is actuall /run/current-system <ss2>hey, I'm trying to (modify-serices ...), and this declaration is always failing on me now. <ss2>Or, how do you add authorized-keys to %base-config? <seeg123>"/run/setuid-programs:/root/.config/guix/current/bin:/root/.guix-profile/bin:/run/current-system/profile/bin:/run/current-system/profile/sbin" <roptat>sneek, later tell raingloom I've looked closer at the files, and even though there is no license file, the files contain some metadata, including "Braille Institute of America, Inc. provides Atkinson Hyperlegible for use, without derivatives or alteration, to the public free of charge for all non-commercial and commercial work. No attribution required." which I believe is OK for a font, but not sure exactly how to represent it <roptat>seeg123, oh if you're running guix as root, you need to run guix pull as root too <roptat>so you have /root/.config/guix/current/bin/guix? <roptat>try "hash guix" then, see what "type guix" tells you <seeg123>ah actually 'which guix' is ok, /root/.config... <roptat>cool, now you should be using your guix pulled version of guix, instead of the system-installed guix (which is always lagging behind because guix cannot define a guix from the future) <roptat>that should help, though your next reconfigure might take some time <nckx>seeg123: Use ‘type <command>’. ‘which <command>’ isn't very accurate. <nckx>civodul: <that's half of our aarch64> Yeah, it's a big problem. <sneek>Welcome back raingloom, you have 1 message! <sneek>raingloom, roptat says: I've looked closer at the files, and even though there is no license file, the files contain some metadata, including "Braille Institute of America, Inc. provides Atkinson Hyperlegible for use, without derivatives or alteration, to the public free of charge for all non-commercial and commercial work. No attribution required." which I believe is OK for a font, but not sure exactly how to represent it <roptat>so I didn't have to call them in the end :) <roptat>also, I packaged otfinfo in lcdf-typetools, so I could read the metadata <ss2>Hi again, would anyone be willing to help me regarding authorizing keys in /etc/guix/acl? <roptat>so, I could find the license in the OTF, TTF and SVG files, but not the others. I think they mean to distribute them with the same license, but you could argue the license only applies to the files it's present in <ss2>And it always fails, and I've tried quite a few other ways trying to get the (modify-services) into my decleration. <ss2>What am I doing wrong there? <roptat>ss2, how does it fail exactly? what's the error message? <janneke>ss2: one thing that's wrong: remove %desktop-services at the bottom, and use it instead of %base-services for the modify-services call <janneke>but yeah, an error message could be helpful <roptat>ss2, oh I see what's the issue: (modify-services ...) returns a list of service (where the guix-service-type is modified, and the other services are the same as %base-services) <roptat>so I'd rather remove your modify-services line, and replace the %desktop-services at the end with (modify-services %desktop-services ...) <janneke>ss2: yeah, thanks -- use what roptat says; i missed the list-within-list issue :-) <ss2>oh man, thanks a lot.. That just took me ages. I was wondering why it should be inside the list, and it failed early on while I had it outside (and then didn't modify base to desktop) <roptat>maybe we could propose to use flatten by default? then you can happily mix services with lists of services <ss2>I think I'm still confused about the differences there. <ss2>anyway, so, desktop-services inherits from base-services, and I can modify the properties? <nckx>Does anyone use LibreOffice Base? <nckx>Maybe I should rephrase that: has anyone used Guix's LibreOffice Base, ever? It seems fundamentally broken. <roptat>nckx, never used it before, it tells me it's missing "hsqldb" <roptat>which sounds like a java package I've encountered before... <nckx>That's the old format, I wouldn't bother with it. The new one is Firebird, which seems like a pretty cool DBMS in itself, so I've packaged it and it fixes LO. But I'm a bit puzzled by the implication that nobody, not a single Guix-using body, uses Base. <nckx>Or do they all use it as a phpMyAdmin on drug^Wjava and connect to postgres? I am puzzled... <roptat>I never used base before, I have no idea what it could be used for ^^' <nckx>It's what most people thorougly abuse spreadsheets for 😉 <nckx>Just a database in office suite form. <nly>how do you answer the question "why am i building webkitgtk" on a reconfigure? <civodul>maav, roptat: how d'you translate "code staging"? i don't have a good answer for that, perhaps we need help from an native English speaker <maav>hmmm, i understand it as a platform/level system, like layering but in a sense of construction <maav>like phases in a factory line <maav>and the stratum in a geological way has certain similarity too, i've used mainly this for my translation <PotentialUser-53>hey guys o/ does anyone know how to specify a 9p mount point in the config.scm ? <maav>in spanish it depends a lot on the context, in construction sense they would be "andamiaje de código", but i used a fuzzier translation and took the long path with "etapas de evaluación de código" IIRC <maav>PotentialUser-53: with stuff do you mean /dev/sdxN? <PotentialUser-53>thats the problem here .. (device ..) seems restricted to a label, a device or a uuid <maav>PotentialUser-53: i thought that everything was a file on plan9 :) <mfg>i only use 9p to mount host folders inside my QEMU guests, never tried to use it inside of an os config... <vagrantc>so, as the spelling/typo/grammar fixes on guix master suggest, i managed to sucessfully build debian packages of guix again finally <PotentialUser-53>but since you cannot put anything into the fstab on guix there ought to be a way to define that somewhere, no ? <vagrantc>and I even managed to track down which test suites fail and disable them explicitly (mostly network and bootstrap binary related) <roptat>civodul, mh... the manual has "mis en scène du code" which is obviously wrong <maav>vagrantc: great! thank you :) <maav>you may be helping to bring a friend of mine from the debian world too ;) <vagrantc>there's one thing I don't understand though, "make install" installs ./usr/lib/x86_64-linux-gnu/guix/guile ... what is that? where does it come from? <vagrantc>is it a wrapper around guile? a copy of the bootstrap guile? <maav>it's a wrapper, you can find info in Makefile.am: # This is our variant of the 'guile' executable, one that doesn't complain about locales. <roptat>civodul, maybe we can find something with "échafaudage" <maav>i didn't know the word, but andamiaje in spanish is exactly that :) <vagrantc>maav: thanks for the explanation and pointer ... now need to figure out how to install it in libexec instead of lib *vagrantc has some crazy ideas to try and package all the guix and/or mes dependencies with both guile-2.2 and guile-3.0 support in a single package... <civodul>roptat: "mise en scène" is most likely incorrect, "échafaudage" sounds interesting <vagrantc>as i'd prefer guix to use guile 3.0 but mes still requires guile 2.2 ... and ... too much beurocracy makes me not want to make a new version-specific package in Debian ... that's one thing I love about guix :) <civodul>maav: i didn't know "andamiaje", neat <maav>vagrantc: isn't being installed there? it's defined as pkglibexec <vagrantc>maav: ah, interesting ... i might need to fix my packaging then <maav>civodul: i didn't knew neither scaffolding nor échafaudage :) <civodul>maav: oh right, i hadn't realize scaffolding was related to that meaning of "staging" <vagrantc>weird, guix-cookbook.de.info is missing dircategory and direntry ... but the others aren't? <vits-test>PotentialUser-53: labels work only for some fs-types, btw. <vits-test>PotentialUser-53: > Currently Guix System only supports ext4, btrfs, and JFS file systems. In particular, code that reads file system UUIDs and labels only works for these file system types. <civodul>maav, roptat: but then, how d'you translate "the code is staged"? <vagrantc>and of course, the guix suite of manpages is all boilerplate :) *vits-test *ржание* "du codee le stagee"! <maav>civodul: probably you cannot translate that alone, but some other context is implied, e.g. meaning staged for future execution you have to incorporate that into spanish as "el código se prepara para su ejecución en una fase posterior" <vagrantc>"program - manual page for program" is not a particularly helpful manpage summary :) <winux>Hi! I would like to know that if there is a guix package for qtile? <sneek>winux, nckx says: That card's driver's firmware is non-free. We don't support non-free software here, sorry. (RTL810xE is a series, not a model, hence the ‘x’, but none of those firmware files look free.) <maav>civodul: even "prepara para una fase posterior" alone could be enough if that context is implied in the translation too. <civodul>maav: yeah, "para una fase posterior" is explicit and clear at least <roptat>civodul, do you have a sample sentence with "the code is staged"? <PotentialUser-53>Yea that's how it looks like, but it's sortof very restrictive. Is there any other way to put custom contents into fstab ? Or is there perhaps a startup hook where I can run that mount command ? ... I guess i could create a daemon out of a mounting script - a bit excessive :D <maav>civodul: otoh, if you are using the scaffolding path in spanish we have a verb for that but would be too metaforical, as "andamiar" is used a lot by construction workers <vagrantc>WAT? I: guix: file-references-package-build-path usr/lib/x86_64-linux-gnu/guile/3.0/site-ccache/gnu/ci.go <roptat>il est échafaudé pour être exécuté plus tard <maav>roptat: i would go to something like "it's moved to an upper stage of the scaffolding" if that metaphor sounds good to you <maav>because here the reference implicit in the staging are the workers doing the job on each of the stages <maav>(i don't want to remember Bourdieu too much here, as that omission involves more topics than the ones of this channel;-) ) <vagrantc>i should follow-up to the debian bug regarding packaging guix and CC devel with some of these remaining issues <vagrantc>and work out patches to force disabling network tests <vagrantc>but it's basically within plausibility at this point <zimoun>civodul, roptat, maav: it is interesting how bad Git translates “stage” in Fr and Es. Section 7.2 of Pro Git book <vagrantc>ah, the other major feature would be to do with gnu/packages/patches/* the same as the bootstrap binaries... <vagrantc>e.g. download from a substitute if not present ... but maybe there's a bootstrapping problem there? <vagrantc>i guess i could write a service that checks out the git repository at the appropriate tag and symlinks them? <roptat>oh, échelonné c'est pas mal aussi <vagrantc>if i changed a few more grammar issues in package descriptions, would this be a bad time? there seem to be some translation discussions going on <maav>zimoun: it's really hard to translate, as it has to be done consistently through lots of places... but yes, i feel the translation too loose in this case as only translates the what but not the how implicit in stage <vagrantc>mostly 'allow(s) to FOO' to 'allows(s) FOOing' or 'enable(s) FOOing' <roptat>my understanding is that staging is putting code somewhere that's being picked up later <maav>i think packages aren't receiving too much attention, only roptat has some work there :( <roptat>it's not the process of picking it up <maav>roptat: exactly, like leaving the tools in the second floor to use them later <roptat>it's very hard because you sometimes need expert knowledge in the field to translate the description of packages... <roptat>maav, more like leaving them on the floor or in a bucket that is raised to the upper level later <maav>roptat: that works too, i guess we can agree they are being used later :) <roptat>maybe simpler: "le code est préparé pour être exécuté au niveau suivant"? <maav>roptat: i just checked mine and i have "se ha @dfn{preparado} para una ejecución posterior", so changing metaphors (scaffolding vs. production line) it's exactly the same :) <zimoun>roptat: « pour une execution posterieure » c’est pas mal aussi :-) <roptat>the only sentence it appears in right now is "it is staged for later execution" <roptat>la traduction actuelle est malheureusement "elle est staged pour une exécution ultérieure" <roptat>donc "elle est préparée pour une exécution ultérieure" ou "elle est mise en place pour une exécution ultérieure" <maav>c'est bon pour moi, mais je ne suis pas français :) <roptat>that was more for zimoun and civodul <roptat>user_oreloznog, you're Hubert right? <maav>there is another place where i kept staging as a quote, because the paper isn't translated, btw <maav>together with the translation <roptat>speaking of quote, how do you translate "quoting" as in "using the quote function" <zimoun>hum? my French is not good enough to see a difference between “posterieur” and “ ulterieur” :-) "elle est préparée pour une exécution ultérieure" sounds good. <roptat>I think you need to say after what it is "postérieur à qqch", not sure you can use it as an adjective (well, it sounds weird to me) <maav>roptat: in the context index i use the meaning: literals and evaluation inhibition <maav>unquoting is introducing the value into the literal expression <civodul>roptat: « mise en place pour une exécution ultérieure » me paraît bien <zimoun>roptat: I do not know about {pos,ul}térieur and cnrtl.fr is not really helpful here. About quote/unquote how Guile does? Well if the Guile manual is translated. ;-) <civodul>not metaphorical, but very explicit and clear <roptat>ok, we need Hubert to make the change, otherwise we'll have a conflict <zimoun>yep I am not able to find it. And Racket neither. <zimoun>civodul: cool for the transformation update. <Bumblehorse>Hey guys, I told myself if I could recreate my debian system on guix, then delete it and reproduce it again with profile manifests etc., I would make Guix my daily driver. I was wondering if you could double check something for me before I delete all my hard work. <mfg>I would suggest to just do a backup of it, before erasing anything :-D <maav>i'll be afk for a bit, later i'll send v5 for grub locales (44196) as the second test with btrfs subvolumes showed that i had the same problem that i was solving elsewhere here... <civodul>ah yes, i'm sorry for the delays there... <civodul>it takes time for me to wrap my head around it <civodul>would you like it to be part of 1.2? <maav>i hope it can, they are bug fixes, the /boot part might wait, but it's a bug open long time ago <Bumblehorse>My plan is to take the file from "guix describe --format=channels" and use "guix pull -C FILE." on it, then use "guix system reconfigure FILE.scm" and then use "guix package -m FILE.scm". I have two questions is the manifest file simply the file called manifest within my .guix-profile directory? Second, is there anything else that I am missing! yesterday I was told about some files in /etc and /var that <Bumblehorse>services would write too but I can't find anything about that and I'm not sure whether it's even worth it <Bumblehorse>Thank you guys very much in advance. After this I should be out of your hair <maav>civodul: i had prepared the trivial ones to be pushed for sure, this one was a bit trickier as some use cases were not so clear to me at the beginning <lfam>Bumblehorse: I recommend looking at the files in /etc, looking at config.scm, and seeing if there is anything that looks like it wouldn't be recreated based on config.scm. Maybe there is some stateful configuration created by the service but I can't say one way or the other <lfam>As for /var, I recommend a similar approach <Bumblehorse>lfam: other than that I am all good? (besides backing up my home directory along with the files for recreation) <lfam>In principle, yes. But I echo civodul's recommendation of a backup <lfam>Services are where the stateless design of Guix meets the stateful reality, and there could be surprises, especially for a workstation or personal computer rather than a server <lfam>I echo mfg's recommendation of a backup <Bumblehorse>lfam: this is all on a separate machine. If it works out I will then do it on my main machine, which I will backup <civodul>"stateless design meets the stateful reality", i like that <lfam>Okay. I think your plan sounds good bumblehorse <lfam>civodul: I'm paraphrasing something that Dave Thompson said a long time ago <zimoun>Bumblehorse: no the file ~/.guix-profile/manifest cannot be used with --manifest file.scm; if it is what you are asking for. <lfam>Also listen to zimoun :) <civodul>lfam: sounds a bit like "it's where those functional hippies have to face reality" ;-) <lfam>I misread the part about your package profile manifest and zimoun is correct <lfam>Heh, I don't think that was Dave's intent <lfam>More like, "this is where it gets hard" <lfam>For example, a mail server. The mail spool can't be recreated based on config.scm <Bumblehorse>zimoun: Is that so? How do I create a manifest file to load with guix package -m then? <zimoun>roptat: quote/unquote is hard to translate. I was reading a bit the Emacs Lisp intro spannish translation for inspiration. Nothing! )-: <civodul>zimoun: what d'ya mean "nothing"? how do they translate it? <zimoun>Bumblehorse: basically a manifest is a list of packages. <Bumblehorse>zimoun: Ooh, I'm sorry. So guix package -m FILE creates the manifest? I thought that was to command to install packages from a manifest <zimoun>civodul: by beautiful sentences. ;-) Well, “littéral“ or ‘expression’ or ‘inévalué’ <zimoun>Bumblehorse: No ‘guix package -m file.scm’ populates the profile with the packages listed in file. <zimoun>lfam: thanks, I was trying to find this link :-) <Bumblehorse>zimoun: Well thats what I thought, I'm just having trouble generating the manifest file, I don't have to write it by hand do I? <lfam>I think this discussion shows that we could use a cookbook section like "Recommendations for recreating a stateless system" <mfg>I always write those bu hand :D <zimoun>Bumblehorse: you have to by hand. For now the converter is not ready. <Bumblehorse>I have been confused about this for quite some time now lol <lfam>It's possible to write a little script that translates a profile into a manifest, but this is fundamentally a "lossy" conversion. These scripts have been shared before but I sadly don't have one handy <zimoun>and this converter cannot be a real converter, just something that approximate a stateless manifest from the internal <profile>/manifest. <lfam>This is good feedback. It shows we are missing some tools :) <lfam>Bumblehorse: Please don't worry about being in our hair. We are all here to help <lfam>I feel like this could conversation presents a good subject for the Guix Days <lfam>Um... sorry about my fractured sentence <zimoun>I have started something but I am running out of time. Wow! One month ago, already. <roptat>I've tried to find some French tutorials about Scheme, but they seem to always use "citation" or "la quote" which is terrible <roptat>"La quasiquote (ou accent grave : ` )fonctionne comme la quote à ceci près que certains arguments peuvent modifier son comportement : notamment unquote (ou virgule , ) qui force l’évaluation et unquote-splicing (virgule arobasque : ,@ ) qui force l’évaluation en supprimant le premier jeux de parenthèse du résultat." <Bumblehorse>Maybe I'm just ignorant, but wouldn't a tool like that be simple? If we are comparing it to writing by hand, I would just go through my "guix package --list-installed" and write them down in manifest syntaxt anyways right? <zimoun>roptat: spannish use « cita » which means ’citation’ <roptat>but then, there's nothing really for "unquote" <roptat>and it can't really be used as a verb either <zimoun>lfam: yes. Or it could be an Outreachy proposal for the next round. :-) <lfam>Bumblehorse: Yes, exactly. But I think the addition of this tool to Guix has been resisted because, like I said, it's a lossy conversion and would give some false confidence <lfam>Good idea zimoun! Well we could talk about it at the Guix days and then welcome people to solve it with Outreachy <Bumblehorse>lfam: Could you elaborate on the "lossy"ness? What kind of data would you be losing? <lfam>Bumblehorse: When you do a profile operation, such as `guix install vim`, Guix decides what to do based on a few factors. E.g. The Guix you are using — created by `guix pull`, including any channels you might have added, and also any external packages from GUIX_PACKAGE_PATH, or even form a development environment using the ./pre-inst-env mechanim (see the manual section Contributing for more info). <lfam>I mean, even from a development environment <lfam>There are also things like package transformations, input rewriting, etc, and they would need to be eaccounted for <lfam>So, it's not as simple as just adding the package names and versions to a manifest.scm file <lfam>This theme of "lossy transformations" pops up in a lot of contexts with Guix, by the way <lfam>We are slowly adding the tooling to overcome it <lfam>We are lucky that the project is so busy :) <lfam>Well, it's not just a matter of luck, IMO :) <zimoun>lfam: the good point of the story is the really fresh #44321 :-) <Bumblehorse>I've never submitted and contributions to anything before, but I may now out of necessity. <lfam>Now that looks really interesting zimoun! <roptat>civodul, I just saw your message. texinfo doesn't seem to be doing the right thing with ``...'' <zimoun>roptat: difficile sujet, peut-être littéral, évaluation {pos,ul}térieure, expression etc. <lfam>That looks really useful Bumblehorse! We have a couple musicians active in Guix, so musical software is pretty well represented <Bumblehorse>lfam: Packing something like that isn't very hard though correct? It only gets hard when you have very complex way to build something right? <lfam>Bumblehorse: Glancing at the source code, I'd guess it will not be hard <maav>i'm back; roptat: in spanish unquote is "fin de la cita", end of the quote <roptat>that doesn't really make sense, does it? <roptat>unquote is inside the quote, not the end of it <maav>it's used a lot in written discourses, the spanish president once read it aloud ;) <roptat>maav, yeah we have the same in French "fin de citation" <maav>it would be like "citing somebody text" with something interleaved "because that's the way they ended" <lfam>Bumblehorse: Some might say it's our secret weapon ;) <roptat>but in `(a ,b c), ",a" is not really the end of the quote, is it? :) <maav>translated to natural language (lisp is an AST) it would be "a" b "c" <maav>so it makes sense in a written discourse "a [unquote] b [quote] c" <civodul>roptat: yeah, you're right, i just checked the manual (info "(texinfo) Inserting Quotation Marks") <civodul>so i guess you could amend my patch to use guillemets <Bumblehorse>(Just incase someone out there might want packagin ideas) I also noticed that we don't have ufw or the czech aspell dictionary <roptat>civodul, yeah, and again I'll need Hubert because I'm letting him do all the work. I don't want to get conflicts between our two versions, so I'll send him your patch, is that OK? <roptat>(see how clever I am "look, it's only a small 10% left..." :D) <roptat>user_oreloznog, just on time when we need you, great! <zimoun>user_oreloznog: thanks for your translation work :-) <roptat>as you might have read, we were discussing the French translation, and there a few things we can improve :) <roptat>user_oreloznog, first, let me forward you the message civodul sent me <roptat>it contains a patch with some improvements <roptat>alright, in this patch there are some places where guillemets («») are replaced with `` and '' (texinfo syntax), but that doesn't work. can you revert just that part to use the French punctuation? <maav>btw, po/packages/POTFILES.in needs an update, as the current pot doesn't include most of the files <roptat>the easiest might be to apply the patch, then fix the result <roptat>maav, I think I sent a message about that one or two years ago... :D <roptat>that's not your fault, it's actually civodul who introduced them by mistake <roptat>`` and '' are used in texinfo for quoting. It's useful because it's translated to “ and ” which most people can't type directly, but they're not the symbols we use in French <roptat>yeah, but that should be available on your keyboard (it's just well hidden, and in a different place from my keyboard probably). I suppose you use azerty? <roptat>apparently it's Alt Gr+[ for "«" and Alt Gr+] for "»". <roptat>(I use a bépo layout, so for me it's even easier) <roptat>but well, first step: can you apply the patch as-is please? <roptat>then fix the guillemets in the translation (not in the msgids!) <jonsger>nckx: repairing with efibootmgr was a full success :) <roptat>would you like to send me the file you currently have? I can apply the patch myself and also do some proof-reading, then send the result back to you later this evening? <user_oreloznog>And the only I have, I've downloaded it since One hour, I put it on my computer <roptat>user_oreloznog, did you change anything from the version on the TP? <maav>well, it seems that we're missing almost 20k msgids from the packages <roptat>ok, then I'll just do my stuff, and send the revision to the TP <roptat>maav, isn't that like 4× what we have now? <user_oreloznog>I live in a far fractured numeric zone, you'll receive it in a few seconds <maav>5x+, but it's on the ball park, thinking about the current number of packages <roptat>also, an advice for your future work: you need to put two spaces between sentences, it's important for texinfo to work properly <maav>i've just removed all the lines except for the comments on POTFILE.in and did this: find gnu/packages gnu/services | sort >> po/packages/POTFILES.in <roptat>user_oreloznog, also you can translate the content of @dfn and @emph <maav>roptat: is that really like that? I don't do that in my translation... :( <maav>oops, i missed those two in my cheat-sheet :) <maav>roptat: should i push the change to POTFILES.in? <roptat>I'd ask civodul first, but no objection from me <jonsger>has someone experience in Guix system on 4k. So all about scaling <mfg>jonsger: i'm using sway on a 4K monitor with guix system. <mfg>but i don't use scaling :/ <apteryx>jonsger: I do. It's a bit of a pain with xorg. <jonsger>mfg: so how and where do you make your settings <roptat>maav, mh... maybe I need to re-read the texinfo manual, but I'm almost certain it uses the double-space to properly detect end-of-sentence <apteryx>It defaults to like 72 DPI (but perhaps full blown desktops like GNOME would do something about it -- I use ratpoison). <mfg>just that: output DP-3 pos 0 0 in my sway config the rest works automatically... <apteryx>So I configured manually my DPI, and it's a global thing, so if you have a 2nd monitor which has lower DPI you'll have to scale it otherwise it'll look huge. This is the reason I'm looking forward to try Wayland (it has per monitor DPI). <maav>user_oreloznog: no problem at all, i learned some of them by trial and error like this roptat just said, because it might be the reason some paragraphs are justified weird <roptat>user_oreloznog, also, and that's related to a bug in po4a, when you see things like "[#:option value] A description of some sort", you should insert a line break, so have two lines: "[#:option value]" and "Une description traduite" (don't translate the content of [#:option value]) <mfg>though i did change icecat settings: layout.css.devPixelsPerPx to 1.25 <mfg>otherwise everything is to small in there <jonsger>apteryx: I guess I will give sway another shot. It's years since I tried it last <jonsger>hopefully is at easy as i3 to run on Guix... <maav>roptat: reading texinfo manual seems that it should work as soon as there is a punctuation symbol and any space after it (@. avoids that) <maav>civodul, what do you think about pushing the POTFILES.in for packages generated with 'find gnu/packages gnu/services | sort'? the file is more than five times bigger, but that's the real picture of such a big project :) <user_oreloznog>zimoun: happy to be able to contribute, thank you! But I need to learn a lot... <roptat>maav, wait, that's taking too many files into account <roptat>at least add a -name "*.scm" to the find command <maav>yes, i didn't copy that part :) <maav>the only bit that changed is that tex.scm is now after texinfo.scm, i'm not sure if sort order is better, or should i use other flag? <roptat>user_oreloznog, oh also something that amazed me when my teacher told us that in high school: "actual" means "réel"; "actuel" can be translated into "current" in English; "current" is not the same as French "courant" which is actually (= "en réalité", not "actuellement" see?) translated to "usual" :D <roptat>no sort is alright, I don't think gettext will have any trouble with that <roptat>yeah, there are so many... English is hard <maav>just a matter of keeping it consistent, thanks <maav>wow, is taking ages to update all the po files... <maav>civodul, roptat: about the POTFILES.in issue... the following make is taking ages, so maybe it's worth to follow that up, or include in the same commit the updated po files to avoid regenerating them, wdyt? <user_oreloznog>roptat: I have to go, feel free to send me your return if you need. I think I can continue tomorrow, for learning and apply new knowledge <roptat>user_oreloznog, sure, I'll send my proof-reading to the TP, and to you <maav>* note to self, test the other cases! <davidl>when I run guix system reconfigure -L /path /path/config.scm guix just silently errors out for me... echo $? prints 1. <apteryx>davidl: you are running this command with sudo, correct? <davidl>apteryx: as root after sudo su -l <davidl>cuz I haven't set a root password I think <davidl>apparently, there is no complaint about the config. I switched config and now it works... <apteryx>weird, can you share that config to see if others can reproduce? <apteryx>does it also occur when you do 'guix system build your-config.scm' <davidl>however, I have messed around forever with getting cuirass to work, and been getting fibers errors. I was also able to cause some other bug, like it outputted something about gmail in the end when searching for packages. <davidl>I am able to start cuirass and successfully build from 2 additional channels, but when picking packages that seem large, I get errors from fibers. <davidl>apteryx: Ill share the config probably later tonight. <maav>davidl: su shouldn't be needed, you could use sudo -s to get a root shell <davidl>maav: indeed, and now it complains about the config! <maav>davidl: ur welcome, i hope the problem can be easily fixed :) <roptat>but it's hard because po4a doesn't split the file correctly <roptat>civodul, I'm having troubles translating "keyword parameters" properly <roptat>user_oreloznog, suggested "paramètres de mot-clés" <roptat>but I'm not entirely happy with that <roptat>maybe "paramètres sous forme de mot-clés" would be more accurate, but that's too long <civodul>roptat: "paramètres nommés" could also work, although that's a bit distant from the initial term <roptat>(also side question: should we apply post-1990 reform?) <roptat>I have "paramètres par mot-clés" <zimoun>roptat: ah la fameuse… je ne sais pas ce qu’il en dans d’autres pays francophones. <civodul>i'm old fashioned :-), but it'd be good to be consistent <roptat>clef is the pre-1990 spelling, clé is the post-1990. I'm younger than the reform and I still use clef all the time ^^' <roptat>zimoun, je crois que la Belgique l'applique exclusivement depuis quelques années, le Québec attend que ce soit la norme en France, je sais pas trop pour les autres pays <zimoun>pour moi, le plus bizarre, c’est les circonflexes qui disparaissent… <zimoun>et certains accents comme règlementaire à la place de réglementaire. Bref! Le plus dur c’est la consistence (-:)- <roptat>c'est pas si bizarre, ça se rapproche plus de la prononciation, et les circonflexes se perdent juste là où ils sont inutiles <roptat>enfin, j'ai l'impression que si on veut que nos textes restent compréhensibles plus longtemps, il vaut mieux essayer d'adopter la nouvelle orthographe (pas si nouvelle du coup) <zimoun>oui oui tout le débat… pour moi, il faut décider pre- ou post- et s’y tenir. C’est à dire corriger quand une incohérence est trouvée, comme une typo en fait. (-: <roptat>civodul, in " It extracts the file name of for out, the default output, and passes it to ./configure" what is "for" for? <apteryx>is there GNU Aspell post-1990 reform French dictionary? <apteryx>or perhaps a way to configure it for it? <roptat>there's hunspell-dict-fr-reforme1990 <roptat>there's also hunspell-dict-fr-moderne not sure what the difference is *nckx never heard of ‘clé’ at school in the 90/00s. <roptat>me neither :p our teachers were not always aware of the reform even in France... <nckx>(I should add: BE, although you all know 🙂) <botsnack>Cette distinction clef / clé peut être utile si on doit faire la distinction entre deux concepts, par exemple une clef de chiffrement et un mot-clé (ou inversement) <civodul>roptat: "file name of for out" looks like a typo <apteryx>when is the file /etc/environment taken into account for a login session? <brendyyn>apteryx: when i packaged a display manager i did that with unix-pam-serivce, which includes pam_env.so <brendyyn>you will see the pam services for GDM, SliM, SDDM include that one way or another <Dzhigit>Guix does not have any repository mirrors, does it? <civodul>each one of us is hosting a mirror :-) <apteryx>brendyyn: oh, indeed, it seems we have to relogin to have this PAM thing reload it <user_oreloznog>civodul: I'll take care of your recommendations. I'll have a look tomorrow <brendyyn>apteryx: yeah the environment is essentially lexically scoped like scheme <apteryx>I was wondering why after installing a network-manager-plugin nm-connection-editor wouldn't have support for it immediately. <roptat>user_oreloznog, I just sent my version to the TP, so you can download it and start working on it whenever you want :) <lfam>raghavgururajan: Are you using x86-64? <lfam>I do get a substitute for 5.9.1 on that system. That's using commit de5659d7d65fe8a8ff579ad2353e6036fb8db2bd (the last commit before 5.9.2) <lfam>I don't expect there to be substitutes for 5.9.2 yet <lfam>Did you try from that commit? If a different commit, which one? *raghavgururajan does `--allow-downgrades` <lfam>I think you can use `guix time-machine` <lfam>There won't be substitutes for any kernels if you are after that commit. I just pushed the update a couple hours ago <lfam>That's with /gnu/store/36yg6a8jj3c3454xdcf5jzmzx1qvcza8-linux-libre-5.4.72.drv ? <lfam>I don't get a substitute for that one either. I'll try building it locally. My guess is some problem with the build farm <nckx>Hullo raghavgururajan (just because it's been so long).