<vagrantc>guix could have dependency packages for recommended package sets, such as the way gcc-toolchain pulls in several dependent packages ... that could be where you include the "default" font <kmicu>I tend to think those are real dependencies and should be declared as inputs for a package but that bring some unpleasant trade offs. <nckx>Yeah, I don't think anybody's happy with the current situation but the alternative is (IMO) also bad. <kmicu>But I also make home read only so don’t listen to me. <nckx>kmicu: They're not even inputs, really. They have no effect on the build. That's certainly not unique to icons/fonts but I think it's a sign. <kmicu>(I do that because e.g. an error in gnupg config can render your desktop unbootable so I prefer to delegate everything to Guix.) <nckx>How the hell do you boot, friend. GPG-encryted dotfiles? -everything? <kmicu>nckx: a font can segfault an app. For me that’s an input, it affects a program and must be handled as a real dependency. <kmicu>nckx: ‘unbootable desktop’ not os. <nckx>Dependency, yes, not an input. <kmicu>desktop as in desktop environmet* <nckx>kmicu: Desktop to me is desktop environment. <nckx>How does a broken GPG lock you out your desktop? <kmicu>Then if you source something in xsession and that thing sources some dotfiles the whole xsession can fail. <kmicu>But we cannot rollback in such case because dotfiles are not managed by Guix. <nckx>Sure, that I get, I'm just very curious what set-up you're describing. Not questioning that it exists. <nckx>kmicu: I also assumed read-only home + Guix → guix home (manager), I guess not. <kmicu>A setup with home‑manager where Guix also manages home but I’m not sure whether that’s your question. <kmicu>I’m also lost on dependency is not an input remark. <fnstudio>(since you mention dot files... so i understand those are not handled by guix directly? so, once i manage to switch to guix system, a complete description of my machine will be guix scm file + dot files (+ storage of course); is that correct?) <nckx>kmicu: I wonder how a broken gnupg configuration could break your DE. I do a lot of weird stuff in my .xsession, but a failing command just silently fails unless the subsequent commands actually rely on it. <nckx>fnstudio: Basically, currently, everything under /home/you is your problem. To Guix there's no difference between dotfiles in ~ and documents in ~: it's all ignored. <nckx>This as opposed to /var where Guix will leave most things alone but there's no promise. Services will chmod, mkdir, or remove things if they think it's necessary. <fnstudio>nckx: right, cool; me thinking out loud: but then each app goes and reads its conf from my .config/myapp file, yes, ok i think i get it <fnstudio>nckx: and the idea of migrating from my host system to a hybrid that's more and more guix-based til the point where i can do the switch to guix system... does that sound as a generally sensible migration plan? <kmicu>nckx: it’s a real world scenario so old it’s from Nix world. Somehow a change in gnupg config renderd my Xorg dead. I was able to fix it quickly thanks to snapshots diffing. Iirc it was one incorrect line in gpg-agent’s config. <pkill9>fnstudio: that's kind of what i did, starting with guix as a package manager <nckx>fnstudio: It sounds OK. I've never actually tried it. <kmicu>Details are not important, what’s important is that rollback could not fix the issue because home was not managed by Nix. <nckx>I was curious about the details. <kmicu>I assume that was a combination of NixOS’ desktop services mess… logic. <fnstudio>my cursor/pointer is considerably reduced in size when hovering on my (guix) icecat... do you know if that can be any missing (soft-)dependency? <fnstudio>(icecat 68.7.0esr, although i suspect it's more of a problem with some underlying graphical library...) <nckx>My guess is that Guix's IceCat doesn't completely integrate with your host desktop's mouse pointer theme. <kmicu>fnstudio: it’s possible you need to set DPI for GTK from Guix. <nckx>Oh, Guix doesn't even have cursor themes. <kmicu>You basically cross streams by using desktop elements from Guix and desktop elements from your native distro. <nckx>Theming is *hard*, with so many toolkits and versions of them. <nckx>PotentialUser-57: Thanks for the report. What goes wrong? They work here (Guix System, IceCat). <nckx>I'm playing the 3 under ‘Discover Guix’ on the home page. <kmicu>[Jokin’] Maybe Apple only allows DRMed videos 😺 <nckx>We use utterly standard & boring HTML <video> elements, nothing fancy whatsover, this should work everywhere videos are played. <lfam>I bet the ipad does not support this codec <nckx>Now I don't know much about the Internet, but webm… I mean, it has Web in the name. And M! That has to stand for something! <nckx>(Anyway, I'm surprised; it's not obscure AFAIK.) <lfam>iOS does not support Opus or Vorbis audio so they probably do not support VP8 / VP9 <lfam>I've been struggling with this regarding internet radio. The only thing that works on all platforms is MP3 <nckx>pkill9: IIRC it was Paul Garlick. <lfam>PotentialUser-57: Firefox on iOS? <nckx>If it provides its own codecs it will. Otherwise, probably not. <lfam>I would try it. It's a big "maybe" <nckx>lfam: I am genuinely shocked. <lfam>So that video is VP9 / Opus. I don't think it will work in Safari or any of the native iOS players <lfam>I think we try to also have mp4 videos available <lfam>Basically, if it requires royalties, iOS supports it <lfam>If it's designed to contribute to the common good, iOS does not support it <lfam>Thankfully MP3 has / is entering the public domain <nckx>I'm going to stop reacting because whatever more I say will sound exaggerated & put on. <lfam>Yes kmicu, but youtube detects client capabilities to send the right thing <lfam>I think that iOS supports VP8 <nckx>Because it's mandated by WebRTC. <lfam>It sucks but we should re-encode in VP8 and another audio codec <nckx>(I'm just reading from the URL you just gave.) <PotentialUser-57>Well I am using text manual anyway. Just wanted to give a heads up because it doesn’t look good to have broken things on the main website <nckx>We could do some nginx magic to detect iOS. <nckx>PotentialUser-57: It's ‘not our fault’ but of course that's no excuse. Thank you. <nckx>lfam: You sound like you know this stuff? <kmicu>Just switch to AV1 and let’s all suffer. <lfam>PotentialUser-57: I'd appreciate if you could try in iOS Firefox <lfam>It's kinda my day job that I don't have anymore <lfam>Okay, thanks PotentialUser-57 <kmicu>lfam: thank you for confirming. TIL. <lfam>AAC is actually a great cross-platform audio codec but there may or may not be licensing issues <lfam>I keep a private "my-ffmpeg" package that adds libfdk support <PotentialUser-57>I’m sure there may be an app that can do it, but the point is easy on boarding for the general public, if Guix community desires that... <kmicu>lfam: iiuc licensing issues does not affect users (in this case a clip on Guix website). <lfam>kmicu: AV1 / Opus is soooooooo good. It's a shame that Apple doesn't support it <fnstudio>kmicu: sorry... when you said "set DPI for GTK from Guix", could you give me some other pointer? i tried to look that up but couldn't find anything <lfam>I agree kmicu, but it might get complicated socially <lfam>Basically we should set a good example in terms of licensing and be better than what would be required of a private user <lfam>PotentialUser-57: If you'd like to test another thing, you could try the iOS VLC app. I believe it should work and it would be a useful data point <lfam>We can't rely on users using that app but it would still be useful to get a "it worked" result <lfam>Apparently we can expect iOS support for AV1 / Opus in a year or two. Then everyone will be able to just use the same thing. I can't wait <kmicu>lfam: yes, thank you, considering trade‑off between users reach (and a11y) vs using only teaching materials created with libre tools is also important. <lfam>Not at all PotentialUser-38! It's important to report these issues, since many of us don't have iOS devices and won't notice the problem <kmicu>PotentialUser-38: you’ve stirred a good discussion. Nothing wrong about that. <lfam>kmicu: Right, but I think we can use h264 / aac in an mp4 container <lfam>We have freely-licensed implementations <kmicu>fnstudio: it looks like IceCat from Guix is based on GTK2. <nckx>lfam: Cool, I was going to suggest those. But you said something about potential issues earlier I think? <kmicu>fnstudio: in such case it’s probably a missing cursor theme but as pointed out by nckx I don’t see any of them in Guix 🤷 <nckx>kmicu: I guess all of Guixdom is happy with the default black one? I know I am. <nckx>PotentialUser-38: Did you mean to link to a shell script? <nckx>I mean, it is a tutorial… for computers. <fnstudio>kmicu: yeah, i'd also think it's a missing cursor theme as icecat otherwise looks good (eg font-size wise) <lfam>nckx: FFmpeg is GPL2+, and FSF says that the AAC encoder implementation's license is not GPL compatible. So, we can't distribute the FFmpeg / libfdk program <lfam>But, we can use it. It works and I use it all the time <nckx>I think MP3 is fine quality-wise, really. <lfam>Yes it's fine at high bitrates <nckx>As a fall-back sent to hostile devices I mean. <lfam>Just annoying when much better options exist now <lfam>Where are the videos kept? <kmicu>(We cannot distribute program but we can distribute data created by such program if they are libre licensed.) <nckx>Well, it's the device being (deliberately?) inferior, not us. <kmicu>fnstudio: could you install xcursor-theme and restart icecat? <nckx>kmicu: That's very true, but… we're a GNU project, hosted at .gnu.org… I can understand a reluctance to do it. <nckx>I'm personally fine with it if it's only sent to crippled devices to spread the good news of GNU. <kmicu>That’s true. Is reaching more users to share GNU message worthy? That’s not my call. (Personally I think that’s a good trade‑off). <lfam>I know that the greater GNU audio / video repository includes MP4 because of this issue <lfam>I don't recall where to find this but I've seen it before <kmicu>On the other hand using Twitter/Faceebok to post about Guix… in my opinion is not worth it. <kmicu>(As in having official Guix account on those platform and sharing news about the project.) <nckx>kmicu: Our videos are also deliberately shipped as ‘source’, with a Makefile & all. Putting a reference to libfdk in there would make our own video repo non-FSDG. Weird. <fnstudio>kmicu: nope, xcursor-themes didn't do it... <bavier[m]1>the guix-hpc folks have a twitter account. I don't know who is driving it. <nckx>So that complicates matters compared to a random Guixer encoding them once on their laptop. <nckx>Maybe it's nothing, IDK. <lfam>nckx: Where are the videos kept? <nckx>I thought artwork but I guess not. <lfam>Mark used to handle this IIRC <lfam>Everything is a Git repo <nckx>lfam: I thought Mark only uploaded them to GNU A-V. <lfam>Fascinating, but there must be another repo that contains the "non-slideshow" videos <nckx>The Twitter thing? Nope. <nckx>I really try hard not to Tweet. <kmicu>Making a bridge from fediverse and making twitter pots read only is even bigger gray area. <nckx>Unfortunately one cannot study contemporary politix and not have a Twitter account. <fnstudio>kmicu: also tried gnome-themes-standard and gnome-themes-extra but no luck with them either <lfam>Did somebody say they'd file a bug report about this issue? <kmicu>fnstudio: I would find out what cursor theme uses your distro by default and contribute it into Guix as a first patch xD <kmicu>nckx: for those I have rss feeds via niiter. That way I can still read Twitter but not directly load any Twitter resources. <lfam>nckx: Do you know how to do the nginx stuff? <fnstudio>kmicu: good point! and i'll seriously work on that... but i now started questioning whether i had really restarted icecat <kmicu>pkill icecat, sudo pkill icecat, reboot, sudo reboot, should do the trick xD <nckx>lfam: Sure. Assuming the Schemey syntax doesn't get in the way. <nckx>‘sudo pkill icecat’, for when you need to kill your ‘sudo icecat’. <lfam>Lol I can't believe they are suing nginx <lfam>No matter what, once you get a lot of money, you'll get sued <nckx>kmicu: That's the boring kind of conspiracy tho' ☹ <lfam>And if you don't have any money, you won't get sued. And that's how software and patent licensing really works *nckx will now be AFK in suspiciously innings-sized chunks. <lfam>It's gonna be a while before Linux has a truly secure boot <lfam>Only Apple has pulled it off so far and they design their own silicon <lfam>I love GRUB but "a serious bug has been found in the GRUB2 bootloader code which reads and parses its configuration (grub.cfg)" <lfam>There is a "grub.cfg" validator tool in the GRUB codebase and I wasn't able to make it fail no matter what I fed it <lfam>It's a small team and a legacy codebase. And if you mess up people can't boot. It's hard to make progress in that case <lfam>And similar for Linux itself <nckx>Seems to be staying up now. <nckx>NieDzejkob <yacc>: Yes, that's what I meant. <NieDzejkob>hmm, a glance at guix weather says tcc is not available as a substitute, but the build succeeds locally. Could I ask for some mild percussive maintenance on CI in this area? <nckx>tcc builds locally for me too. <NieDzejkob>hmm, the go-sctp logfile was truncated, so I checked locally and it built <nckx>I noticed that too, but it does actually fail tests on berlin. <nckx>I'm building everything again with -c1 (again)… <nckx>Now go-sctp is silent at ‘starting phase ‘check’’… <NieDzejkob>do you have, like, a sikret backdoor access to the logs that sidesteps the part that truncates them? <nckx>Now it built (I wish I'd been tracking which build nodes got the job each time but I wasn't) and is seemingly stuck retrieving 1 store item from '220.127.116.11'... <nckx>NieDzejkob: Er, the sekrit power of ‘zless’. The log wasn't truncated then 🤷 <nckx>So go-sctp finally succeeds and now is definitely stuck on the build node 😒 <nckx>because the clock just struck four and the big GC lock has been taken 🤦🤦🤦 <NieDzejkob>the tcc failure looks like it was caused by parallel tests <nckx>And yet: guix build -c1 /gnu/store/prcxypiarca3cah8h65bpdwg856zdbhj-tcc-0.9.27.drv <nckx>(Which doesn't mean the makefile honours that, of course, maybe a good place to start debugging.) <nckx>The whole clock striking four thing reminds me I should get some sleep. *NieDzejkob is in the same timezone <nckx>Unless things have changed a lot the GC lock is held for hours anyway. <nckx>NieDzejkob: I didn't save the previous logs (they are overwritten). I didn't expect to be locked out. <NieDzejkob>BTW, balsa seems to be another package that needs encouragement <nckx>(There's no DB entry for the latest /gnu/store/l7f2fl5b8wxcy8pb7bcy46nlkslw3mja-balsa-2.6.1.drv). <NieDzejkob>(that link's showing only failed builds for me right now...) <nckx>Unless IceCat has a sekrit bonus cache, I don't understand. <nckx>Now it's changed *again*. <nckx>Failed → dependency failed.
***terpri_ is now known as terpri
<nckx>TBH this is why I don't run Cuirass myself and avoid getting involved with the CI one. <nckx>It's not that I don't want to help but you get sucked into a recursive bug-hole before bed time and here we are. I need to go to work in 3 hours; I'm calling it a night. See you! <PotentialUser-74>Hi. I’m trying to install guix. At the very end of the process I get an error. <jorge[m]3>En la versión guix 2.0 es que viene con hurd? <NieDzejkob>PotentialUser-74: is that a manually-edited config, or one generated with the graphical installer? <NieDzejkob>PotentialUser-74: okay, could you post the config, then? You can pipe it to | nc termbin.com 9999 <NieDzejkob>(I think nc is available on the install image. If not, guix install netcat)
***tricon` is now known as triconium
<NieDzejkob>wait, so are you trying to install Guix System, or Guix the package manager on top of Debian? <NieDzejkob>do you have access to your configuration file from here? <NieDzejkob>or do you need to reboot into the installer again/ <NieDzejkob>okay, and you're saying that copy-paste is not working on your debian? <NieDzejkob>you can use a command like cat config.scm | nc termbin.com 9999 <NieDzejkob>oh, I think you copied the #+BEGIN_SRC scheme line too <NieDzejkob>that's just how the author delimits their codeblocks <NieDzejkob>yeah, on IRC you usually post links to pastebin sites for larger blocks of code <NieDzejkob>paste the full error, with line numbers and everything <NieDzejkob>what do you usually have after the @ in your bash prompt? <NieDzejkob>I think dots aren't allowed in a hostname. not sure. maybe it's the numbers at the start <NieDzejkob>why do you think that the IP in the hostname field is the problem? <NieDzejkob>did you read what you pasted? there's lots of garbage, there's only a random excerpt of the config, and no error message <NieDzejkob>it's supposed to end only after all the fields are specified <NieDzejkob>it seems that whatever editor you used inserted "smart" quotes <NieDzejkob>Unicode Character 'LEFT DOUBLE QUOTATION MARK' (U+201C) <PotentialUser-42>hint: Perhaps this `guix' command was not obtained with `guix pull'? Its version string <PotentialUser-42>hint: Perhaps this `guix' command was not obtained with `guix pull'? Its version string <NieDzejkob>how about /root/.config/guix/current/bin/guix describe <PotentialUser-42>hint: Perhaps this `guix' command was not obtained with `guix pull'? Its version string <NieDzejkob>so basically, 'guix pull' created a guix that thinks it doesn't come from guix pull <NieDzejkob>and somehow it's also much older, such that it doesn't have the openssh-sans-x variable <NieDzejkob>don't you feel that you should mention things like "I had to interrupt it"? <NieDzejkob>sorry, it's a bit late over here, I'm a bit cranky <NieDzejkob>yeah, in linux stuff, when you interrupt a command, it may sometimes have strange consequences <NieDzejkob>so you should always mention it if you ask for help <NieDzejkob>also, if you do interrupt it, think of it as failed <pkill9>where are all these PotentialUser-XX users coming from? <PotentialUser-42>I've never tried a fully free software. only other linux flavors. but I like SICP and Scheme so much!!
***rEnr3n1 is now known as rEnr3n
***rEnr3n0 is now known as rEnr3n
<kmicu>Hi folks, does Guix patch its BootHole? <roelj>I'm using "guix pack --profile-name=package-name-0.0.4 --format=docker ..." and it teels me "guix pack: error: package-name-0.0.4: unsupported profile name". What makes a profile "unsupported"? <nckx>roelj: --profile-name= must be either ‘current-guix’ or ‘guix-profile’, i.e. refer to an *existing* profile. Although the list is hard-coded, not actually based on directory existence. <nckx>alextee[m]: Do you have problems? <alextee[m]>nckx: i remember having some, reinstalling wine now <nckx>kmicu: Guix does not do ‘secure’ boot. Nobody's rushing to add it either. <bhartrihari>Hello, is there any un/official package for Tor Browser? Has anybody got it to work on Guix? <mroh>trying to fix current docker (go-sctp) breakage on master and realize that docker is a mess ;) <nckx>☝ they are free to use this slogan. <kmicu>If I’m using Guix System there’s no hole in my boot? Is that correct assesment? <nckx>Meh. Guix boots the way computers have always booted before ‘secure’ boot was introduced, we haven't jumped onto the ‘but now that's evil!’ bandwagon. Guix doesn't sign anything, so there's no signature check to be circumvented. <leoprikler>SecureBoot can't be broken if you boot "insecurely" 😉️ <nckx>There are no restrictions to circumvent. <nckx>I was trying very hard to avoid marketing propaganda, Leo 😛 <bhartrihari>Hello, is there any un/official package for Tor Browser? Has anybody got it to work on Guix? <nckx>It's okay to have an iris scanner at your door (I guess), it's OK to point out gaping hol^Wflaws in existing iris scanners, but suddenly calling houses without builder-mandated iris scanners ‘insecure’ is something I feel uneasy about. <leoprikler>bhartrihari: tor-browser is currently not yet there, but have a look at 42380 <leoprikler>Your iris scanner is only really secure if it uses Facebook's patented security solutions, that surely don't do anything evil 🙃️ <kmicu>“The infosec community will tell you that Secure Boot has been broken for 10 years, and yet nobody cared” so it looks like my boot is safe afterall. <nckx>And just happen to film the street as part of a drone network of your neighbours' doorb^Wscanners uploading everything to the cloud. <nckx>kmicu: Basically. The freedom zealotry community (proud member) will tell you that, in theory, secure/restricted boot is fine if users can easily add and remove keys. In practice, Linux distributions are forced to ship a MS-signed ‘shim’ to boot out of the box, and vendor firmware is buggy in just such a way (what an odd coincidence) that you can't remove their or MS's key. <leoprikler>kmicu: Well, not entirely "safe", but about as safe as any other boot *nckx takes off tinfoil hat, puts tasty burrito in it, eats tasty burrito. *nckx needs to go AFK, you're all lucky 😉 <alextee[m]>er how do you check the files of a package again? <rovanion>I have an issue where IceCat is unable to resolve domain names after reconnecting to WiFi, any ideas on how to debug or gather info for a bug report? And should that bug report go to IceCat or Guix? <janneke>alextee[m]: possibly do: find $(guix build <package>) <alextee[m]>mingw-w64-x86_64 doesn't have anything under "bin" there <raghavgururajan>Should I do (description "Everything has gone \\(really\\) foo bar.") <jlicht>rovanion: does `sudo herd invalidate nscd hosts' help? <janneke>alextee[m]: yes, try guix build --target=i686-w64-mingw32 hello <janneke>hmm, after a guix pull, i get "guix environment: error: Unbound variable: ~S" <alextee[m]>this package needs to run mingw GCC and then normal GCC <pkill9>no it's not necessary raghavgururajan <alextee[m]>it's creating bridges for windows binaries -> gnu/linux <alextee[m]>so first it uses mingw to build the windows stuff and then normal gcc to build the bridges <pkill9>only double-quotation marks need to be escaped, e.g. \" <janneke>alextee[m]: that's usually done using something like ./configure "CC=i686-w64-mingw32-gcc" "CC_FOR_BUILD=gcc" <janneke>yeah, but i'm also talking "conceptually" <janneke>many softwares do not support cross-building very well, if at all <alextee[m]>janneke: which package is i686-w64-mingw32-gcc in btw? i'm looking for that binary <alextee[m]>i thought the x86_64 version was in the mingw-w64-x86_64 package but it's not there <janneke>alextee[m]: we don't have that in a regular package (yet) <NieDzejkob>ah, I should've read the backlog entirely first -_- <janneke>hmm, guix pull seems OK now, sorry for the noise <Kimapr>kmscon doesn't work for me with DRM. Is there a way to disable it using kmscon-service-type (and kmscon-configuration)? <rovanion>NieDzejkob, jlicht: It seems to have worked permanently, even after suspending and resuming again. I will suspend the machine for a couple of days and see if it still works when I return. <rovanion>It could also be that the machine wasn't suspended long enough to trigger the bug. <rovanion>Which is why I'll try the longer suspend. <jlicht>rovanion: I just have the nscd cache invalidate snippet bound under a hotkey, as my guix is extremely enthusiastic in caching negative lookups <str1ngs>janneke: hello, just a FYI I'm gearing up for a new nomad release. The new release supports emacys windows and now uses gobject introspection almost exclusively. Also in regards to emacsy it has a better looking modeline and echo input area <janneke>str1ngs: very nice, looking forward to that! <str1ngs>janneke: regards to emacsy windows it only handles full windows. splitting is quite the problem to handle with webkit as you probably know. <str1ngs>I also added frame support but only in the context of nomad right now,. not emacsy I'm going to try and back port that eventually to emacsy <Jacob_>Hi. I'm having a difficulty installing my custom package. I've added (setenv "GUIX_PACKAGE_PATH" "/my-path/to-package-modules/:$GUIX_PACKAGE_PATH") to the top of my manifest file, but I'm getting an unknown package error. My question is: is exporting GUIX_PACKAGE_PATH variable in my .bash_profile the only way to get this work? <Jacob_>bavier[m]1: The one that stores a collection of packages I want to be "installed" to my local guix profile <str1ngs>Jacob_ normally I export GUIX_PACKAGE_PATH in ~/.bashrc or ~/.bash_profile <apteryx>hello! Is there a way to configure VLANs at the operating system definition level? <bavier[m]1>Jacob_: the manifest is loaded at a point where, I think, setting something like GUIX_PACKAGE_PATH cannot affect the packages known to guix <pkill9>str1ngs: does nomad do window management? Or is it still just a browser? <str1ngs>pkill9: it does handle window management. though it's not complete ie. it cant split windows. you can though have multiple emacs like frames open with different buffers activated. <str1ngs>pkill9: splitting webkitgtk controls is not easy. only one instance is visible at a time. <str1ngs>pkill9: also nomad supports more buffer then just <web-buffer> it can handle <text-buffer> and also <terminal> <hendursaga>I can replace git with git-minimal and be fine? Since Guix depends on so many Git-related things. Do I do a 'guix install git-minimal' and then 'guix remove git' ? <bavier[m]1>hendursaga: you can also do a single transaction with 'guix package -i git-minimal -r git' <str1ngs>pkill9: probably eventually I'll improve the window management support so that you can split windows. just you won't be able to view the same buffer twice. if that makes sense. that's the easiest solution for now. <str1ngs>no web browser that I now of can split web views. :P <pkill9>can it open any kind of windows? <str1ngs>though in nomad you can use ibuffer to list buffers and switch think single window model <str1ngs>pkill9: nomad uses emacs terms. so a window is not a application window but a split. and application windows are called frames. nomad does handle bot emacs like windows and frames. <str1ngs>janneke: though i have though of using a forest of webkit controls to handle splitting. but that really makes thing extremely complicated. <c4droid>Hi, I just installed dwm and st on my guix system, someone know where stored dwm and st's config.h and config.def.h? <pkill9>c4droid: you'll need the source, run `guix build -S st` to get it <pkill9>c4droid: you'll need to write a package definition that patches those files with configuration as you want it <c4droid>pkill9: Thanks. I'll going to download the source <Jacob_>bavier[m]1: Thanks for help. I'm going back to the drawing board then <NieDzejkob>I'm packaging zxtune, it has a multitude of audio output backends. Is a specific one (like alsa or pulseaudio) preferred, or should I enable all of them? <guixy>NieDzejkob, If you enable all of them and run guix size, what is its total output? <NieDzejkob>guixy: guix size pulseaudio alsa-lib -> 180MB, guix size alsa-lib -> 73 MB <mbakke>apteryx: should we ignore the inkscape update for now due to the aarch64 issues? *mbakke still mourns the OSS deprecation in Linux <mbakke>how different the world would be if OSS had not (temporarily) gone proprietary <guixy>NieDzejkob, I think it's worth enabling all of them if possible. <NieDzejkob>I'm reading up on audio history on Linux, did OSS support mixing (i.e. multiple programs outputting audio at once?) <mbakke>NieDzejkob: I think so (at least on FreeBSD according to Wikipedia). <mbakke>if it didn't back then, it definitely would now without such hacks as PulseAudio :-) <nckx>God damn you all to heck. You tricked me into reading audiophile. <nckx>‘In my opinion that antialiasing makes sound be feminine while OSS corresponds to the northern and viking like school with crisp highs. Thus for me alsa = jewish school of audio engineering, oss = indoeuropean school. By the way in jewish culture mother dominates over father. Interesing coincidence.‘ <nckx>I demand some form of compensation. <nckx>Isn't PA the default for new Guix System users? *mbakke has high hopes for Pipewire <mbakke>nckx: by the way, I cheated a bit and picked your git update to master :P <nckx>I'm sure you have your sinister indoeuropean reasons. <nckx>(Did something else trigger a basically-rebuild?) <mbakke>it's probably futile, but I don't want git to become a 'staging' package; currently we have a ~60 rebuilds buffer... <apteryx>mbakke: I guess so :-/. Although it seems lib2geom has a new maintainer which is showing some activity, so hopefully they will get interested in fixing that issue. <apteryx>mbakke: I'm curious though, if Inkscape built with its bundled lib2geom dependency could actually run on aarch64; if it does perhaps we're doing something wrong with our lib2geom package. <NieDzejkob>huh, git triggers more rebuilds than git-minimal. Interesting, it's usually the other way around <NieDzejkob>my guix refresh says 330 rebuilds, is the threshold at 400 now? <nckx>NieDzejkob: git-minimal inherits from git, which is less common. Staging is still 300 AFAIK. <nckx>Ah, I'd missed mbakke's response. I don't get the ‘become’/‘buffer’ part either. <apteryx>mbakke: thanks for the tip about VLANs, I had the iproute idea but my Guix services skill is always on the rusty side so the blog link will help greatly. <mbakke>nckx, NieDzejkob: on current 'master', git (and git-minimal) has 240 dependent packages :-) <mbakke>where those 90 extra rebuilds went is left as an exercise for the reader... <mbakke>NieDzejkob: ./pre-inst-env guix refresh -l git git-minimal <mbakke>Building the following 105 packages would ensure 240 dependent packages are rebuilt: ... <nckx>But then it wasn't cheating. <nckx>Please, cheat more if this is how you do it. <NieDzejkob>I'm surprised there were useless inputs like these <mbakke>maybe they weren't at some point, I notice some other distros also try to avoid needlessly adding git to the build dependencies <nckx>Some packagers add inputs that the build system mutters about but don't actually affect the build. <nckx>mbakke: How did you detect these? <nckx>PotentialUser-31: Yes. Hullo! <kmicu>(Handy for folks saying Guix is like Docker.) <PotentialUser-31>I am a current/former archlinux/pacman user, but today morning I installed Guix <kmicu>You are not alone PotentialUser-31 there’s plenty of those here (includingme). <PotentialUser-31>I've heard a lot about Nix and I love scheme/lisp so I wanted to try it out *kmicu still remembers constantly fixing xmobar with custom habs setup after a periodical pacman -syolo. <kmicu>Guix infra is quick as Flash. <kmicu>Build farms idle cuz they are so fast. *kmicu paste [gotta go fast lizard meme]
***apteryx_ is now known as apteryx
<kmicu>PotentialUser-31: 30 minutes for guix pull…? That could be a network issue.
***apteryx is now known as Guest50856
***Guest71056 is now known as apteryx
***apteryx is now known as Guest31086
***Guest50856 is now known as apteryx
***Guest31086 is now known as apteryx
<kmicu>pacman will be always faster because it’s mutable, less work to do when we can overwrite (aka break) a working setup. 😺'
***Guest76184 is now known as apteryx
***apteryx is now known as Guest63051
***Guest91398 is now known as apteryx
<PotentialUser-31>maybe its because my system is new and requires several dependencies, but still I was getting <1mBps while on pacman I regularly get 20-30 mBps
***Guest63051 is now known as apteryx
***apteryx is now known as Guest78835
***Guest71714 is now known as apteryx
***apteryx is now known as Guest80569
***Guest78835 is now known as apteryx
<nckx>It's very unlikely that GNU systems were to blame, unless you spent minutes actually downloading from the Savannah git repository. That's the only part of ‘guix pull’ that they handle. All the rest is Guix hardware (downloading substitutes) and your hardware (building the new guix, which can take quite a while).
***Guest80569 is now known as apteryx
***Guest82480 is now known as apteryx
***apteryx is now known as Guest32675
***Guest89814 is now known as apteryx
<kmicu>It’s possible you live next door to a Arch CDN.
***Guest32675 is now known as apteryx
***apteryx is now known as Guest60729
***Guest86117 is now known as apteryx
***Guest60729 is now known as apteryx
***Guest24260 is now known as apteryx
***apteryx is now known as Guest59980
***Guest18934 is now known as apteryx
***ChanServ sets mode: +o nckx
***apteryx is now known as Guest90833
***Guest59980 is now known as apteryx
***apteryx was kicked by nckx (apteryx)
***Guest90833 is now known as apteryx
<mbakke>nckx: I painstakingly went through each package with a git input, ripped it out, and checked what happened :/
***apteryx is now known as Guest90921
***Guest96147 is now known as apteryx
<mbakke>the OPAM test suite detour made it worth it though <nckx>ChanServ why you no akick. <rndd>hi everyone! trying to pull guix from local directory. where i can take openpgp key for introduction?
***Guest90921 is now known as apteryx
***Guest55612 is now known as apteryx
***nckx sets mode: +b apteryx!*@*
***apteryx was kicked by nckx (apteryx)
***Guest33117 is now known as apteryx
***nckx sets mode: +b Guest*!*@*
***apteryx was kicked by nckx (apteryx)
<nckx>PotentialUser-31: Sorry about that. It's a regular user (maintainer even) whose client/bouncer has gone mad. <NieDzejkob>rndd: you can have a channel without an introduction, if that's what you're asking <mbakke>rndd: you can use (channel (inherit (find guix-channel? %default-channels) ...) in your channels.scm to inherit the canonical introduction <mbakke>eh, missed a paren after %default-channels <rndd>mbakke: got error "extraneous field initializers (inherit)" <nefix>is there any reason why Telegram Desktop is not packaged? (licenses or something like that) <nefix>NieDzejkob: nice! (and not nice :( ) <nckx>If you need a client and need one now, we have emacs-telega. I'm sure it's a vastly different user experience but it works. <nefix>I'll try porting the Nix declaration, let's see if I succeed <bavier[m]1>there might be a few dependencies that need to be packaged to <nefix>Both telegram and vscodium are the two packages that make me not being able to fully switch to Guix
***PotentialUser-31 is now known as jotaru
<bavier[m]1>agreed, give it a go, and we can always help if you need <nckx>I've never heard of viscodium. <nckx>*vs, and aha: ‘Visual Studio Code is a source-code editor developed by Microsoft for Windows, Linux and macOS.’ <nefix>vscodium is vscode without the microsoft stuff <rndd>NieDzejkob: how i can do it with no auth? <Formbi>much nicer than the electron shits such as discord <nckx>That vscodium isn't packaged is no surprise: it manages to require both electron (=Chrome) and at least some NPM stuff (=hardish to package). <nefix>I failed miserably like half year ago (and even with Spacemacs) <Formbi>also you can install some things thru flatpak <jotaru>and aren't there free versions of chrome? <Formbi>but kinda shaky technology-wise, I imagine <bavier[m]1>free, but I'd imagine we'd want something based on our ungoogled-chromium instead <jotaru>yeah thats what I had in mind ::p <nckx>Are there free eversions of Electron? <jotaru>does guix package free software that isn't gpl? <Formbi>>Electron avoids most Google-related parts of Chromium by virtue of not using most of the //chrome directory. *nckx .oO but it's not in Debian so really <jotaru>what sort of network speeds do you all normally get with the savannah mirror? <mbakke>rnddcan you paste the channels.scm? the example I pasted works for me. <jotaru>alright in that case something is wrong with my network :p <jotaru>because for me its very noticable <nckx>Of course now it's slowish here but I'm on shite wi-fi, I'll try elsewhere. <nefix>jotaru: for me it's not the network but the disk speed (I'm running Guix in a USB stick :P) <jotaru>I get 100% packet loss when I ping gnu.org for some reason <mbakke>jotaru: are you in a VM? ping does not work through Qemu (by default) <nckx>jotaru: Are you in a Qemu VM? <nckx>Or a firewalled network? <jotaru>could be some firewall problems, but idk why one computer would work while the other doesn't on the same network <jotaru>I just pinged for 2 minuted to gnu.org 94% packet loss <nckx>I get several MiB/s pulling from Savannah. <mbakke>rndd: ah, the (inherit ...) statement must be the very first entry <jotaru>On my mac I get only 3% packet loss <rndd>mbakke: well, now i have "guix pull: error: Git error: cannot locate remote-tracking branch 'origin/keyring'" <jotaru>I must have offended Dr Richard Stallman somehow <jotaru>its like my computer has been personally blocked from achieving a proper ping >;( <mbakke>rndd: yes, you need to checkout the keyring branch in your guix clone :/ <nckx>jotaru: Is that only for gnu.org (and friends)? <nckx>E.g., try guix.gnu.org, which is on a different continent. <nckx>I fail to lose even a single packet ☹ rms must love me too much. <jotaru>guix.gnu.org is 0% packet retention <nckx>Nah, DNS is a one time IP look-up. <rndd>mbakke: what is the ideo of keyring branch? <nckx>jotaru: mtr is in Guix, might give you a hint. <nckx>jotaru: guix.gnu.org was my bad. I forget (not for the first time) that it blocks ICMP completely. <nckx>Ping is not a great diagnostic tool, part n. <rndd>mbakke: anyway, thank you <nckx>s/a/the/ to prevent flaming. 🙂 <jotaru>yeah so I just did a guix pull and its giving me about 22Kib download speed <nckx>jotaru: For what, exactly? <nckx>Pasting the entire terminal output to paste.debian.net is fine. <nckx>I mean what's it downloading. <nckx>OK. Those are downloaded from guix.gnu.org, not gnu.org. <jotaru>(so rms isn't hating on me then :D) <nckx>We don't just block your pings, we pinch your pipe. <jotaru>is there anything I can do about it? <nckx>I thought this was fixed long ago. <nckx>jotaru: Long story short the server had a bad NIC, so everything was moved over to a new machine with a better one. <jotaru>is there only 1 server containing all the guix expressions <nckx>jotaru: Could you paste a URL that was slow? Ideally a nice big file. <mbakke>rndd: I'm not sure why they keyring branch is separate, I suppose it adds a little flexibility/separation of concerns <nckx>ci.guix.gnu.org builds them all and serves them as binary substitutes. <jotaru>im not used to a source based/binary based distro so im still getting used to the terminology and stuff <nckx>(For background info search those two last words in the manual.) <jotaru>I def have some manual reading to do <jotaru>do you want me to just pastebin the output of guix pull? <nckx>paste.debian.net or another Tor-friendly pastebin, please, so not pastebin.com. *nckx → AFK for dinner now. <jotaru>also, in case it matters, I am not yet running the full GuixSD distro, I am using the Guix package manager on top of pacman. If that is not supported here, sorry for the troubles... <jotaru>I wanted to experience the package management powers first before moving into fully declarative systems configuration, etc. etc. <mbakke>jotaru: guix supports any reasonably modern GNU/Linux distro :-) <mbakke>jotaru: some of those speeds are pretty good, and others are super slow, weird <lfam>Could somebody using an Intel Skylake platform try using our Blender package? For me it crashes when trying to add a video clip <nckx>to see if it's deterministically slow. <lfam>It's slower than 100 kbps for me in eastern US <PotentialUser-84>hello! I'm new to IRC in general, so please let me know if I'm breaking some kind of conduct. If anyone has sometime, I have a question about system installation(3), specifically about UEFI mounting <jotaru>when I manually download about 260 Kib speed <jotaru>oh wait nvm average speed is 160 Kib <lfam>PotentialUser-84: Please ask your question :) If anybody knows the answer they will help you <PotentialUser-84>thanks! So in the documentation (18.104.22.168), it has the user set up file systems to prepare for mount. When it's discussing the efi mount point, it says: "if you have opted for /boot/efi as an EFI mount point, mount it at /mnt/boot/efi" ... my question is, should the command then be: "mount /dev/sda1 /boot/efi" or "mount /dev/sda1 /mnt/boot/efi" ... <kmicu>Hi PotentialUser-84: during the installation process we use /mnt/boot/efi and after installation that will turn into /boot/efi. <PotentialUser-84>gotcha, and it is normal to have to mkdir both /mnt/boot and /mnt/boot/efi right? <kmicu>Iirc graphical installer should do the right thing so you could use that for an effortless installation. <PotentialUser-84>unfortunately it fails when attempting to partition in graphical, no idea why, doesn't throw any codes, just reboots from the <kmicu>If you find time you could share your partitioning scheme. Maybe there’s a workaround. <PotentialUser-84>will do, I'm going to try the manual installer one more time with the second option first just to make sure something else was wrong. I tried that option yesterday, and was given a successful install message from the guix init, but when I rebooted, it just cycled the intel boot manager <nckx>jotaru: Ouch. Said server's in Germany (as is guix.gnu.org), I'm not much further, but 28M to 160K is a big drop just to cross one little ocean… <nefix>how am I supposed to clone a git repo with submodules? I tried to add (recursive? #t) but it doesn't seem to do anything :/ <nckx>nefix: Did you change the hash? <nefix>nckx: nope, but it doesn't complain <nckx>It won't, since the hash is the primary identifier and it has something that matches. <nckx>Change a character (not the first one) & try again. <nefix>yep, that worked. Does something like Nix's lib.fakeSha256 exist? It's basically a string with zeroes with the correct length that is really useful <nckx>nefix: I just use emacs' Esc-52 0 to create 52 zeroes. <nckx>I'm not aware of anything like that. <mfg>Does usb redirection in QEMU need my user to have a specific group membership? i always get permission denied when i try to redirect a usb device ... <nefix>I could create something like that for vim... but it would be really useful! <nckx>mfg: apteryx (Maxim Cournoyer) had a suspiciously similar problem with Gnome Boxes! Unfortunately, they're currently banned.
***nckx sets mode: -b apteryx!*@*
***nckx sets mode: -b Guest*!*@*
***nckx sets mode: -b Guest!*@*
***ChanServ sets mode: -o nckx
<mfg>kmicu: thanks for this :) <vivasvat>im the user from before who was having network connectivity issues from guix.gnu.org *kmicu wonders wheter 8119b7c6f7 gnu: rust-1.45: Update to 1.45.1. triggers an IceCat rebuild. <nckx>mfg: Thanks for debugging. Using a udev rule to chgrp it would definitely work. I don't know if its teh best practicez. I'll look around. <Formbi>pkill9: initially ECMAScript was supposed to be a Scheme :< <nckx>Most depressing anecdote. <Formbi>but ECMAScript isn't even very similar to Java *kmicu checks if a gopher website about Guix exists. <Formbi>msavoritias[m]: I think websites should have general interfaces <nckx>That's the point. Marketing said: Java big $, nerd make thing look like Java! make $. <Formbi>and different things could connect with said interfaces <NieDzejkob>kmicu: it doesn't, because the rust variable points to rust-1.39 <Formbi>browser, Emacs, some ncurses thingy etc. <NieDzejkob>what a weird times those were. Java hippier than Scheme... <kmicu>[Jokin’] There are big cybermoney in Guix so I started new project: GuixScript. <msavoritias[m]><Formbi "browser, Emacs, some ncurses thi"> and all of it written in some scheme language <msavoritias[m]>i would definetily like it. sites are too tied to specific browsers at the moment <nckx>kmicu: I should say more than that :) The default rust is 1.39, so update others at will. <kmicu>Thank you NieDzejkob nckx somehow I always forget about defaults. <vivasvat>Sorry, I had to go do something for awhile, does anyone know the reason that I may getting 160 Kib (vs ;;normal for others;; of 30 Mib) download speed when doing guix pull? <nckx>lfam: Has it always been this slow for you? <vivasvat>but I got it only for the past 24 hours ish <nckx>vivasvat: Did you see my message earlier, about the 28M server being in Germany (=500km from ci.guix)? I wouldn't expect that outside of Europe. <nckx>lf_am has been using Guix for years. <nckx>I used to have a shell account on the East Coast but it moved to… Germany >_< <GNUtoo>hi, I've managed to build guix by using ./bootstrap && ./configure && make -j3 <GNUtoo>However ./pre-inst-env guix build <some new package> gives me that: <GNUtoo>guix build: error: failed to connect to `/usr/local/var/guix/daemon-socket/socket': No such file or directory <MtotheM>how do you deal with the missmatch between manually installed programs and your config file? <nckx>GNUtoo: You need to start the daemon (either in the background using & or in a different terminal). <nckx>See Running Guix Before It Is Installed in the manual. <GNUtoo>that'd be easier than with 04-packaging-part-two.webm <nckx>GNUtoo: Please let us know if anything is missing from or inaccurate in the videos! <GNUtoo>ok, I'll re-listen to them and report here <nckx>We try to keep the up to date but it's easy to forget (unlike the manual). <nckx>MtotheM: Do you mean programmes installed using ‘guix install’ and those in provided by your Guix system configuration? <MtotheM>Yes, let's say I want to re-deploy my setup on another system. and a bunch of things aren't included in my config as they have been installed manually. nckx <NieDzejkob>GNUtoo: you want to pass --localstatedir=/var and use your system's daemon <nckx>MtotheM: I recommend creating a manifest file and running ‘guix package -m <the file>’. To add/remove a user programme, edit the manifest and run that command again. *GNUtoo re-compiles with ./configure --localstatedir=/var now *GNUtoo doesn't understand well the consequences of not using --localstatedir=/var though <nckx>GNUtoo: Did you get ‘./configure’ from anywhere? A tutorial or documentation? <GNUtoo>does it use /var/gnu instead of /gnu <nckx>GNUtoo: No, /var/guix instead of /usr/local/var/guix. <nckx>/gnu would be /gnu either way. <GNUtoo>It did mention --localstatedir=/var in a video I was looking at <GNUtoo>but I skipped that as I assumed that the default was fine <nckx>But Guix tracks /gnu/store's state in a database in /var/guix. <nckx>You like to live dangerously. *GNUtoo doesn't really understand why I've to tell the guix cloned repository to use my system's guix stuff <GNUtoo>intuitively these should be separate <nckx>I have the equally dangerous habit of reading ‘by using ./bootstrap && ./configure && make -j3’ as ‘I did roughly this, y'know, 3 steps’. <GNUtoo>like ./configure typycally picks /usr/local instead f /usr <nckx>So thanks again to NieDzejkob for *checks notes* ‘being able to read’. <lfam>nckx: I use a private mirror so I don't really notice. I'm sure I've seen it be faster on occasion though <nckx>GNUtoo: Sigh. I agree. It's because these are standard default locations: the default --localstatedir for all GNU projects is /usr/local/var (=$prefix/var). <nckx>So that's the reasoning, and it's been discussed before, but it's a risky default. <nckx>‘./configure typycally picks /usr/local instead f /usr’ because that's also standard. <GNUtoo>"and it's been discussed before" -> Is there some documentation about that? <nckx>If tomorrow some ./configure uses /CoolApps/Bob/Documents and Settings as default instead of /usr you'd be annoyed, the same argument applies here, even if the default is bad. <GNUtoo>So that I don't bother people on IRC and read what has already been discussed instead <GNUtoo>what I don't understand is that I want to send a patch, so I clone guix and build it in its own directory <GNUtoo>so I assume it's not supposed to use any of the host sutff when using ./pre-env <GNUtoo>and I assume it would use different defaults <GNUtoo>so as I don't understand a thing here, I probably miss some concepts or understanding of how Guix work <GNUtoo>So I just leanred that there is a database somehwere (which is separate from /gnu) <GNUtoo>and I know there is a daemon that can spawn builder process <GNUtoo>And I mostly read the official documentation here and there, and part of the cookbook + saw the videos <nckx>If you're already using Guix, just configure with the right option and ./pre-inst-env guix will use your already running daemon. <GNUtoo>ok, so the idea is to tell the source guix to use the host daemon, I see <GNUtoo>So now I assume that all the scm of the source are used thanks to pre-env but that the host daemon and /gnu and all that stuff is still used <nckx>GNUtoo: Even if you used ./pre-inst-env guix-daemon it would still use /gnu and you'd still want --localstatedir=/var. *GNUtoo also looked at ./pre-inst-env <nckx>lfam: Ever consider sharing?