<thelounge>Blackbeard: Unit guix-daemon.service could not be found. <Blackbeard>thelounge: $ ls /etc/systemd/system/ | grep guix <Blackbeard>thelounge: ok please remove guix and let's start again. Please do not close the terminal as the messages will help me to figure it out were things went wrong. <thelounge>Blackbeard: ok running for i in `seq -w 1 n`; do userdel guixbuilder$i; done gives the error bash: syntax error near unexpected token `do' <Blackbeard>thelounge: please copy paste the command and error in debian paste <thelounge>Blackbeard: the error is bash: syntax error near unexpected token `do' <Blackbeard>jayspeer`: or maybe the command needs root, but I figured is easier to see everything <Blackbeard>thelounge: please copy paste the command and output <reepca>ah, makes sense. sudo runs executables, it does *not* treat its arguments as shell commands <reepca>you'd want "sudo sh -c 'for i in `seq -w 1 n`; do ... ; done'" <reepca>(without the outermost double-quotes: "") <lispmacs>I keep getting build error "configure.ac:89: error: possibly undefined macro: AC_CHECK_LIB" <lispmacs>I've added inputs autioconf, autogen, automake, gettext, and libtool. what am I missing? <lispmacs>I can build the source just fine inside environment --pure --ad-hoc gcc-toolchain autoconf automake grep sed coreutils tar xz findutils diffutils m4 autogen libtool pkg-config libusb gawk <reepca>thelounge: did you delete the double-quotes on both sides? <reepca>the following line, exactly copied, works fine on my system: <reepca>sudo sh -c 'for i in `seq -w 1 10`; do echo $i ; done' <reepca>thelounge: okay, then just replace echo $i with userdel guixbuilder$i <thelounge>reepca: userdel: user 'guixbuilder01' does not exist <Blackbeard> /etc/systemd/system/guix-daemon.service, /etc/systemd/system/guix-daemon.service.d, and /etc/profile.d/guix.sh <joshuaBPMan>Hey guix...I seem to have to run ./pre-inst-env as root... <Blackbeard>thelounge: to be honest, I don't know. What I would do is install guix manually and use torsocks where necessary <plstohelp>is there any GNU reason a package containing the branch of visual studio code that lacks branding or telemetry not be accepted into guix? <ecbrown>gnu reasons of lingering bitter animosity toward microsoft? <civodul>as long as VS Code is free and can be made to follow these guidelines, it can get it <civodul>it might be hard to package but that's another story! <plstohelp>iirc the build from source does, and the build from ms does not <plstohelp>:ecbrown GNU developers have a not unentirely unearned repuation for pettiness <civodul>plstohelp: please avoid broad statements like this, it's not helpful IMO <plstohelp>what would one do in a system config file to have guix build from core-updates instead of master by default <civodul>like i wrote, Guix follows the guidelines above <civodul>plstohelp: you can try "guix pull --branch=core-updates" <civodul>but note that it's very much under development *ecbrown scrambles for pencil <plstohelp>i am just wondering if there is anyway to configure it as the "default behavior" <plstohelp>anyway to do that from the config.scm that the instaler generates, or do you have to do that after your first boot <civodul>just do that next time you run "guix pull" <pkill9>isn't vs codium an electron application? <pkill9>unfortunately electron packages can't be packaged because electron is an unpackageable monster <pkill9>but otherwise it could be, e.g. ungoogled-chromium is packaged, even though google provides a nonfree alternative (chrome) <plstohelp>wait, how is electron not packagable but chromium is? <pkill9>electron uses tonnes of node packages i think <pkill9>that all have many more dependencies <pkill9>i think there was work on an npm importer <plstohelp>every day, my hatred for js grows deeper and stronger <plstohelp>god, guix REALLY confuses the gnome disk space analyzer <plstohelp>keeps throwing innacurate LOW DISK SPACE warnings <plstohelp>since tinkering in /gnu/store is discouraged <pkill9>oh that is bad, you should post an email to bug-guix@gnu.org <joshuaBPMan>hello, I'm trying to test a change I made to wlroots. I am trying to update it to version 0.10.1. <pkill9>plstohelp: is there anything you could [remporarily] remove from /var/log? <plstohelp>lol i nuked and started a fresh install 10 mins ago <pkill9>/var/log/messages tends to get quite big <joshuaBPMan>when I run sudo ./pre-inst-env guix build wlroots0.10.1 (I am unable to run pre-inst-environment as non-root), I get an error: no code for module (gcrypt hash). <joshuaBPMan>if I run the above command as a regular user, I get an error: failed to run program guix/scripts/guix ... <joshuaBPMan>running this command...sudo -E ./pre-inst-env guix-daemon --build-users-group=guixbuild <Blackbeard>joshuaBPMan: please use a guix environment and see if it fixes it <joshuaBPMan>Blackbeard: ok guix environment is giving me an error: <joshuaBPMan>guix environment: error: failed to connect to `/var/guix/daemon-socket/socket': Connection refused <pkill9>joshuaBPMan: maybe try restarting the guix daemon with `sudo herd restart guix-daemon` <Blackbeard>joshuaBPMan: is the guix daemon inside the environment? Is the environment --pure? <plstohelp>is there an elegant way to export a list of packages assocated with a user profile? <plstohelp>like pacman -Qqen, but for a guix user profile <plstohelp>is there anyway to feed that data to a user profile, as to say "please duplicate this package list" <joshuaBPMan>Apparently wlroots needs to build with meson version 0.51.2 <pkill9>are you updating wlroots joshuaBPMan? <plstohelp>so the "manifest"file under .guix-profile has your user package profile? <plstohelp>if i were to "manually" update it, how would i force the equivalent of a system reconfigure? <plstohelp>like, for example. if i wanted to export a useer profile to a new machine <xelxebar>The manual has a couple nice examples of how to maintain a declarative user profile. <plstohelp>so, guix package --manifest mymanisfest.scm would import tha user profile? <xelxebar>plstohelp: That sounds right, as long as by "import" you mean "create new generation with specs of file" <plstohelp>this is a core concept about guix that has been confusing me <xelxebar>Yeah. It takes some time to get a handle on all the terminology. <plstohelp>my understanding has been that, when you guix pull, you download package updates, and those updates are applied with a guix upgrade, or perhaps a guix system reconfigure <plstohelp>these packages are stored in the gnu/store until they are needed <plstohelp>why is it then, that whenever i do a guix upgrade or guix system reconfigure, i always see "downloading from ci.guix.gnu.org" for dozens of packages <plstohelp>shouldn't everything have already been grabbed and put in the store by guix pull? <pkill9>guix pull doesn't grab dependencies <pkill9>it does grab dependnecies for guix itself <pkill9>because the package 'database' is compiled <pkill9>so it downlaods substitutes for that compilation if htey're available <plstohelp>iirc i didn't see it download the new patch for linux libre 5.4 until i ran reconfigure <pkill9>plus the guix command which is a package <plstohelp>is that a "dependency" in some esoteric guix way? <plstohelp>the kernel itself as a dependency for something else? <plstohelp>when i ran my first reconfigure, AFTER running guix pull, only then did i see it download kernel 5.4.33 <plstohelp>the kernel is a dependency of some guile thing? <plstohelp>or is it because i did not guix pull as root, so i only updated my user packages <ecbrown>how did you "become root" to do the reconfigure? sudo -E ? <ecbrown>i think there's a difference between sudo -E (inherits user) and sudo su -l (login as root) <ecbrown>that should inherit your user env, iiuc <ecbrown>so a pull as user would provide that to a sudo -E ... reconfigure <plstohelp>so why did it pull the kernel update when it did <ecbrown>not sure. are you on guix system or e.g. debian+guix <ecbrown>(i suppose everything eventually depends on the kernel...) <plstohelp>im just trying to understand the underlying behavior here <plstohelp>like, i understand when and why pacman or apt does what it does <plstohelp>but this is something of an edge case i suppose <ecbrown>plstohelp: one thing you may want to do is plot the dependency graphs <ecbrown>guix graph coreutils | dot -Tpdf > dag.pdf <plstohelp>ill read the manual entry and try your suggestion <plstohelp>i dont have a package called dot, and none seems to exist on the repos <plstohelp>too be honest, i know people say "guix packages take a while to install", and they're right <plstohelp>a lot of that time is the 1-2 sec pause between downloading each package <plstohelp>other package managers just rip through with no delays <plstohelp>that and the downloads are slow but what can you do <plstohelp>not everyone can have a mirror on every street corner <plstohelp>it only seems to show downstream deps, not upstream <plstohelp>and i have 2 different gnu store entries for pulseaudio 13 <plstohelp>guix is downloading some things when i UNINSTALL a apackage <lfam>Don't worry, it's supposed to do that plstohelp <lfam>plstohelp: Basically, when you install or uninstall programs, a new generation of your profile is created. You can view these with `guix package --list-generations` <lfam>The profile includes things that must be built, like man page databases <lfam>My guess is that since the last time you changed your profile you ran `guix pull`, and so now the programs that are to be used to build the profile have changed <lfam>So, even though you are removing something from your profile, the collection of programs required to make profiles has changed, so you see those things being downloaded <plstohelp>yeah, but why would package remove require you to redownload ghostscript or glibc? <lfam>I'd guess those packages were changed with a graft. Otherwise it's not really possible to update ghostscript or glibc very often <plstohelp>if only that damn 2 second gap between package downloads could be eliminated <lfam>Yes, we do have a bug report about that <plstohelp>anyway thank you for answering my inane questions guix irc <plstohelp>i will be a bit busy but i am going to try to package vscodium <plstohelp>or if it doesnt fit the guidelines, ill find something else <Blackbeard>plstohelp: vscodium will probably a bit of work to start. Maybe something smaller and easy could be a good start? <ecbrown>(all the "easy" stuff has been done ;-) <Blackbeard>ecbrown: emacs packages are easy and there are a lot from melpa <apteryx>ecbrown: updating simple packages is often easy, and useful. <sneek>Welcome back apteryx, you have 1 message! <sneek>apteryx, raghavgururajan says: Thanks for pushing the patch. Btw, I tested it on master and the application worked fine. I did not get Mesa related errors. I tested by `./pre-inst-env guix environment --ad-hoc linphoneqt --pure` and `linphoneqt &`. <apteryx>raghavgururajan: I'll try to see if we can install that default ring sound file. <apteryx>hm, well the file is not included in the package source files. *thvk-ivgf (yrffre guna, guerr) <raghavgururajan>apteryx Cool! Btw, I still think it is better to change just package name of liblinphone to linphone. Because, 1) Upstream have not renamed the released source directories yet, as qt-version is still in heavy developement. 2) The working linphone client in master is currently liblinphone, and user has to execute by using `linphone &`. So just changing package name (not variable name) to linphone is g <usney>Is it possible to install guix to an SD card instead of the main memory the OS is installed on? <sneek>MYMYRSRS, you have 1 message! <sneek>MYMYRSRS, rekado_ says: you can’t ping ci.guix.gnu.org because the server doesn’t respond to pings. <raghavgururajan>Where does the "LatGrkCyr" console font come from? Not to able to find which package it belongs to. *thvk-ivgf terminus have larger fonts <efraim>binutils-final on powerpc retains a link to bootstrap-binaries, I just removed the #:allowed-references for now <brendyyn>Does anyone have an idea how many people get paid to work on Guix? *thvk-ivgf "hi-hi, brendyyn the Amateur Tax Inspector" <leoprikler>does %base-services come with vt1-6 or do I need to add something? <leoprikler>next question: what do I do if that doesn't spawn? <leoprikler>It appears ssh-service-type runs, but I can't switch VTs and there's no login prompt on VT1 <craftyguy>I just started reading about guix, and I'm wondering where I can find out more info on how packages are updated in the distro. several blogs list 'no need for package maintainers' as a perk for guix, but a number of packages on guix.gnu.org/packages are out of date compared to upstream. who updates them if there are "no package maintainers"? <brendyyn>craftyguy: anyone can at any time. they packages supported grows as the community as a whole grows <craftyguy>anyone can update any package listed on guix.gnu.org/packages ? <brendyyn>which packages in particular are you looking at it? <craftyguy>1.17 was from March 2019, 1.20 is out now. how does one update that package for all guix users? <craftyguy>(I don't actually have guix installed.. yet. I'm just a long time arch linux user trying to learn more about how this distro works because it's shiny and new :D) <brendyyn>for wayland im unsure. you see updating wayland would update over 1000 other packages <brendyyn>basically you would download the guix git repo, make a patch, test it and email to guix-patches@gnu.org <craftyguy>yeah, that's why I picked it. it's >1yr out of date, and lots of things depend on it <craftyguy>I see, does a human review/accept the patch in that mailing list? <brendyyn>yes there are 40+ volunteers with commit access i think <craftyguy>oh ok, so there are maintainers of the package repo, but not necessarily 'official' maintainers for individual packages. or something <craftyguy>brendyyn: thanks. back to more reading I go :) <brendyyn>yes. civodul believed it may create barriers to contribution if there were such "official" maintainers. there are people that care closely about certain packages and may be effectly maintainers of it <brendyyn>for core packages like gcc and bash etc updating them is done in a separate branch like core-updates. <craftyguy>That's actually quite similar to the way Alpine Linux is run. There are package maintainers that 'keep the ball rolling', but they also accept patches from anyone that wants to update something, provided it passes review and build test <brendyyn>wayland im unsure about. you could ask on guix-devel <rekado>you can provide a patch for the core-updates branch <efraim>wayland is still at 1.17.0 and wayland-protocols at 1.18 on core-updates <raghavgururajan>Quixk question, what is short-cut key bind for auto-indent scheme code in emacs? <brendyyn>raghavgururajan: can you highlight the region and press tab <brendyyn>if you find it via M-x and it will tell you the binding? <raghavgururajan>bash: ./etc/indent-code.el: /gnu/store/4aih30r5zgaivdwpid9bcpyzvicq6drv-profile/bin/emacs: bad interpreter: No such file or directory *raghavgururajan have packaged Midori Web Browser into Guix --> #40712 \o/ <brendyyn>oh god dont remind me how disfunctional that thing is <brendyyn>My friend used to run Puppy Linux with it :). This was maybe 7 years ago <thvk-ivgf>is that ok to mix phases if that reduces a boilerplate (same let*)? Like put `make` in 'configure? <leoprikler>raghavgururajan: (indent-region (point-min) (point-max)) <leoprikler>point-min is the first character in the emacs buffer <leoprikler>if you want to indent just a subregion you have to be more creative in how you select that, but that's not the kind of thing you'd do in an auto-indent script <civodul>i just got a message like that from vlc <TZander>why are webbrowsers like firefox, chromium, waterfox etc not packaged? <civodul>TZander: Chromium contains non-free software and poses serious threats to privacy, so we have ungoogled-chromium instead <civodul>Firefox has privacy issues as well and doesn't comply with the FSDG, so there's icecat instead <civodul>but there's also epiphany, next, nomad, etc. <TZander>oh, I looked through web-browsers.scm, which had nothing mature. <leoprikler>not all browsers are in web-browsers, e.g. epiphany is in gnome <thvk-ivgf>TZander: emacs-w3m inspired me to use an HTML file (with bookmarks) as the home-page *TZander is not an emacs or gnome person. <TZander>hmm, ungoogled-chromium crashes at startup. <TZander>FATAL:platform_font_skia.cc(97)] Check failed: InitDefaultFont(). Could not find the default font <thvk-ivgf>"Essential font packages include gs-fonts, font-dejavu, and font-gnu-freefont-ttf.", "Chinese, Japanese, or Korean ... font-adobe-source-han-sans or font-wqy-zenhei." <TZander>yeah, installing those makes things work <TZander>seems that running guix on top of arch has issues finding the arch installed stuff. I still have no icons in some apps either since the icon-theme I use in KDE isn't available on guix. <TZander>but these browsers now run fine, thanks! <PotentialUser-60>I'm in trouble with a package called "vigra" that not compile. Someone know how to find which package depend on it? To avoid compilation of "vigra" <raingl>PotentialUser-60: seems to be a dependency of LibreOffice and it takes a looot of RAM to compile <ecbrown>guix seems to give numerous choices as to how to do business. a question i have: is there advantage to aliasing emacs="guix environment --ad-hoc emacs -- emacs" or just guix package -i emacs <raingl>PotentialUser-60: had 2G + 4G swap, still got killed by earlyoom <ecbrown>(having realized that i have six or so commands that i use routinely) <ecbrown>perhaps it's an unanswerable question, but hey its sunday am ;-) <raingl>PotentialUser-60: i saw in the `guix package` output that it wanted to build the libreoffice derivation and the vigra one, so i guessed that vigra was a dependency. turned out i was right. *ecbrown wonders if the answer is "pretty much only in cases where one needs frequently needs speed" <PotentialUser-60>raingl: In gentoo exist command to discover that kind of things, It also exist for guix? <raingl>PotentialUser-60: well, i know there is guix graph, but i haven't used it much. there are almost certainly others as well. <thvk-ivgf>PotentialUser-60: /run/current-system/profile/share/guile/site/2.2/gnu/packages/ -- ebuilds are there (as fresh as `guix pull`) <thvk-ivgf>`git clone ...` -- one need it to contribute <raingl>for per user pull: /var/guix/profiles/per-user/$USER/current-guix/share/guile/site/3.0 <raingl>AFAIK that's the one affected by guix pull when you run it as your own user (ie.: without sudo) <raingl>is there a way to detect cyclic dependencies? i think i just introduced one during a recursive import from pypi... <raingl>guix build just keeps growing in RAM usage without printing anything. so i'm preeeetty sure it's a dependecy cycle, but not sure. *thvk-ivgf (if recursion? (invoke "sed" "s|^u^|@u@|g" "faces.txt") "B)") <thvk-ivgf>raingl: idk; is `guix graph gitfs` able to end it's work, or hangs? <raingl>not sure, but i don't wanna experiment too much. i'm 99% sure it did what it did because of a cycle. <raingl>it has like 30 native dependencies for testing, so i'll just disable tests i guess. <abralek>so basically (equal 31622399 31622400) step is failing <abralek>And I am not sure is it a leap second or something else <PotentialUser-60>My system reconfig failed with this error: /gnu/store/gwgjamz3svc2q72piyx1fhbyz2yp85pj-grub-efi-2.04/sbin/grub-install: errore: /gnu/store/gwgjamz3svc2q72piyx1fhbyz2yp85pj-grub-efi-2.04/lib/grub/i386-pc/modinfo.sh non esiste: specificare --target o --directory. <PotentialUser-60>The error was launched by this: grub-install --boot-directory //boot --bootloader-id=Guix --efi-directory //boot/efi/ <thvk-ivgf>PotentialUser-60: please paste.debian.net your sys. config.scm <xelxebar>lfam: In a previous discussion, you mentioned that there is an existing issue on the tracker related to the delay between package downloads. For the life of me, I can't seem to find it. Would you happen to have a link? <mbakke>PotentialUser-60: that error means that you are using 'grub-efi-bootloader', but your system has booted in "BIOS mode" <lafrenierejm>Anyone available for TeXLive help? I'm testing a recipe I wrote for https://www.ctan.org/pkg/moderncv, but I get the error "File `ifluatex.sty' not found" when attempting to compile a sample tex file with pdflatex. How would I go about finding out which CTAN package profides `ifluatex`? *raghavgururajan has to sleep <civodul>i had an idea: we could provide ready-to-use system configs with communications services (VoIP, IM, Git, etc.) <civodul>so that people under lockdown have an option to easily deploy the stuff <civodul>instead of relying on centralized services <g_bor[m]><civodul "i had an idea: we could provide "> civodul: that would be great <raghavgururajan>thvk-ivgf xD. Yeah, we don't need that. https works when 'glib-networking' is added and gio crash fixes when 'gsettings-desktop-schema' is added. <ecbrown>i don't even know what GNU systems can do in this arena. big fan of decentralization <g_bor[m]>civodul : how did the ungoogled chromium thing finally evolve? <ecbrown>some day i wish free software could solve the problem of how to send someone a 10G file <g_bor[m]>It would be a great enabler if we could safely use the method we used there to liberate software. <civodul>the ideal solution would be a web site where you select services and it gives you the config <bricewge>civodul: +1 It would be nice to be able to setup a Video telephony service with Jitsi and Matrix <civodul>but a simpler option would be a blog post or sections in the cookbook <civodul>yeah we'd need to add Jitsi, but i guess that's quite some work <civodul>heh, that's already very good, and obviously Debian is more widespread :-) <bricewge>Packaging such stacks (java, javascript) seems hard <civodul>yeah the Javascript bits in particular are pretty much out of reach <civodul>unless we make a separate channel with not-built-from-source JS, but that'd not be leading by example <g_bor[m]>civodul: I have a feeling for a while... it goes like this: the packages part of guix could be reduced by a huge amount in the following way: ***dddddd_ is now known as dddddd
***deesix_ is now known as deesix
*raghavgururajan thinks about creating ready-to-use VM images (generated from configs based on civodul 's idea), where it can be run on VPS, SBC etc. This establishes decentralized communication channels very fast. <g_bor[m]>Have the output of a recursive importer as a fixed output derivation. <g_bor[m]>Have also a diffset as a fixed output derivation. <g_bor[m]>And use the resulting set of package definitions as a channel <civodul>g_bor[m]: heh, it's tempting, but the work we do "manually" also includes curation work, QA, license checks, etc. <civodul>but "someone" could do that for Node/JS, for instance <civodul>or just do the on-the-fly import we've discussed before <civodul>also, importers don't work for cross-language packages: Jitsi (Java+JS), Mattermost (Go+JS), etc. <divansantana>what to do for binaries that are looking for /bin/bash (guessing). like this -bash: ./vagrant: No such file or directory *divansantana hoping to get vagrant working this way on #guix. <g_bor[m]>civodul: yes, I was not proposing this as a replacement, but instead as an automation mechanism. It could do a lot of the more laborous but boring stuff, and we could do all what we have been doing in the diffset. <civodul>divansantana: if you're working on a package for Vagrant, the patch-shebang phase should take care of it <thvk-ivgf>divansantana: aren't there 'bash:' telling that it can't find ./vagrant? <civodul>g_bor[m]: i guess Someone could give it a try and see how well it works <divansantana>civodul:unfortunately not working on a package. Just trying to get a quick win to get it to work on guix. <ngz>Hello. I have a (mostly) working Guix installation on top of Debian, but it seems that late installation scripts do more advanced configuration than they used to. In order to test new installation script, is it possible to run it on top of an installed Guix? <civodul>ngz: it would be best to test it on a pristine system <mbakke>PotentialUser-60: replace 'grub-efi-bootloader' with 'grub-bootloader' in the config.scm <ngz>Hmm. I'm particularly interested in sys_create_init_profile() phases in the script. <ngz>IIUC, this phases merely copies stuff in /etc/profile.d/guix.sh. Does that mean no ".profile" file in $HOME is necessary then? <civodul>ngz: i think it depends on whether the distro picks up /etc/profile.d/* automatically <civodul>apparently some do, but maybe some don't <ngz>Debian does, from /etc/profile IIUC. So, I can create manually the script there, and remove spurious configuration from my $HOME/.profile? <ngz>Thanks. I'm going to try this. <ngz>Hopefully, this will fix bug#35308 <ngz>Once you know the fix, it is not nasty anymore, but unsetting XDG_DATA_DIRS is not great either. <mbakke>hmm the cython update should probably have gone through staging with its 450+ rebuilds <ngz>Restarting GDM to check if "guix.sh" script saved me. <ngz>At least my ~/.profile is slightly cleaner now. <lle-bout>hello, is there any reason why GNOME is not 3.36 yet? is there a blocker? if so what is it? <rekado>lle-bout: Gnome upgrades often require a full rebuild of all packages due to glib upgrades <rekado>that’s why we’re a bit slow with upgrades <rekado>there’s a branch for a newer version which hopefully will be merged soon <sneek>Sneeky bot running on Guile version 3.0.2 using bobot++ 2.3.0-darcs <lle-bout>rekado: okay, well GNOME 3.36 has stunning performance improvements so it's probably a good idea! <rekado>if you’re adventurous working on that branch might be a good idea *thvk-ivgf "hm, genuine sneek" <ecbrown>rekado: pinging you about r-bridgesampling diff sent to list. like to know whether format is acceptable before hacking on others <lle-bout>how would I test GNOME in a dev env? QEMU/KVM? <rekado>ecbrown: haven’t been able to look at it yet, sorry <rekado>(my wrists started hurting a lot yesterday, so I’m trying to do a bit less typing these days) <rekado>probably not because of typing but because of carrying the baby so much <rekado>but I better take a break to minimize damage <rekado>turns out to be less portable than advertised. <ecbrown>you can eke out some one handed typing, sometimes. quick edit && make -j 16 <lle-bout>been getting 30KB/s while I have ~200Mbps otherwise <rekado>“guix publish” and our nginx config might not be optimal <rekado>ci.guix.gnu.org has two 10G uplinks, so there really shouldn’t be a problem with bandwidth <bricewge>I have the same issue for a dome days now <bricewge>With Free on ADSL and 4G, but civodul doesn't <lle-bout>maybe it's a peering issue, it's possible <lle-bout>Free has had peering issues for some parts of the Internet at times <bricewge>lle-bout: Ludovic is on Free to but wasn't affected yesterday <thvk-ivgf>`ls /run/current-system/profile/share/guix/` what about trying "berlin" instead of ci.guix, can it make any diff? <Kimapr>I have a problem with kobodeluxe: it can't save player profiles <ecbrown>now i'm getting good speeds from ci. <Kimapr>[5829] ERROR: Couldn't resolve player profile path 'SCORES>>Kimapr.1000'! <Kimapr>[5829] Created new player profile '<NO FILE!>' for player 'Kimapr' <Kimapr>[5829] Failed to save player profile - no file name! <rekado>disappointing that we can’t reproduce this reliably <bricewge>lle-bout: When you download a small substitue, what's the speed? <rekado>I don’t have control over the network infra here, so I can’t check to see what’s going on when it’s slow <lle-bout>rekado: I mean you have access to the server, correct? If you open an iperf server we can debug things a little more *rekado doesn’t know iperf <bricewge>lle-bout: I'm quite sure that tunneling through a VPN would make the issue disapear <lle-bout>bricewge: well, it's "fast" but I don't think the speed measurement can be accurate in that case. <rekado>FWIW we do have a firewall on ci.guix.gnu.org, but that’s just for fail2ban <lle-bout>rekado: well you can install the "iperf" package, and then I'll give you a simple command to run, you also need to allow a TCP and UDP port, and that's it. <rekado>the annoying firewall is that managed by the IT department, which applies to the whole subnet. <rekado>allowing any new ports would require IT approval <rekado>best if we can use SSH tunnels or bind to localhost <rekado>lle-bout: could you perhaps write to guix-sysadmin@gnu.org? <rekado>I need to leave soon, but I could follow the instructions later/next week. <lle-bout>rekado: hmm well. I'll try to do with mtr or whatever <rekado>Kimapr: could you perhaps run this under strace -f -o log -s 1024 to see what file it tries to write to? <rekado>we may need to patch the program. <rekado>lle-bout: I can install iperf but I can’t promise a publicly accessible port <lle-bout>rekado: well looks like mtr could figure it tout <rekado>ecbrown: I’ll try to comment on your patch later today <lle-bout>there's loss somewhere around Hamburg on a Level3 router <rekado>gotta go now, be back in two hours or so <lle-bout>It shows loss at some Free backbone router and then at some Level3 one, weird. <lle-bout>how would you report packet loss to Free... <bricewge>Biggest challenge how to bypass L1 support <bricewge>I have around 25% packet loss on my part <vincelevi>hello lle-bout, I am also on free, and also experiencing slowness in dl'ing guix substitutes, so I'm trying mtr, and it also shows significant loss on one of free's routers along the path... <lle-bout>I have Xavier Niel's phone number just in case.. maybe that could help <vincelevi>65% loss on : p11-crs16-1-be1109.intf.routers.proxad.net <ngz>alextee[m]: ungoogled-chromium works <lle-bout>alextee[m]: you can probably figure out what web feature to unblock in the settings of IceCat <lle-bout>It may not be properly labeled with LibreJS <bricewge>lle-bout: I'll try that too. Good luck... <alextee[m]>says acess to the file is denied if you click on those b lank buttons <lle-bout>alextee[m]: well IceCat disables some web features and blocks some JavaScript because it's nonfree by default <Kimapr>rekado: i'm confused. output doesn't change at all when i run kobodl under `strace -f -o log -s 102` <alextee[m]>chromium is asking me to install google calendar extension <lle-bout>alextee[m]: that's a website, not ungoogled-chromium <lle-bout>you can always install Ublock Origin or something to block those annoyances I guess, but yeah it's not default yet on GNU Guix. I'm certain you can submit a patch to install Ublock Origin by default it would be great. <alextee[m]>o.o so jitsi wants me to install google calendar? weird <lle-bout>It's integrated with Google Calendar, AFAIK. But it's Jitsi Meet the official website, if you install an instance on your server it wont do any of that. <Kimapr>logs come out to be too long for debian paste <lle-bout>like, better integrate it within Google Chrome, but it confused you with Google Chrome because you got a similar User Agent to Google Chrome with ungoogled-chromium <lle-bout>alextee[m]: I think it's because the WebRTC situation used to be that Screen sharing wasnt in the web standard so it required an extension, but that's not the case anymore now. <Kimapr>my browser gives me a `ERR_NAME_RESOLUTION_FAILED` <lle-bout>Kimapr: paste.fedoraproject.org might work <lle-bout>yes, they delegated the hosting to them now <Kimapr>'Got a packet bigger than 'max_allowed_packet' bytes' <lle-bout>yeah well if the log file is too big narrow it down to the relevant part where there's an error maybe? <thvk-ivgf>Kimapr: 178.238.45.121 is paste.opensuse.org (at least today) *thvk-ivgf "dolbaniy Cloudflare :P" <bricewge>lle-bout: There are no technical support on sunday. Will try tomorrow <Kimapr>on 178.238.45.121 i was met with this: <Kimapr>> Access through untrusted domain <Kimapr>> Please contact your administrator. If you are an administrator, edit the "trusted_domains" setting in config/config.php like the example in config.sample.php. <Kimapr>> Further information how to configure this can be found in the documentation. <Kimapr>> Nextcloud - a safe home for all your data <lle-bout>bricewge: hmm, they tell me that "there's no problem on my line" when there is.., and then they ask me to do webcam based support, wtf? <lle-bout>Kimapr, yes it's normal, accessing IP addresses directly over HTTP doesnt tell the server what website you want to access to they point you to their default web page <bricewge>lle-bout: lol Well technically it's not a problem on your line but in their network or upstream. <lle-bout>bricewge: "Nous sommes désolés mais nous ne pouvons pas prendre en compte votre message 😔. Nous vous invitons à réessayer plus tard." <lle-bout>if ISPs were a bit more local there wouldnt be such support problems like this <bricewge>True, you can subscribe to FDN if you want a local ISP or another one from the FFDN <lle-bout>bricewge: I'm waiting for FDN to get fiber and then I'm in <bricewge>Hum, newsgroups, maybe that's the way to get someone technical ***jonsger1 is now known as jonsger
<bricewge>lle-bout: Are they plans to provide fiber subscriptions trough FDN <lle-bout>bricewge: yes but currently stuck with legalities and funding <bricewge>I tought the legal framework around fiber vs ADSL was so much in favor of the big ISP that it won't happen anytime soon <lle-bout>I would be amazed if there's anymore from Free on the newsgroups actually but let's try <bricewge>But if there is someone it'll probably be the right kind of person <lle-bout>bricewge: doesnt look like it exists anymore.. rip <lle-bout>lfam: yeah that would be great, also HTTP 2.0? <lle-bout>bricewge: ohh no maybe it's still up after all <lle-bout>bricewge: I used 'Pan' as a client instead of Gnus, it's much easier to use <Kimapr>'The Legend of Edgar' doesn't have button texts show up in menu <efraim>i'm aproaching 10 hours on building gcc on powerpc ... <lfam>What kind of computer is it efraim? <efraim>they have an IDE slot, but I found an IDE drive enclosure with an m.2 slot inside so I'll switch it over one of these days <ecbrown>sweet, might make it a little more pleasant <efraim>ifixit says it's about an hour to disassemble down to the HDD <efraim>that's why I've put it off for 3 months already <jfred[m]>older thinkpads have spoiled me when it comes to repairs <efraim>stock is a 60GB 4200 RPM harddrive <lfam>I had an ibook g4 back in the day. Teardown was suuuuuper hard <lfam>I seem to remember there were a couple dozen different kinds of screws. Just to replace the hard drive <lfam>Better than using a spinning disk to build GCC, though! ***drakonis1 is now known as drakonis
<efraim>I have 2, a 12" and a 14". I had to actually burn a CD to install debian on it <lfam>They had the cool feature where you could "boot" the computer as a firewire external hard drive <efraim>yeah, I pulled out my CD burner and what's left of my spool of CDs and I said "... wait, what do I do now?" <efraim>can't exactly dd an image onto a CD <lfam>I guess everything isn't a file after all <efraim>and that's basically why we now have xorriso with a GUI output <lfam>GNU/Linux is still missing a really solid CD ripper <lfam>There are plenty of options but none that meet a really high standard <lfam>Basically, you have to use EAC (windows) or XLD (mac) <lfam>I guess that whipper is supposed to be good enough now <lfam>It's a long term project without much (any?) industry support <sarg>I've noticed that in my guix on top of debian setup, qutebrowser doesn't use pulseaudio to output sound. When I start the browser - I see Cannot open shared library libasound_module_conf_pulse.so (/gnu/store/nyylgcnzmbw8wrn4sna2crl0g7zxxh33-alsa-lib-1.2.2/lib/alsa-lib/libasound_module_conf_pulse.so: libasound_module_conf_pulse.so: cannot open shared object file: No such file or directory) <bavier`>I've had a good experience with brasero in the past, but some things are broken in the guix package <lfam>It turns out that it's possible to read an audio CD and get files that sound good but they don't actually reflect what's on the disc. It's a tricky early-digital implementation with plenty of room for error while reading <lfam>I'm on Debian too, trying it <apteryx>lfam: I don't get it. Digital files are either successfully read or not at all. It's not like there's some analog to digital conversion or vice versa going on. Or perhaps you are discussing the merits of various audio compression technologies/settings, and that some rippers have better defaults in that regard? <lfam>I don't see the error when starting qutebrowser either <lfam>apteryx: No. It's more complicated than that. <lfam>With CDs, it's obviously possible to fail to read the data due to scratches. But there is a lot of error correction built in to work around this. And it can actually misread but still pass the primitive checksumming in use <lfam>Remember... audio CDs were designed in the 1970s <sarg>lfam: could you please check that qutebrowser's stream is visible in pavucontrol? <reepca>technically that can happen with any checksum, it's just lower-probability with modern ones <lfam>At the moment I can think of any solid citations apteryx, but they do exist <lfam>Audio CDs are a really interesting example of cutting-edge digital technology at the very beginning <reepca>(where "modern" here means "more bits of redundancy") <lfam>You're right sarg, it's not showing up there <lfam>I guess it suspended pulse and used ALSA <lfam>And going back to the other subject, there is *always* an analog to digital conversion going on ;) <lfam>But with CDs the physical layer is really crappy so it doesn't work well <sarg>lfam: yup, and when it happened you should've seen an error in console <apteryx>lfam: Ah, so it's about the performance of the error correction (recreating missing parts). So your argument is specific to the case where the input data (the CD) is damaged to start with, if I'm following. <lfam>Yes, basically. But sometimes the CDs were burned poorly <sarg>guixsd installs /etc/asound.conf where paths to pulse plugins are specified. I've tried to install alsa-plugins:pulseaudio and create this file manually. Now I get another error and still no sound output :() <lfam>There is at least one commercially sold CD that crashes CD players! <lfam>Not on purpose, it was supposed to contain music <lfam>But it somehow wedges the error correction code <lfam>Anyways, music archivists have to care about this stuff and it's a reason they still use Windows or Mac <OriansJ>well there are commercially sold CDs that intensionally install malware on your computer <lfam>Looks like the whipper program is getting there <lfam>Yes, but that's a different problem <roptat>trying to build from git: error: failed to load 'gnu/system/install.scm': Nom de fichier trop long <roptat>maybe my guix is too old, I'll update <cbaines>Does anyone know what the current state of importing golang things is? <sarg>So, after installing alsa-plugins:pulseaudio, making proper ~/.asoundrc file, and restarting pulseaudio - qutebrowser started outputting audio to pulse. Hooray! <sarg>But how can I create ~/.asoundrc with guix? <Enes>I am trying to install gnu guix but i can't pass Wi-Fi section of graphical installation. I pluged in my ethernet cable but i am stil seeing same Wi-Fi selection section <apteryx>raghavgururajan: I managed to test audio calling using linphoneqt with my voip.ms VoIP provider. I got some video stream to show up using VP8 in the preferences dialog :-). <apteryx>I have a patch fixing the missing sound files; I'll push it shortly. <TZander>cbaines: was successful compiling and running something golang based just last week <apteryx>raghavgururajan: works with SIP TLS and SRTP as well for encryption, that's neat! <apteryx>one annoyance is the sound; it plugs directly in alsa (rather than using pulseaudio), which makes it unreliable (I must kill other sound applications otherwise it's muted). <rekado>ecbrown: your patch does not apply cleanly on the master branch <davidl>How can I use the ruby "bundle install" in a docs folder with a Gemfile after I've installed ruby and bundler with guix? <cbaines>davidl, the same way as normal I guess? <cbaines>davidl, I'd be careful with the bundler package in Guix though, if you downloading things from Rubygems anyway, you might be better off gem install'ing bundler. That avoids issues with Ruby versions. <davidl>cbaines: Ok, Ill try uninstalling guix bundler package <cbaines>davidl, I usually use the direnv "layout ruby" to keep rubygems things contained <cbaines>that works nicely with use guix --ad-hoc ruby <davidl>cbaines: Im not very familiar with direnv unfortunately. The main thing Im trying to do is to serve the docs folder as a jekyll website (from some nbdev template), so I need to cd docs ; bundle install ; jekyll serve - but Im failing the dependencies in the Gemfile. For now I think Ill try without direnv. I have: export GEM_HOME=/home/myuser/gems which let me use bundler succesfully, and now it seems <davidl>to work after I removed the guix version of these packages. <ecbrown>rekado: i submitted a new patch to bug#40696 <emys>hi, how can I find out the recipe used for a package? Let's say I want to have a look at how `python-isort` or `python-numpy` have been packaged <ngz>emys: guix edit python-numpy <rekado>ecbrown: thanks, I’ll take a look <xbox>where i can find config files in guixsd ? (bspwmrc, compton.conf and other) <ecbrown>xbox: if you clone the guix git repo: gnu/packages/compton.scm <xbox>i use baremetal install with (gnu packages compton) but i cant find compton.conf on system <thvk-ivgf>xbox: `guix package -I` shows the dirs inside of read-only store (it can hold the examples, default configs that can be copied to ~/) <xbox>thnkx. and second newbir question - how to updatedb with mlocate? <xbox>updatedb: can not open a temporary file for .....blah blah <xbox>i need chmod or something? <thvk-ivgf>xbox: chmod may help if the file is outside the /gnu/store (which should be always ro) <roptat>ok, now from a recent guix (april 16) I get "error: failed to load 'gnu/system/install.scm': Nom de fichier trop long" while building the guix checkout from "guix environment guix" <roptat>(I did run ./bootstrap and ./configure again, just in case) <civodul>roptat: could you strace it to see which syscall returns ENAMETOOLONG <roptat>How so? that's from running make <roptat>also it's on armhf, so it will be long to get to that point (might try on my x86_64 too...) <civodul>hmm can you try to reduce it to something shorter? <roptat>trying ./pre-inst-env guile build-aux/compile-all.scm gnu/system/install.scm <xbox>also i found a bug. guix cant encrypt 2TB hdd (graphical installer) <xbox>i replace it with 128gb and it works <emsyr>efraim: about the enlightenment apps menu problem I wrote about a few days ago. I looked it up today, followed your instruction, deleted ~/.cache/efreetd and did the trick. Now I can lunch apps from enlightenment's menu. <roptat>didn't work: ice-9/eval.scm:293:34: In procedure string-append: Wrong type (expecting string): #f <ngz>I get this : "gnome-session-binary[1526]: GLib-GIO-ERROR: No GSettings schemas are installed on the system" when launching Gnome (after GDM prompt) on a foreign distro, and cannot log in. It disappears if I unset GI_TYPELIB_PATH. Does that ring a bell? <ngz>I can log in if I unset both GI_TYPELIB_PATH and XDG_DATA_DIRS, more precisely. ***amiloradovsky1 is now known as amiloradovsky
<joshuaBPMan>that's weird...I have built guix locally. and It can't seem to find my new updated meson package. <ngz>joshuaBPMan: you may have stale compiled files somewhere. I usually "make clean-go" and compile guix again when that happens. <ngz>(assuming you use ./pre-inst-env script) <joshuaBPMan>that's fine. I'll rebuild it again. And see what happens. <xbox>is it possible to force startx in config.scm ? if i want rebuild guix without any DE. <joshuaBPMan>xbox. You might need to make a shepherd service for it. <carapace>Sorry if this is a dumb question. I'm using guix-system-vm-image-1.1.0.x86_64-linux with qemu; is ssh installed by default? <rain>carapace: if you try enabling it with `sudo herd start ssh` and it doesn't print any errors, then it is. <carapace>In other words, what do I need to do to be able to ssh into the qemu VM? I tried installing openssh. herd start ssh-daemon says it 'could not be found'. <carapace>I just tried 'sudo herd start ssh' and it also 'could not be found'. <rain>did you install it through guix package or through your system config? i think it might need to be in the latter. <rain>as regular user or root? <rain>yeah, the system won't see it then. <rain>i recommend adding it in your config.scm alongside a (services) entry. (eg.: (services (list (service openssh-service-type))) <xbox>bspwm works good. thanks. also the last question - do I always need to update the config.scm or can I just install packages on the system <carapace>where would i find the config.scm file? Do i write one from scratch? (I saw /etc/config.scm mentioned in the manual but there is no such file.) <_tss>you can find examples on the install media in "/etc/configuration" <carapace>I'm using the GNU Guix 1.1.0 QEMU Image from the https://guix.gnu.org/download/ page. I can download the installation ISO, mount it, and look in there, of course but is there a config.scm in the qemu vm image? <carapace>(btw the git repo doesn't seem to be linked from the main website, I was just looking for it.) <carapace>In the manual under 8.16 Running Guix in a Virtual Machine it says "You can also reconfigure the system based on its initial configuration file available as /etc/config.scm" but I'm using the pre-built Guix VM image and that file doesn't seem to be there. <civodul>carapace: yes, it's a bug that we've now fixed on master <carapace>(I was worried I was just doing something dumb.) <civodul>i guess we should fix the on-line manual <roptat>civodul, strace failed building on armhf :/ <craftyguy>hmm, so there's guix the package manager, and guix the distro? that's not confusing at all.. <roptat>the end of the log looks weird too: Throw to key `decoding-error' with args `("scm_from_utf8_stringn" "input locale conversion error" 0 #vu8(1 2 48 11 4 55 12 5 56 9 7 57 10 8 15 49 13 14 55 17 56 31 32 92 39 34 60 60 48 58 58 48 62 62 49 126 127 128 255))'. <roptat>but that's after make: *** [Makefile:7963: check] Error 2 <roptat>(due to 16 test errors in strace's test suite) <civodul>you can spawn the build from another machine <roptat>I haven't set it up for offloading I think, but I'll do that <roptat>oh it's still building the kernel from yesterday, but it has two defunct processes, run from make (so nothing is taking CPU for now...) <civodul>no need for offloading, just do: GUIX_DAEMON_SOCKET=ssh://foo guix build -s armhf-linux strace --no-grafts <roptat>I sent a SIGHUP to one of them, the CPU is doing stuff again (I shouldn't have since I have to build strace now, I'm stupid...) <roptat>oh, the build failed (did I send the wrong signal?) <pinoaffe>just out of curiousity: what's the guix position on packaging foss software that cannot be used without using / interacting with proprietary software? <roptat>scanning it quickly, I couldn't find a sentence that corresponds to the situation, but I think we wouldn't package it (it's useless on guix) <pinoaffe>roptat: not necessarily, since it could be meant to interact with nonfree code running on other systems <roptat>oh interacting with non free code on another system that you don't control anyway is not an issue at all <roptat>that's why we have web browser in guix :) <roptat>do you need non free software for it to be useful at all, or can it be used on a completely free system? <roptat>civodul, re-running make with strace now, it will take some time... ***atw` is now known as atw
<ngz>Still trying (sorry) : does the error shown there https://paste.debian.net/1141418/ which prevents Gnome from starting, ring a bell somehow? This happens on Debian stable and is fixed by setting both GI_TYPELIB_PATH and XDG_DATA_DIRS to an empty value in ".profile". <rekado>ngz: these variables can tell an application to load incompatible binaries <rekado>it’s a real problem, not only for Guix software on foreign distros but also for foreign distro software when Guix sets these variables. <rekado>generally, I think that we need to isolate Guix more by never setting any variables (other than PATH) that haven’t been prefixed with GUIX_ <rekado>this would require patches to a lot of applications, though <lfam>pinoaffe: What do you mean by "interacting with proprietary software"? Do you mean online services? <ngz>rekado: But when I unset XDG_DATA_DIRS, my foreign Gnome does not have access to Guix installed applications. I don't know what I'm losing by unsetting GI_TYPELIB_PATH, tho. <rekado>we’ve got a similar problem with PYTHONPATH which cannot be shared between Guix applicaitons and software provided by the foreign distro <g_bor[m]>rekado : i agree with the isolation you proposed. <rekado>ngz: well, that’s the problem, isn’t it? GNOME from Guix does need XDG_DATA_DIRS, but that same variable will be respected by foreign distro software. <rekado>there really is no good way around this <rekado>if only we could do some more magic with environment variables and make the lookup non-global <ngz>Oh. So everyone's Gnome is crashing on foreign distros? oO; <g_bor[m]>this would also mean that the foreign distro Gnome would not see guix stuff by design. And also that the guix gnome would not see the foreign disto gnome by design. They would live in complete isolation... <rekado>I don’t think anyone uses Guix GNOME on a foreign distro. <rekado>I’m sure there are cases where you’d want to use the plain variable, though. <rekado>with GNOME, for example, you would want to be able to see all applications, not just those that were installed with Guix <ngz>Note that my problem is the opposite : GNOME I use is from Debian. <rekado>I wonder if we could just change getenv in glibc…