IRC channel logs


back to list of logs

<jab>unmatched-paren: thanks for the tip!
<Andronikos>Using Guix on Gentoo. I have made a nerd-fonts package and installed it on my system. slstatus does not render my icons although in Emacs they are displayed fine. If I install the fonts in /usr/share/fonts slstatus display them fine. Does someone know why?
<sneek>Andronikos, you have 1 message!
<sneek>Andronikos, nckhexen says: I do.
<Andronikos>nckhexen: So you have migrations turned off?
<Andronikos>Also using guix gc the value of freed was always the same as the latest value in the brackets. Now it is not and it has weird precision (
<gabber>unmatched-paren: what do you think about this ? i currently use it as a home-shepherd-service
<nckhexen>Andronikos: Yes.
*nckhexen → 😴💤 'nait gweks
<Andronikos>nckhexen: Thanks. Guess I will try it as well.
<lechner>mroh: thanks for the pointer! how about docker. also, someone told me about ganeti
<mroh>lechner: docker works (very) well.
<the_tubular> podman seems borked since the last commit that changed the version
<the_tubular>I gotta fill a bug request but I got too much to do :/
<the_tubular>The network is borked
<the_tubular>Can't pass ports to your host
<jab>cwebber: just said in his live chat that he hopes spritely would be useful in guix...
<jab>I wonder how guix would use that code...
<lechner>mroh the_tubular: thanks!
<lechner>Hi, is anybody using a Pinyin input method like fcitx, fcitx5, or ibus?
<lechner>right now i need it for Gnome under Wayland, but eventually for other WMs, too
<lechner>actually, the graphical session is Gnome in X
<jts>anyone else having issues launching nautilus?
<jts>oh wait just found this... why are dates inverted? anyway,
<cwebber>jab: she/her btw
<jab>cwebber: yeah I know. dang on it. Sorry.
<lechner>cwebber: thanks for being nice about it!
<cwebber>lechner: mistakes happen, I know jab didn't mean to, so it's fully forgiven, though I still wanted to be explicitly clear for those who didn't know... misgendering is painful. I think I know what you mean, that you appreciate that I was being polite, though I have mixed feelings about being thanked for it kinda. but that's kinda complicated and this isn't necessarily a great forum to discuss that in
<cwebber>back to talking about cool guix things ;p
<cwebber>I did talk about spritely and guix working together, I do have lots of hopes and plans there. have been kind of being quieter about it until the guile-goblins with captp release is out
<jab>cwebber: speaking of cool guix things, I must have missed the bits of your talk where you talked about guix and spritely integration...I was occasionally pulling out my earbud and doing work, throught your talk btw. also thanks for being gentle
<xd1le>cwebber, <3
<Cairn>When marking verbatim in a (description), should I use `this syntax'?
<Cairn>Is there a different way?
<dirtcastle>I'm big noob but hear me out. A file manager list the files. It somehow knows where the beginning of the memory of the file is and where the end is. Write an app that will shred everything that is not a file that is viewable by file manager. So all hidden malicious files will be destroyed.
<dirtcastle>Similarly shred everything that can't be listed in /usr/bin. See the paths that are in the /usr/bin and "secure-shred" (I just named it rn) that too. I think this should be enough to clean all the malicious software? So u don't have to erase Everything and reinstall? Doing this on every boot will be beneficial especially if there is spyware.
<dirtcastle>your thoughts . I'm no cs grad so my idea will be full of flaws.
<HP-UX>Hello. I am fine with binary packags, but I need only one package such as plasma-framework, to be patched with my patch, so it can work with Xmonad. I'm new to nixos/guix etc. and I'm trying to figure out how to custom compile certain packages
<HP-UX>Basically I need to get the source package from guix/nixos etc. and then patch the source, then recompile and install. Please advise
<brendyn>HP-UX, there are a variety of ways to modify guix packages. depends what you are doing as to which way is most convenient
<HP-UX> patch at bottom, would need to redo the patch for the nix version
<HP-UX>Without this patch, kde panels in Xmonad fail to open
<HP-UX>Whatever would be the best way. I'm collecting information right now, before beginning my installation and preperation
<brendyn>HP-UX, are you trying to run KDE? We don't have KDE working in plasma yet, although we are working on it slowly
<brendyn>working in guix*
<HP-UX>Oh. Darn.
<HP-UX>I use plasma kde panels with Xmonad.
<HP-UX>I could use a diff panel I guess. I used to use kicker but then they upgraded to a new design. I still think kde 3.5.10 was the best kde
<brendyn>currently you can do GNOME and XFCE and i think MATE for major desktops
<HP-UX>How is nixos different from guix?
<brendyn>HP-UX, one cool thing about Guix is you can generate virtual machines of your system to test out.
<brendyn>you just need to install Guix on your current distro
<brendyn>then you just run `guix system vm my-config.scm`
<HP-UX>What does that do?
<HP-UX>I have ubuntu 22.04 right now and I'm unhappy with it in many ways
<brendyn>generates a script that launches a virtual machine with your exact system configuration
<HP-UX>which system?
<HP-UX>my current running live ubuntu 22.04 system?
<brendyn>your one that you will create if you intent to install Guix System
<HP-UX>I need to learn more before
<HP-UX>I still don't really understand how all this fits in together
<HP-UX> watching this video on nixos
<brendyn>Ok so you know how when you install and configure Ubuntu, Arch, or most other distros, you do so by following a sequence of actions; install this, that, configure that, etc... With Guix the process is fundamentally diferrent. You create a .scm file that describes your entire system, the services and desktop you intend to run. then you run a command `guix system vm config.scm` to /instantiate/ or upgrade your entire system in on one
<brendyn>step. Guix therefore also allows you to generate a virtual machine using that same configuration that you can use for testing.
<HP-UX>Where do you run this command?
<brendyn>In any terminal, provided you have Guix Installed
<HP-UX>You boot into something minimal and then do it what?
<HP-UX>I don't get it!
<HP-UX>What do you mean install guix on my already linux distro?
<HP-UX>I'm confused by this, please.
<brendyn>Ok. So Guix as a project does includes a lot of things, so its a bit confusing
<HP-UX>I am reading this
<HP-UX>so I boot a live linux cd, then I get guix, then prapre my hdd, isntall guix there, then reboot?
<HP-UX>Something like this?
<HP-UX>I'm interested in buildilng the whoel OS from guix and not having to intall it over on to of my ubuntu which is confusing I dont' get that part.
<brendyn>For one, Guix is a "package manager" rogram that you install on other distributions like Ubuntu. However, it also includes its own complete distribution that you can install on your computer instead of Ubuntu
<HP-UX>Ah I see.
<HP-UX>Is there a base OS distributed by guix on top of which I can specify my application software?
<HP-UX>And where do I get this base os image
<brendyn>Not 100% sure what you mean
<HP-UX>How do you install guix from without a previous OS.
<brendyn>the Binary Installation page you liked to describes how to install Guix ontop of Ubuntu
<HP-UX>Is there a installer?
<HP-UX>I am interested in installing guix from scrtach if that makes sense.
<brendyn>on the left here.
<HP-UX>Wipe ubuntu clean, start fresh, install guix
<HP-UX>USB/DVD ISO installer of the standalone Guix System.
<HP-UX>Thank you
<HP-UX>Does guix support btrfs (it's in the linux kernel)
<brendyn>Two things you need to know when installing Guix. 1) Guix by default only installs Free Software, and so many wireless cards and recent graphics cards will not function well or at all, and 2) It is not as complete as mainstream distributions like Ubuntu, so you will find usability bugs and need to get your hands dirty to make some things work
<HP-UX>I don't need prop drivers
<brendyn>yes it supports btrfs. I think you can install if for the root filesystem too
<HP-UX>So long as you can run my 5800X CPU
<brendyn>thats the same cpu i have
<HP-UX>Phantom Gaming 4 motherboard.
<brendyn>it has no integrated graphics though. what graphics card are you using then?
<HP-UX>I have nvidia gtx, 275, needs binary driver nvidia 340, which nvidia abandoned, but nouveau works which I use.
<HP-UX>lshw output:
<HP-UX> lspci output
<brendyn>my lshw doesnt even tell me my graphics card name
<HP-UX>That hw should be okay in linux-libre?
<brendyn>probably. only way to find out is to test it.
<HP-UX>Thank you
<brendyn>are you keen to learn how to use Guix?
<HP-UX>I'm hungry right now.
<brendyn>... like, for food?
<HP-UX>I have food dancing in my vision!
<HP-UX>I will be able to think more later.
<brendyn>cant say ive halucinated food before
<HP-UX>I mean figuratively ha
<sarg>hi there, fastboot build is failing on my machine because I have disabled ipv6 using kernel parameter (ipv6.disable=1). Is it considered a bug?
<sarg>namely they're using this code in tests:
<sarg>which fails, because they're trying to create ipv6 socket specifically
<brendyn>sarg, i think it is a bug yes
<brendyn>networking functionality in general should not be needed to build packages
<brendyn>sarg, can you please email about it?
<itd_>Hi. 'guix pull' fails for me with '[...]invalid package license~%\\n\" arguments: (#<package expat[...]' and I'm struggling to understand what it's unhappy with. Full error: Any ideas? Thanks!
<brendyn>itd_, it relates to some new license validity checking that ludovic added recently
<unwox>hi guix!
<nashdidan[m]>Hi guys. When packaging an email client such as `himalaya` which is developed in... (full message at <>)
<unmatched-paren>nashdidan[m]: Put the executable in either mail.scm or rust-apps.scm, yes.
<unmatched-paren>And the crates in the appropriate crates-*.scm.
<nashdidan[m]>Ok, didn't see rust-apps.scm, but I guess mail.scm is probably better? Also, there are MANY deps, lots of them not in the store yet, obviously many are mail/imap/notmuch libraries, tls etc...
<unmatched-paren>Unfortunately, that's often the case with Rust.
<unmatched-paren>The dependency trees are *h e l l*.
<nashdidan[m]>I see... one more question, does it make sense to submit the current in-development version, or wait for the official first stable release? The developer advaises to not use in production before v1.0.
<unmatched-paren>1.0 doesn't exist yet, right?
<nashdidan[m]>No. I've packed v0.6, which works for me fine with the imap backend, but not extensively tested.
<unmatched-paren>Cool, just use that tag then.
<nashdidan[m]>Thanks for your help help!
<itd_>brendyn: Makes sense, thanks. I'm probably missing something, but: the error doesn't mention non-Guix files; Guix's CI is happy (i.e., Guix itself is likely working). I do use non-Guix channels, but I'm unsure how to determine the offending code from this error.
<brendyn>itd_, it tells you #<package expat@2.4.1 gnu/packages/xml.scm:120
<brendyn>doesnt make much sense though since i dont see whats wrong with it
*itd_ nods
***wielaard is now known as mjw
<mroh>itd_: I guess, in one of your channels is code that has (license #f) or expat w/ xml module or so. perhapse try w/o channels and then one by one...
<unmatched-paren>mroh: but gnu/packages/xml.scm is a file in guix itself
<brendyn>itd_, oh i think i get it
<brendyn>itd_, so basically there is a license called expat, but also a package
<brendyn>itd_, so in this case they need to make it pick the right one
<unmatched-paren>A (guix licenses) import is missing #:prefix, I guess.
<unmatched-paren>itd_: Somewhere, #:use-module (guix licenses) needs to be changed to #:use-module ((guix licenses) #:prefix license:)
<unmatched-paren>And then (license FOO) in that file changed to (license license:FOO)
<unmatched-paren>hello zimoun!
<efraim>ugh locked out of my debian VPS, blocked my IP address by mistake. Really need to replace that with a Guix VM
<efraim>back now finally
***Dynom_ is now known as Guest6166
<sektor[m]>Hrmm. Is there a pre-start for shepherd things that require kernel modules?
<unmatched-paren>sektor[m]: do you mean "can you make a shepherd service depend on a kernel module"?
<unmatched-paren>sektor[m]: because you can do that by extending kernel-module-loader-service-type
<sektor[m]>unmatched-paren: Pretty much.
<sektor[m]>Although I am not sure how that would work; espeaup itself does not provide a module. The module itself lives in the kernel; it is called speakup_synth_soft. I think that the loadable-modules-service-type only arranges things to be loaded, but doesn't actually modprobe it into existance. Unless I'm misreading.
<unmatched-paren>sektor[m]: Yes, but kernel-module-loader-service-type actually seems to modprobe them.
<unmatched-paren>Whereas linux-loadable-modules-service-type makes them loadable.
<sektor[m]>unmatched-paren: Interesting.
<sektor[m]>Ah I've found it.
<sektor[m]>Now let's try this again; hopefully I can get speech working now.
<itd_>mroh, brendyn, unmatched-paren: Thanks for the input (and sorry for the delay). I'll look for something like that. :)
<brendyn>itd_, it also happens with zlib because thats also a license and package name
<unmatched-paren>and ruby
<sektor[m]>I didn't know there was a Ruby license.
<linj>what is the difference between the shapes of nodes
<linj>* what is the meaning of difference shapes of nodes
<unmatched-paren>linj: I wonder whether ovals are ones that extend shepherd, and squares are ones that don't?
<nckhexen>sneek: later tell sarg: Be sure to report that upstream, they should fix the test so it's skipped if IPv6 isn't available, but still fatal if it is. Requiring local networking in builds is fine, and packages with networking functions *should* test them — that's good!
<sneek>Got it.
<nashdidan[m]>Follow-up question on packagine, when working off a guix checkout, how do I build/package after adding a new definition of the package I'm woring on? I've update mail.scm and I would like to build from $GUIX_CHECKOUT with ./pre-inst-env guix build --keep-failed himalaya@0.6 return guix build: error: himalaya... (full message at <>)
<nckhexen>itd_, brendyn: That's a fun one! Looks like somebody used ‘expat’ the package variable where it should have been ‘expat’ the licence variable (often imported as ‘license:expat’). That was bound to happen, but went unnoticed until recently.
<unmatched-paren>nashdidan[m]: maybe try ``guix shell -D guix -- make -j$(nproc)'' again?
<itd_>nckhexen: I like this definition of fun. :)
<unmatched-paren>also, you should probably call the variable ``himalaya'', not ``hymalaya-0.6''.
<unmatched-paren>nashdidan[m]: put the *entire* package definition on
<civodul>nckhexen: oh, did the license 'sanitize' field find actual problems?
<unmatched-paren>apparently, yes! :)
<nckhexen>civodul: Seems like it! My guess was that it was a third-party channel, but I didn't ask yet.
<linj>unmatched-paren: probably thanks
<nckhexen>civodul: So congratulations ;-) If that's the actual error message it could use some work, although I'm not sure what I'm looking at (why repl?).
<civodul>nckhexen: nice! :-)
<civodul>i didn't see the error message
<nckhexen>itd_: Yeah, it's not great. It requires ‘knowing’ that this gotcha exists (zlib is similar: both a package and a license). How exactly did you produce this output?
<nckhexen>civodul: (repost)
<unmatched-paren>so, it's just spitting out the raw exception? surely we have some kind of machinery for translating that to a proper error message?
<nckhexen>We do.
<nckhexen>FSVO proper, but we do.
<civodul>nckhexen: ah yes, the message is "not great" if you notice during 'guix pull'
<civodul>it's better when developers notice while working on the code
<civodul>so yes, the exception is turned into an actual error message, but not if it happens while building the package cache during "guix pull"
<nckhexen>itd_: Cancel that question ☺
<nashdidan[m]>Ok, so this is what i've added to mail.scm... I didn't add the dependent libs yet, just wanted to check what happens when I try to build/search for this package... i.e. guix package --list-available=himalaya
*civodul was almost disappointed that it didn't catch any error in the guix-{hpc,past,science} channels
<nckhexen>My impulsive ‘fun’ above was for a similar reason.
<xd1le>hi guix
<sektor[m]>Bah still no dice; not sure if the ISO actually booted or not.
<zimoun>Considering integrity checksum, why does Guix use plain serializer for tarball vs nar serializer for other?
<zimoun>civodul: maybe? ^ :-)
<itd_>brendyn, mroh, unmatched-paren, nckhexen: Now I feel silly. It was indeed a '(license expat)' & '#:use-module (gnu packages xml)'. Not sure why I did not use 'license:' like everywhere else (in my third-party channel). Thanks for your help! :)
<rekado_>sbcl-closure-common has no license
<sektor[m]>Good news: it did try to load the module, but apparently it couldn't find it.
<lil>hello ><
<lil>what characteristics does the build server have?
<lil>i want to rent dedicated server for building packages in private network
<nckhexen>lil: Head node: AMD EPYC 7451 24-Core Processor | 192 GiB RAM? | 100T storage + about 30 or so build nodes: AMD EPYC 7551P 32-Core Processor | 128 GiB RAM | ~500 GiB storage. The nodes might be slightly heterogenous, I don't remember, but won't differ much.
<nckhexen>You do not need this.
<nckhexen>(That's only x86_64. Build nodes for architectures connect to the head node but are much less numerous, powerful, and homogenous.)
<apteryx>the test suite of nss is so ridiculously long
<apteryx>I wonder if it could be made to run in parallel at least
<Kabouik>Does r-tinytex not provide the pdflatex command? It isn't found on my system after installing that package
<Kabouik>No, it's probably just the R side of things. Any recommandation as to which LaTeX package is the most minimal in Guix?
<Kabouik>I installed texlive-base
***kasper is now known as boo_
<boo_>Hi! I'm having an issue where my Guix system fails to reconfigure if I add the bluetooth service to my config.scm. It works until it tries to restart services and then it fails with this error: No applicable method for "#<<generic> provided-by (1)>" in call (provided-by "#<unspecified>")
<boo_>My config can be found here:
<iska>anyone used rocm on guix? rock seems to be missing
<sneek>iska, you have 1 message!
<sneek>iska, pkill9 says: does pulseaudio get run for you when you start an audio application if you aren't already running pipewire?
<iska>pkill9 don't think so
<nckhexen>iska: but just from the description, it doesn't sound very useful on a free distro.
<iska>nckhexen pytorch seems to hang because rock is missing
<nckhexen>Oh, and it's from AMD. Further increasing the likelihood it's useless.
<nckhexen>(Without blobs, I mean; not snark.)
<nckhexen>iska: That's possible. I can't much help with that.
<iska>nckhexen is it possible to use dkms and how? guix search shows nothing
<iska>info page also shows nothing...
<nckhexen>iska: The description says ‘Note: currently this package does not provide GPU support.’ so I'd expect it to (1) work (2) not try to load the AMDgpu-related rock kernel driver.
<nckhexen>I don't think DKMS is easy or pleasant to get to work on Guix. I'm not aware of a package. We don't provide support for nonfree software here.
<iska>nckhexen well my wifi drivers are free but need dkms
<nckhexen>Nothing needs DKMS.
<nckhexen>It's one way to build out-of-tree modules, but it's not the only way to do anything.
<jcmdln>is this an official stance or personal opinion
<nckhexen>Guix can do that (IMO better, but people say I might be biased) with the linux-module-build-system.
<nckhexen>jcmdln: Which part?
<iska>nckhexen T2U nano does but it's verified on h-node
<nckhexen>I think you misunderstood me.
<nckhexen>A quick search: <> no DKMS needed.
<jcmdln>`This driver can be installed using [DKMS]`
<nckhexen>Can != must.
<nckhexen>If a driver uses a special non-standard build system, DKMS *can't* build it either, because it just issues standard (make &c.) build commands.
<unmatched-paren>I should remember to add qualifications when I search for acronyms. I tried searching "dkms" and got a German bone marrow donation nonprofit.
<nckhexen>> build instructions unclear, donated bone marrow.
<nckhexen>Weird. I searched for it on DDG, got the ‘Wir besiegen Blutkrebs’ result you mean, then searched on Google, then hit the back button to DDG but it has different results now? OK.
<nckhexen>(Without ‘blutkrebs’ anywhere on the page.)
<gnucode>hmm, I am trying to make tests that make sure that various incorrect records throw errors...
<gnucode>but when I write the tests that I believe should work, it is saying that the tests fail.
<gnucode>ok I am not catching the error apparently
<rekado_>Kabouik: tinytex comes with a command that installs a statically linked texlive subset
<rekado_>the executables require the linux loader to be located at /lib64/
***mark_ is now known as mjw
*podiki[m] hears something about executables and loaders, mumbles something about fhs containers coming to guix shell, and disappears
<apteryx>iska: T2U nano is not free per GNU FSDG, unfortunately. The driver has firmware blobs in C arrays and was removed from Guix recently.
<nckhexen>Did someone tell H-node?
*nckhexen didn't verify that it's actually listed there.
<gnucode>now my tests pass, but I am getting a lot of output to the screen...
<nckhexen>Oh, it's this one: (‘tested on: GNU Guix System’, hehmph).
<nckhexen>Someone has raised the blob issue, but I'm not sure what happens then — if anything.
<nckhexen>iska: So the H-node page does mention this, but it was added relatively recently (past few months at most).
<iska>nckhexen ohhhh
<iska>even the github one?
<apteryx>nckhexen: it's the same firmware that you removed recently
<apteryx>at least that's the one I was using with my Archer T2U Nano dongles
<apteryx>it was rtl8812au-aircrack-ng-linux-module
<nckhexen>apteryx: Yep.
<nckhexen>I'm pretty sure I added it, too, so my bad, and at least it didn't die at the hand of a stranger.
<futurile>does anyone have contact details for who ever runs I have been using the public inbox to keep up with guix development, but the nntp server seems to have gone down in the last few weeks.
<Guest3333>god i am SO TORN between which declarative package manager i want to use
<Guest3333>i'm wanting to plan out my next arch install in a way i can come back to *next* year and change
<Guest3333>guix has a better home manager, but nix doesn't use lisp. i would really give a lot to be able to use a shell script instead, but those aren't as futureproofed...
<unmatched-paren>"I would really give a lot to be able to use a shell script instead" isn't something you hear every day :)
<vagrantc>Guest3333: well, you've come to a largely biased channel to discuss the topic :)
<unmatched-paren>Lisp syntax isn't a disadvantage in any way.
<Guest3333>i am in the nix discord as well to discuss this with them
<roptat>hi guix!
<Guest3333>i simply don't prefer lisp syntax, because the indentation is based on word length instead of character number
<vagrantc>Guest3333: but you said arch install ... and while guix or nix could be used on top of archlinux ... generally they won't touch your native OS installed packages
<unmatched-paren>Evening roptat :)
<roptat>I'm trying to run gcstar, which is a perl program that uses gtk3. It looks like we have an issue in our perl-gtk3 package; no window is showing
<roptat>any perl expert around? :)
<unmatched-paren>Guest3333: What do you mean by "word length instead of character number"? :)
<roptat>this is a reproducer, it blocks right after saying hello:
<Guest3333>unmatched-paren: here is an example of lisp from the docs
<roptat>(you should be able to run it in a guix shell perl perl-gtk3)
<Guest3333>this is normal stuff
<unmatched-paren>Guest3333: sure...?
<unmatched-paren>looks normal to m
<Guest3333>however, the indentation seems to want to have each word end at the same place, rather than just having "levels" of 2 characters per indent, like every other language
<Guest3333>or however many characters
<unmatched-paren>that's a legitimate criticism
<Guest3333>is that a /s or not a /s
<unmatched-paren>and scheme's long variable names (see ``string-append'') do make preserving horizontal space quite annoying
<unmatched-paren>Guest3333: not a /s
<Guest3333>the indentation is the main reason i don't want to have my config files in scheme
<vagrantc>Guest3333: you can write it all on one arbitrarily long line if you want.
<roptat>after some time you probably won't see it anymore though :)
<Guest3333>vagrante: no
<Guest3333>woops missed the c
<vagrantc>Guest3333: e.g. you can write your own files however you want
<vagrantc>it's not indentation-sensitive
<roptat>and you can use any indentation you want, you don't have to follow the same conventions, indentation doesn't have any meaning in Scheme
<Guest3333>yes, but i'll constantly be fighting vim because it wants me to indent in an idiomatic way rather than my own way
<Guest3333>no i will not be using emacs
*roptat uses vim too :)
<unmatched-paren>Guest3333: doesn't vim's scheme filetype do the indentation the schemey way?
<vagrantc>Guest3333: you might need to have a conversation with your editor about allowing you to edit files how you want to :)
<unmatched-paren>back when i used neovim it seemed to work fine
*apteryx fails to see the indentation issue
<Guest3333>unmatched-paren that's true, and it doesn't want to do it *my* way
<unmatched-paren>Guest3333: also,
<Guest3333>yeah, fair
<unmatched-paren>but for short procedure names, it's often nicer to do it the first way
<unmatched-paren>that looks fine, right?
<unmatched-paren>it's just when you have stupidly long procedure names that it gets a little ridiculous
<civodul>Guest3333: you can also run "guix style" to obtain the conventional indentation if you want
<Guest3333>civodul i thought that'd put it in the idiomatic indentation instead
<unmatched-paren>e.g. (use-modules ...) probably should have been (use ...)...
<Guest3333>yo o/
<unmatched-paren>user_oreloznog: hi!
<user_oreloznog>Hi unmarched-paren and Guest3333 !
<unmatched-paren>Guest3333: I found Lisp formatting to be pretty weird when I first tried it, but it's just a different way of doing things, not necessarily a worse way
<unmatched-paren>There are issues, of course, but C-style syntax has issues too, e.g.:
<nckhexen>If you were not a regular I'd have my finger on the quiet button after that ominous ‘:’.
<unmatched-paren>Ah, yeah, it does look like I'm about to flood the channel.
<Guest3333>civodul how would i get it to do conventional indentation instead of idiomatic indentation?
<unmatched-paren>I'll just refrain from typing the message until I've made the paste next time.
<unmatched-paren>Guest3333: surely the two phrases are equivalent?
<Guest3333>unmatched-paren idiomatic does that "put the ends of each word together" thing
***HP-UX is now known as Linux
<Guest3333>oh hey a fellow steno boye
<Guest3333>wait no the stenography chord for UX would be different nvm
<unmatched-paren>HP-UX is HP's POSIX OS...
<unmatched-paren>AcTuAlLy, it's HPE's OS.
<unmatched-paren>Guest3333: As I was saying, C styling has problems too:
<nckhexen>Guest3333: <"put the ends of each word together"> I have no idea what that means. I'm with unmatched: your distinction between ‘conventional’ and ‘idiomatic’ sounds like something very… idiomatic to you (which is fine, but we'll need moreinfo to help).
<nckhexen>unmatched-paren: <I'll just refrain> Oh, no, I didn't mean anything by it. I thought it was funny.
<Guest3333>"idiomatic" would mean something that scheme inherently does. "conventional" means something from the conventions of other languages, specifically from outside of scheme
<nckhexen>It was nice having my spidey sense tingle & then be reassured by the nick. No offence meant.
<unmatched-paren>nckhexen: No offence taken :)
<Guest3333>if i could just write it like a bash script that'd feel so much better
<unmatched-paren>Guest3333: it's a completely different style of syntax
<Guest3333>very much
<unmatched-paren>you're not gonna be able to format it like a bash script
<Guest3333>yeah, understood, that's why i said "if i could," rather than "i'm going to"
<vagrantc>Guest3333: i daresay indentation when picking your package management system ... seems a small thing to consider. not much of a lispy-schemey sort of person myself ... but i pick guix for it's commitment to software freedom, reproducible builds more-or-less by design, and an amazing community :)
<nckhexen>But there is no ‘other language’ convention. You can indent Scheme like bash, or indent it like C, or Haskell, or Forth, but none of those are more ‘conventional’. If anything, Lisp conventions outrank them by age 😉
<nckhexen>*most of them. Type it, don't just think it, nckx.
<vagrantc>oh, and bootstrappability being a core consideration
<unmatched-paren>Indeed, the only reason C syntax feels normal is because most people use mostly or exclusively C-style languages.
<Guest3333>vagrantc yeah, it is kinda small, but i do want to integrate guix into my own config files, and having the scheme files not read or write the same way would be jarring
<unmatched-paren>The more home services get added to Guix, the fewer non-Schemey config files you'll need.
<unmatched-paren>ALL IS SCHEME.
<Guest3333>but apparently bash is not scheme yet
<Guest3333>that would be a happy day actually, if i could write a shell script that has the reproducibility and futureproofing of declarative package managers like nix and guix, i would be very happy and comfortable
<unmatched-paren>nckhexen: Scsh? :)
<nckhexen>I thought of it, but I have never actually tried it.
<unmatched-paren>Gash? Bournish? :)
<Guest3333>sh spoils me because i can do `[condition] && {operations}` and not have to do an if statement at all
<unmatched-paren>Guest3333: (and (condition) (begin (operation)))
<unmatched-paren>Or, if it's only one operation, (and (condition) (operation)).
<vagrantc>unmatched-paren: i don't think that will win someone over :)
<acrow>roptat: I've replicated the failure to open a gtk3 window you mentioned awhile ago. So, it's not just you! That aside, not sure I see how to fix it.
<Guest3333>unmatched-paren the functionality is there, but the terseness of the sh way isn't
<Guest3333>the built in extensibility is nice though, because that was the next thing i was going to say about the sh way
<nckhexen>I don't find shell terse compared to Scheme, I find it annoyingly verbose. Which just goes to prove, well, something. That's for sure.
<Guest3333>that's a skill issue tbh
<Guest3333>pardon the inflammatory affect but you could legitimately do code golf with shell
<unmatched-paren>nckhexen: WHAT="Disagreed."; echo "${WHAT#Dis}"
<Guest3333>just `echo "Dis"` then
<unmatched-paren>${foo:-...}, ${bar%%...}, etc are all pretty cryptic
<Guest3333>also shell is horrendous at string editing, that's why sed exists
<nckhexen>Implying that I lack skill in shell is too laughable to parry.
<Guest3333>ok fair
<nckhexen>I am, however, a ‘set -e’ zealot, so have fun with your naked && then 😉
<Guest3333>i will avoid `set -e` because of my naked &&
<nckhexen>I think we must now fight to the death.
<nckhexen>I'll check man bash.
<nckhexen>Yep, there it is.
<unmatched-paren>Guest3333: Sed... is also pretty horrible.
<Guest3333>yeaaahhhh it's unfortunate but i don't really use shell for string editing anyway
<nckhexen>Is this going to end in unironic ed advocacy.
<unmatched-paren>Awk is cool though.
<dgcampea>is there a list of recommended/tested (pcie) ethernet cards that will work with (stock guix) libre kernel?
<Guest3333>nckhexen let me quell that immediately, ed sucks lol
<Guest3333>it's unbearable to use
<unmatched-paren>Guest3333: ?
<nckhexen>dgcampea: ‘Check’ is the stock answer, but it's not a list, I'm afraid.
<unmatched-paren>dgcampea: I believe most should work?
<nckhexen>(I mean, OK, it's a list, but I don't find it very useful as such.)
<unmatched-paren>I think there are a few people who have incompatible wifi cards that rely on ethernet because the driver support for it is much better.
<nckhexen>Yes. One of the reason many cards are missing from H-node is because they are in general less problematic than Wi-Fi. But there are exceptions.
<unmatched-paren>My last laptop had an iwlwifi card that didn't work but an ethernet card that did, so I relied on that for a while after my wifi dongle broke.
<dgcampea>unmatched-paren: just got burned with one (r8169 0000:04:00.0: Unable to load firmware /*(DEBLOBBED)*/ (-2))
<nckhexen>(…so people don't report them as much. I hope my opaque reasoning was clear.)
<dgcampea>(0000:04:00.0: Missing Free firmware (non-Free firmware loading is disabled))
<unmatched-paren>dgcampea: Oh, that's pretty unlucky, I think...
<unmatched-paren>dgcampea: You sure that's not wifi?
<nckhexen>TBH Realtek in general is stayaway time.
<unmatched-paren>Ah, okay, I didn't notice the driver name in the first error.
<nckhexen>It's ethernet.
<nckhexen>I had it once. It was awful.
<nckhexen>Unstable even with vanilla Linux®.
<dgcampea>nckhexen: what do you mean with unstable?
<nckhexen>It would lock up the kernel and take a while to reset, IIRC. This was a few years ago, I don't remember the details, but my dislike of Realtek is 100% experience :)
<unmatched-paren>Linux, now with added arrays of seemingly-random numbers(r)
<nckhexen>They show up a lot in random second-hand hardware grab-bags, and they often don't work good.
<Linux>I might give up on Linux. Linux people are pretty stupid.
<Linux>It's pretty sad.
<Linux>Reason and logic evades these people.
<dgcampea>I thought ethernet cards were mostly equivalent (it just needs to receive and send packets after all)
<unmatched-paren>Linux: What...?
<nckhexen>Linux: Sounds like a plan.
<Linux>I don't get this freedom nonsense.
<Linux>What freedom does anyone have? If you have not the skills, you have not the freedom.
<Linux>As you are still dependent on other devs making the software for you.
<unmatched-paren>Look, I use the vanilla kernel. It was not a "omg we must wipe all the firmwarez even if necessary for use of computah!"
<nckhexen>Linux: Yes, you already posted that bizarre reasoning in handy PNG format. It's just as bogus in plain text.
<unmatched-paren>That isn't even what most people who use Linux-libre think.
<Linux>nckhexen: as they say, it takes one to know it.
<Linux>Bogus minds interpret bogusly.
<Linux>Or so they say. I don't. They do.
<unmatched-paren>Linux: Nobody is forbidden from learning the skills.
<Linux>That's not the inclination or aptitude of most people
<Linux>Therefore impossible for most.
<Linux>Who here has an issue with systemd?
<nckhexen>[Citation needed.]
<unmatched-paren>"That's impossible" != "I don't feel like doing it"
<Linux>Do you control which linux distros replace init with systemd? No. You end up using whatever the distro people decided, or go use a different distro. But if all distros abandon it for systemd, where exactly is your freedom?
<unmatched-paren>Linux: I agree there are some people who act as if systemd is the cause of all suffering in the world.
<vagrantc>my main issue with systemd is the proliferation of appending "d" to everything
<Linux>Your freedom is in your skills, so you can create a new distro to your needs.
<unmatched-paren>But I do think it has problems.
<Linux>But without skills, you're 100% dependent on others.
<Guest3333>Linux that's a very arch/gentoo way of thinking about this
<nckhexen>The vast majority of people are perfectly capable of understanding and writing software. They are not obligated, though. That's freedom.
<Linux>And not everyone has an engineering mind that can do math/programming/these higher level concepts
<Guest3333>nckhexen where's that xkcd comic about overestimating those outside of your field....
<Linux>So people without skills, no matter the adversiting nonsense of FOSS people, or GNU people, regarldess of the freedomds the skilled folks have, other peopel don't have that freedom.
<nckhexen>Again, the vast majority of people can learn these skills. Programming is not hard. Your ‘inferior minds’ premise is just flawed, Linux, and it leads you to wrong conclusions.
<Guest3333>capital U Users absolutely do not have the capability of understanding and writing software, and they don't *want* those skills
<Linux>The vast majority cannot learn these skills.
<Linux>That's a false assumption.
<dgcampea>What's this bizarre line of thinking? It can only be classed as freedom/libre if you have the skills to do so?
<unmatched-paren>I, myself, am not entirely convinced that most people *can't* do "these higher level concepts"; they *won't*. And fair game to them, it might not be their thing. Doesn't mean it's impossible for them.
<Linux>Have you ever talked to the vast majority? Their IQ isn't very high.
<dgcampea>Then almost nothing could be ever classed as 'free' since it will depend on some pre-acquired skill
<Linux>dgcampea: otherise you're not making your software, someone else is, with their thinking, decisions, ways of doing things.
<nckhexen>Linux: I'll get to the point. Do you have one, or are you just here to troll?
<Linux>Your freedom is when you create your own software.
<Linux>Welcome to ignore.
<Guest3333>not necessarily? bundling a bunch of software together can lead to creation as well
<dgcampea>or you can commission someone else to do it?
<Linux>You dislike something you hear, instantly namecall another as troll.
<Linux>People like you are who cause social problems in communities.
<dgcampea>provided you have the source
<Linux>Discussions like this do not cause isues.
<Linux>People like you do.
<Linux>Shame on you.
<vagrantc>Linux: i presume you also object to carpenters building houses
<Linux>vagrantc: again, it's about freedom.
<Linux>The house owners don't whien about freedom.
<Guest3333>Linux: 🙄
<unmatched-paren>Linux: Because, if they actually *own* the house, they are allowed to modify it.
<vagrantc>Linux: so, speaking of whining.... maybe you could do less of it?
<dgcampea>I question what definition of 'freedom' are you using here
<Linux>But where is their freedom? they haven't the skills, so how are they free to modify the house without skills and tools?
<vagrantc>Linux: and if you can't do less whining, could you do it elsewhere?
<Linux>Welcome to ignore my friend.
***ChanServ sets mode: +o litharge
***litharge sets mode: +q $a:kernel
<Linux>I am not going to address insults.
<unmatched-paren>Linux: Then they learn how to do it. With books, or the internet.
***litharge sets mode: -o litharge
<jab>vagrantc: "my main issue with systemd is the proliferation of appending "d" to everything" hahahaha!
<Guest3333>"i am not going to address the flaws in my arguments, such as my patently obvious bad faith throughout this entire conversation"
<Linux>Morons can't seem to understand a single fucking point.
<dgcampea>because freedom/libre in the FOSS world is defined by the "4 essential freedoms" (
<Linux>How far up is your head in your own ass?
<Guest3333>"i am going to take every criticism of my conduct in this discussion as an insult and call you a moron in an ill-fated attempt to look correct"
<nckhexen>Linux: Can you stop.
<nckhexen>I've asked nicely.
<Linux>When you address how something is said, rather than WHAT is said, you're obviously not worried about what is being said.
<hnhx[m]><Linux> "As you are still dependent on..." <- And you have the freedom to modify said software lol... (full message at <>)
<nckhexen>Linux: How it's being said doesn't matter. Improve the content, not the format.
<nckhexen>Low-quality content is the problem.
<unmatched-paren>Linux: We are addressing both. Rudeness will not endear others to your position, however.
<Linux>Instead of hating someone trying to make apoint, and finding faults in their conduct, maybe not be defensive, and call names like troll.
<Linux>You first namecaleld. I replied with moron.
<Guest3333>oh, my criticisms have been pretty obvious, Linux. the moment you mentioned IQ, you colored your perspective in ways that would take too long to properly correct
<nckhexen>Linux: ?
<Linux>You don't think IQ is a real think?
***ChanServ sets mode: +o litharge
***litharge sets mode: +b Linux!*@*
***Linux was kicked by litharge (You are banned from this channel (by nckhexen))
<Guest3333>gg tho lol
***litharge sets mode: -o litharge
<Guest3333>was going to actually try to tell him why iq is bad but honestly this is for the best
<nckhexen>Why do people who rant about IQ never seem to have it.
<Guest3333>because they want a justification for why minorities are bad and why the underclass is necessary
<Guest3333>it's a highly political reason
<podiki[m]>ugh, is it the season for trolls on irc too? keep it to kids in costumes getting candy please
<Guest3333>anyway AAAA my package manager situation is still unsolved
<unmatched-paren>Guest3333: Sorry about that.
<nckhexen>It was weird how they kept dodging criticism of their stupid opinions as if it were criticism of their tone. I wonder if they thought that would somehow work.
<Guest3333>some ppl you wonder if they're going to write a manifesto or join the proud boys
<vagrantc>glad to have things back on topic :)
<jab>who banned the Linux? by the way? I am sure that it is annoying policing the channel, but thanks for doing it! I've always thought guix is a fun room to hang out in!
<nckhexen>What ever was the topic? Blobs in Linux®? I'm sorry to have summoned a troll, but that nick should have tipped me off a lot sooner.
<HP-UX>I thought I was in #GNU. My mistake.
<nckhexen>Oh noes, another OS joins.
<HP-UX>This was right next to #gnu (#guix) in my channel list. I meant for that discussion to be in #gnu. Apologies.
***ChanServ sets mode: +o litharge
***litharge sets mode: +b HP-UX!*@*
***HP-UX was kicked by litharge (You are banned from this channel (by nckhexen))
<nckhexen>That's not how bans work.
***litharge sets mode: -o litharge
<SysV>Man you are petty and small.
<SysV>God bless.
<nckhexen>Yes. Bye.
<vagrantc>i daresay the pot is calling the chromed stainless steel kettle black.
<nckhexen>Ya know what, let's just
***ChanServ sets mode: +o litharge
***litharge sets mode: +b *!*@user/OpenSource
***SysV was kicked by litharge (You are banned from this channel (by nckhexen))
<Guest3333>lmao he's trying so hard
***litharge sets mode: -o litharge
<hnhx[m]>Is there any way to set the keyboard layout permamently in Xorg?
<hnhx[m]>On gentoo I did this:
<hnhx[m]>On Guix I use `setxkbmap hu`, however that sometimes just goes away and goes back to the default keyboard.
<unmatched-paren>hnhx[m]: yes
<unmatched-paren>i believe you can configure it with xorg-configuration or something...
<jab>How is do we know that SysV and HP-UX and Linux are the same person?
<Guest3333>i think you're able to make a Proper keymap but it uses a different syntax
<Guest3333>jab he mentioned "that discussion"
<jab>oh. I thought ya'll were doing some hipster cool think like tracking his IP address of something...
<unmatched-paren>jab: Well, HP-UX renamed themself to Linux earlier, before commencing their unhinged rant.
<Guest3333>why can't unhinged people be funny
<hnhx[m]>unmatched-paren: thanks
<podiki[m]>(ah had missed that too, glad some people are paying attention!)
<acrow>roptat: Could it be that perl-gtk3 fails to include the deprecated gdk-pixbuf-xlib? I see GdkPixBuf being called for in the error stream...
<jab>unmatched-paren: ah. As an observer watching the drama unfold...some of the discussion was a little funny to watch...
<unmatched-paren>hnhx[m]: In (service xorg-service-type (xorg-configuration ...)), of course.
<jab>sorry for being honest. ahaha
<hnhx[m]>people who troll on IRC must be the loneliest and saddest people on earth :p
<hnhx[m]>doesnt make any sense
<unmatched-paren>Oh god, now they're trying to continue it in my PMs.
<unmatched-paren>Can I block people on IRC...?
<Guest3333>i think there's /ignore
<vagrantc>most clients have an ignore functionality
<jab>hnhx[m]: that reminds me of the templeOS guy. People wanted to help the guy, but then he got super racist...and he didn't really want their help.
<podiki[m]>there's this too
<jab>I'm kind of curious how many people have been banned from this channel now...
<nckhexen>unmatched-paren: Yes, I recommend /ignore greatly. I wish I felt able to use it. 😛
<podiki[m]>hnhx: ^
<unmatched-paren>...and apparently senpai doesn't have /ignore yet. :(
<Guest3333>i should probably get this out of my system because i'm going to end up thinking about it for the rest of the day if not
<Guest3333>the reason IQ sucks is because it implies there's a certain class of people who not only cannot do things, but that they cannot learn how to do things, and that justifies an outward hostility towards a lot of the people who free software would actually benefit
<nckhexen>jab: Redonkulously few. I am not the ideal moderator or catalyst (I am learning…) but I dare say I'm proud of that.
<Guest3333>that should be it, it should be out of my system now
<nckhexen>I should not have helped feed them. That was bad of me.
<unmatched-paren>As podiki[m] mentioned, there does seem to be more of them recently.
<jab>nckhexen: sounds fair.
<unmatched-paren>nckhexen: Eh, it's really tempting to try to argue with them in good faith, even when they obviously aren't doing the same.
<unmatched-paren>I don't blame you for that.
<jab>nckhexen: did you change your nick?
<podiki[m]>on topic as well: do all services have tests? or should they all?
<unmatched-paren>podiki[m]: Not all.
<hnhx[m]>jab: he was an untreated scizho though, so...
<hnhx[m]>but im dumb
<hnhx[m]>podiki[m]: ye thanks i found this too
<hnhx[m]>it complains about
<hnhx[m]>`In procedure struct-vtable: Wrong type argument in position 1 (expecting struct): "hu"`
<jab>hnhx[m]: oh I didn't know that.
<unmatched-paren>hnhx[m]: how'd you know?
<podiki[m]> was where I learned about him
<unmatched-paren>hnhx[m]: Oh, right, I thought you were talking about the person who was just banned.
<hnhx[m]>no no
<hnhx[m]>I was talking about Terry
<podiki[m]>you should post your config hnhx with the problem
<podiki[m]>(pastebin of some sort)
<hnhx[m]>(service xorg-server-service-type (xorg-configuration (keyboard-layout "hu")))
<hnhx[m]>and it complains abotu this:
<hnhx[m]>In procedure struct-vtable: Wrong type argument in position 1 (expecting struct): "hu"
<unmatched-paren>hnhx[m]: Please put code blocks or logs like that on a paste site :)
<nckhexen>unmatched-paren: Yeah, but I should have known better sooner.
<nckhexen>jab: It is spooktobers after all.
<podiki[m]>i would pastebin the whole config to see the whole thing
<podiki[m]>but I think it is missing some keyboard-layout (e.g. the sample in the manual, it is more than just (keyboard-layout "hu"))
<podiki[m]>I think you are missing a keyboard-layout, if that is not confusing already
<dgcampea>hnhx[m]: put it as (xorg-configuration (keyboard-layout (keyboard-layout "hu")))
<nckhexen>unmatched-paren: I mean, if you set your nick to a common word, then come sealioning in… good faith quietly takes her leave.
<dgcampea>from my understanding, the first keyboard-layout is a field, the second is a procedure that returns an object representing a keyboard layout
<dgcampea>field of xorg-configuration
<roptat>acrow, I don't know, it can open a window if I don't use an entry, and I see exactly the same messages
<hnhx[m]>dgcampea: same issue
<hnhx[m]>: /
<nckhexen>jab: <How do you know> You can tell, and I actually meant to ban them by account name, but because I am a big failboi I got that wrong.
<nckhexen>Using a different client, so nothing is where I left it, dammit.
<acrow>roptat: Yes, I see the same thing. I tried naively loading gdk-pixbuf but it made no difference.
<hnhx[m]>oh okay I solved the issue, turns out I had to do
<hnhx[m]>`(xorg-configuration (keyboard-layout keyboard-layout)))`
<hnhx[m]>and `keyboard layout` was defined already to huf by the guix installer
<dgcampea>hnhx[m]: did you do (keyboard-layout (keyboard-layout "hu")) ? (note its repeated)
<hnhx[m]>dgcampea: yep
<dgcampea>because the option exists to allow for more keyboard-layouts than the one defined in operating system
<jab>nckhexen: haha. happy halloween
<jab>early halloween
<hnhx[m]>okay cool it works after a reboot
<Cairn>Would it be alright to submit a patch that doesn't work in order to get some help diagnosing the issue?
<hnhx[m]>the issue with `setxkbmap` was that if i disconnected the keyboard and connected it again it would default back the the english layout
<Cairn>To the mailing list I mean
<nckhexen>Cairn: Absolutely!
<nckhexen>You can add ‘WIP’ to the subject but it's not mandatory.
<Cairn>Oh that's a good idea
<Cairn>Thank you
<unmatched-paren>Cairn: git send-email --subject-prefix="WIP PATCH" [...]
<nckhexen>jab: So in this case, it seems that drugs were likely involved. Which is to say, let's not start diagnosing peeps with $whatever over IRC.
<Cairn>Nice, didn't know about that flag. I'll use it.
<nckhexen>That wasn't aimed at you specifically, really. But you know Guix. Big woke softies. Let's not stigmatise folks more than is strictly necessary for achieving world domination.
<Cairn>So I just want some feedback
<Cairn>I'm so close to getting this all working; I've gotten 4 new dependencies working. I just can't figure out this last build issue.
<unmatched-paren>Cairn: What's the error?
<nckhexen>And go ahead and share your code on, probably :)
<Cairn>That's the thing. All the other issues I've run into have had an individual error. In this case though, it's like there's some sort of fundamental incompatibility where a ton of small elements of the code aren't working.
<Cairn>Let me build the package again, then I'll paste the enormous build log
<unmatched-paren>Cairn: What build system does it use?
<unmatched-paren>That seems to be a general indicator of "how painful will this be to package" :)
<jab>nckhexen: I am now working on getting my opensmtpd service with records to have some tests. Basically the tests try to set up faulty records (conflicting, duplicate, or non-sensical)...when they do, an error should be raised. Each test just makes sure that trying to create silly records raises an error.
<Cairn>What I'm doing is updating Stellarium to 1.0
<unmatched-paren>Okay, I've had problems with CMake before, but I don't think it should be too bad...
<Cairn>It's already packaged in guix. But I needed to update a few QT dependencies to 6.3.1 (done) and needed to package two other dependencies (done). All I have left is to fix the definition of the main package
<unmatched-paren>Cairn: Hmm, if you're updating Qt stuff, these patches might need to go to core-updates...
<nckhexen>Pardon the noise whilst I tidy up my flailing.
***ChanServ sets mode: +o litharge
***litharge sets mode: +b $a:OpenSource
***litharge sets mode: -o litharge
***ChanServ sets mode: +o litharge
***litharge sets mode: -bo *!*@user/OpenSource litharge
***ChanServ sets mode: +o litharge
***litharge sets mode: -bo HP-UX!*@* litharge
***ChanServ sets mode: +o litharge
<jab>nckhexen: I like world domination! I wonder in say 10+ years how many kernels guix will support
***litharge sets mode: -bo Linux!*@* litharge
***ChanServ sets mode: +o litharge
***litharge sets mode: -qo $a:kernel litharge
<Cairn>unmatched-paren: That's good to know. Although I could just send those qt package updates separately, since they're rather trivial.
<nckhexen>Yuck. I wish it would cling to ops a bit longer, but that should be all. Sorry again.
<unmatched-paren>nckhexen: Hmm, was OpenSource another HP-UX nick?
<nckhexen>jab: Both of them!
<nckhexen>It's their IRC® account name.
<nckhexen>unmatched-paren: ☝
<nckhexen>jab: Is WSL a kernel? (Serious quesh, without getting into politics please — is it a VM or not?)
<unmatched-paren>nckhexen: It runs a Linux shipped with Windows.
<nckhexen>I think we support that now.
<unmatched-paren>Yes, apparently Windows comes with Linux now.
<nckhexen>Ah, so VMish.
<nckhexen>Guess it doesn't conut.
<unmatched-paren>Yeah, I think so.
<nckhexen>*count. Or coconut, if you're hungry.
<unmatched-paren>I think WSL1 might have been different though?
<Cairn>Alright while I'm waiting for this build to finish, here's my updated stellarium definition:
<nckhexen>I think that's where my confusion comes from (it's been… a long time since I've touched a Window. Long before WSL1 even.)
<Cairn>The Qt dependencies I updated are `qtcharts` and `qtserialport`. And the new packages I've made are `qxlsx` and `calcmysky`. Both work fine and don't seem to be related to the issue.
<unmatched-paren>nckhexen: I've never used WSL, these are just scraps of information I picked up from looking at the internet.
<jab>nckhexen: I hope hyperBolaBSD takes off. Their irc channgel is pretty lively today
<unmatched-paren>Cairn: You don't need #t at the end of phases anymore
<Cairn>As for the issue, I've added every possible extra dependency as tests and none seem to solve the issue. I've also messed with including the headers of `qtbase` itself (in the same way that the definition current includes the headers of `qtserialport`). Also not the issue.
<unmatched-paren>Actually, that was made unnecessary before the new style of inputs was added, I believe.
<nckhexen>jab: Yes! I know I've been critical of them recently, and of course they basically jumped into 5th gear right after that just to make me look the fool. I'm cautiously optimistic.
<Cairn>unmatched-paren: Good to know, I'll take it out. That was from the old definition.
<jab>nckhexen: what makes you say they jumped into 5th gear?
<nckhexen>‘They are doing things now.’
<jab>nckhexen: hahaha!
<nckhexen>Which, from not doing so, it a big jump.
<unmatched-paren>Cairn: Also, if you use gexps, you can do #$(this-package-input "...") instead of (assoc-ref %build-inputs "...")
<unmatched-paren>%build-inputs is deprecated now
<unmatched-paren>as are %output and %outputs
<unmatched-paren>I believe they've been replaced with #$output and #$output:foo, though I'm not sure whether the latter actually works; never tried it :)
<nckhexen>I think a world with a quirky FSDGBSD is better than one without, so I sincerely wish them luck.
<Cairn>Nice! That's also good to know
<nckhexen>unmatched-paren: It should!
<cwebber>WSL is not a kernel, IIUC it implemented the linux abi, but in WSL 2 they switched to using a VM
<cwebber>so now it runs the actual kernel, I think
<Sariboo>So I just randomly found this channel, and I was wondering about how "heavy" guix is compared to Ubuntu (assuming GNOME desktop for both)?
<unmatched-paren>nckhexen: I was told about it with a disclaimer that "I'm not sure whether this actually exists but I'm pretty sure it does"
<cwebber>Sariboo: that's a tough thing to answer, it depends on what direction you analyze
<cwebber>it probably has less heavy non-gnome dependencies, however
<cwebber>it also is more likely to spend CPU time installing a package, since binaries are sometimes not yet available when you want to install something
<unmatched-paren>Sariboo: Well, if you think this counts, it doesn't have systemd.
<Sariboo>unmatched-paren: heh
<jab>Sariboo: I will say that apt based software updates are pretty fast. but guix can take a little bit more time to download more bytes. in my opinion. I don't have any date to back that up though.
<Sariboo>I'm always on the lookout for a better "old/slow hardware" OS option
<Cairn>The build log is too big for
<nckhexen>unmatched-paren: It might be ‘lightly tested’ in some contexts.
<jab>Sariboo: I use sway with Guix System. Updates are pretty manageable with sway. manageable meaning not super time consuming
<nckhexen>Cairn: and (It also has a limit, so it's possible.)
<Cairn>Yes, too big for that too
<nckhexen>Otherwise, try to snip out the relevant bit, but (as I'm sure you know) that isn't always the last error. It's often the first. But not always, because computers hate us.
<Cairn>I wish there was a relevant bit. It gives me like thousands of lines of errors
<nckhexen>I'm not aware of a paste site without limits.
<nckhexen>Folks would probably upload base64'd moviefilms.
<unmatched-paren>Cairn: Snip out everything before 'build starts.
<Cairn>Just a little disappointed that has a limit, since I pay for my account
<Cairn>Yeah, so the error section is about 1.5 MB of text.
<Cairn>And the build log before that is about the same
<nckhexen>You could plonk it into a git repository.
<unmatched-paren>Oh my.
<nckhexen>(Not joking.)
<unmatched-paren>Cairn: this won't help your problem, but here's a reformatted version:
<nckhexen>I have to go, though. Good luck, and I guess ping me if shits hit fans again, but I don't expect them to. And I'm not the only op here. o/
<Cairn>Oh, has syntax highlighting?
<Cairn>Here, I got the error section to paste
<Cairn>This is around 75% of the way through cmake's process. You can see the error happens after a 77% marker
<Cairn>I'm assuming this is some sort of version mismatch between dependencies.
<unmatched-paren>nckhexen: Who else is op?
<Cairn>Oh, I should mention that I also tried using g++ and clang on a whim. That also wasn't an issue.
<unmatched-paren>Presumably maxim and ludo too?
<nckhexen>unmatched-paren: Mostly the expected suspects:
<unmatched-paren>Cairn: It only works if the file name you set has a recognised extension.
<unmatched-paren>nckhexen: Good to know.
<unmatched-paren>Huh, you founded the IRC channel?!
<Cairn>unmatched-paren: Thanks, that makes sense.
<podiki[m]>mostly doing something else, but the kind of errors and the number makes me suspect a version mismatch somewhere
<Cairn>I'm glad you suspect the same podiki =)
<nckhexen>unmatched-paren: Libera is a brave ‘new’ world 😉 But no, taylan grabbed the name first, in a wise and generous move.
<nckhexen>Really have to go no. Byezies.
<podiki[m]>like a qt5 vs qt6
<nckhexen>(I mean, we could have hostiletakovered it by talking to staff, but we didn't have to. o/)
<Cairn>podiki: This version is able to use qt6, and I've made sure all qt dependencies are version 6.3.1
<podiki[m]>just guessing. maybe look at what build options (cmake flags) they have, maybe something sets some version to expect? really don't know offhand though
<Cairn>I checked that as well, but I appreciate the suggestion.
<Cairn>Here's the relevant doc:
<Cairn>It seems to take whatever qt version is provided to it
<Cairn>Although it did suggest using CMAKE_PREFIX_PATH to the location where you've put qt. I tried that out as well to no avail
<Cairn>For that, I used a similar format to the existing phase
<podiki[m]>perhaps take a look at other packagers, like arch: they define a bunch of c flags, which I think is not normally specified
<podiki[m]>also -DENABLE_QT6=1
<vagrantc>Sariboo: guix will definitely consume more disk space than ubuntu or similar distros
<Sariboo>vagrantc: good to know, thanks
<Cairn>podiki: Hm, yeah, I just saw that
<Cairn>Maybe it's too new to have been documented yet? I'm gonna try a build using that flag
<Cairn>They also us -G Ninja, which I can't find yet, haha
<podiki[m]>i forget if ninja is already part of cmake-build-system? search around maybe, but hopefully it is the other parts causing trouble
<Cairn>I'm 🤞that is was the ENABLE_QT6 which I missed. It's a shame that it wasn't documented
<Cairn>Gosh, some of the Arch folks are so good at this stuff. I feel like I'm miles away from being able to package well
<podiki[m]>they have a lot of people and experience
<podiki[m]>and generally try to just do vanilla upstream (patches as needed) so it is a good reference I find
<Cairn>I'm at 70%. Approaching the moment of truth.
<Cairn>Darn. That wasn't it.
***lispmacs[work] is now known as forthmacs
<vagrantc>what's guix's 32-bit x86 architecture called? i686-linux ?
<civodul>vagrantc: yup!
<civodul>or i586-gnu :-)
<Cairn>vagrantc: i686-linux-gnu
<civodul>(the other one)
<civodul>vagrantc: there's now "guix build --list-systems" and "--list-targets"
<vagrantc>oh, i should use that!
<vagrantc>clearly newer than guix 1.3.0
<civodul>yes, it's recent
<vagrantc>i have some systems where i just have the guix package installed from Debian and don't run guix pull ... largely to have "guix hash" available.
<cwebber>a bit of an aside but might be a source of inspiration for a wayland wm thing
<Cairn>I guess the cross-compilation section should be updated in the manual then
<Cairn>With a more recent output of --list-targets.
<civodul>cwebber: looks fun!
<podiki[m]>I saw that recently too, has me dreaming of a stumpwm-like for wayland (so I can switch one day)
<cwebber> it looks like it has a nicely written guide, anyway
<podiki[m]>I know if I try to start writing such a thing I'll never be productive at my computer until it is done and perfect ☺️
<unmatched-paren>podiki[m]: What could be more productive than writing a Lispy wayland compositor? :)
<user_oreloznog>Good night guix!
<unmatched-paren>user_oreloznog: \o
<podiki[m]>unmatched-paren: anything lisp is always fun and I *feel* productive, but....they tell me that doesn't pay the bills (right now)
<podiki[m]>perhaps I should say it would destroy all other hobbies along the way too
<Kabouik>I think Nyxt folks got a EU funding a few years ago to develop Nyxt so that must have paid a few bills, but sadly just for a time (I hope they still have income, or that the funding is still ongoing)
<unmatched-paren>Huh, why'd the EU decide to fund a random web browser?
<cwebber>I'm funded :P
<cwebber>but it took a long time to get to that point
<Kabouik>I might be wrong, and I am sorry if I am, but I think I read something like that
<unmatched-paren>Nyxt felt kind of slow and clunky last time I tried it...
<Kabouik>> This project was funded through the NGI0 Discovery Fund, a fund established by NLnet with financial support from the European Commission's Next Generation Internet programme, under the aegis of DG Communications Networks, Content and Technology under grant agreement No 825322.
<unmatched-paren>Kabouik: Ahh, indirectly funded.
<Kabouik>I like it. It's not something I'd consider stable, but amusingly it tends to be more reliable than Icecat (which also has its quirks) for me
<unmatched-paren>I use qutebrowser at the moment.
<Kabouik>But yeah for some websites that are a mess in the first place, it can be unreliable/unstable. For Youtube too, but I use mpv for that (which reminds me that ytfzf throws an error on my Guix machine, but I can't solve that now, got slides to do for tomorrow)
<Kabouik>I liked qutebrowser, vieb and luakit too, but just scratched the surface of them
<Kabouik>Nyxt was my favourite because I could customize it a lot.
<jab>cwebber: congrats!
<Cairn>Ok, I don't know if this is the issue, but it looks like I didn't realize qtimageformats didn't have a qt6 version. Updating that as well, then I'll try rebuilding
<lagash>unmatched-paren: it's becoming faster and faster - sadly, it seems it's mostly WebKitGTK's fault, once we have an alternative web renderer we shall see better who or what to blame! :)
<lagash>unmatched-paren: you should totally idle in #nyxt sometime!
<unmatched-paren> aaaa stop tempting me, nyxt blog :)
<rekado_>jab: about download speed: I think the major difference is that has no network of mirrors, nor does it use a CDN.
<rekado_>so while the web server behind is well-connected the connection from your ISP to that server can be pretty limited.
<Cairn>Alas, still not the solution
<jab>rekado_: why does guix not have any mirrors? I find that a little odd.
<rekado_>it does have mirrors
<rekado_>but it’s not like we’re making it easy to mirror our substitutes
<jab>rekado_: so there currently is no cordinated effort to mirror the substitutes? I guess I better finish the IPFS work. :)
<jab>that may take me a few decades
<rekado_>jab: I recommend finding a co-conspirator who can help you get it done in fewer decades.
<civodul>actually berlin (the machine behind ci.guix) has rsync set up to export substitutes and a bunch of other things
<civodul>perhaps we could make that rsync module public?
<civodul>and advertise it
<rekado_>a mirror server would still need to run ‘guix publish’, though, wouldn’t it?
<jab>rekado_: I should probably find a co-worker to get me to finish my opensmtpd service with records...job description, "Wake up josh early in the morning and tell him to code. Use a whip if necessary."
<rekado_>two[m]: don’t know if you’ve seen this snippet before:
<rekado_>two[m]: does this address your problem of getting env vars in ‘tabular’ format?
<civodul>rekado_: true, that's kinda annoying
<podiki[m]>nyxt with guix package support seems very cool (though not sure I would use it for anything, love the interplay between the two)
<jab>civodul: I personally have no issue with guix download speeds. :) How much storage should a good mirror server have? 50GB?
<civodul>jab: more space means that older/less popular things can be kept
<jab>civodul: does your rsync module account for "oh geez, this mirror only have 10GB of storage."
<jab>only has* ?
<podiki[m]>i would be happy to help contribute funding for mirrors (a US one would be nice)
<civodul>jab: the rsync module is an "export"; it's up to the mirror to actually run rsync and DTRT
<civodul>podiki[m]: if you'd like to run one for a start, that'll give an incentive to look into making this easier, documenting it, etc.
<podiki[m]>don't we have some people that have gotten guix easily setup on some vps hosts?
<civodul>a research institute in France has offered to mirror substitute, but they're proposing NFS, which isn't as convenient
<podiki[m]>sure, I could help out on that front. there are also the mirrors that (I forget who, sorry) had for some speed tests and I wasn't sure if they were going to stick around
<podiki[m]>I work at a university here (in the US) but not in a compsci department, wonder if that is an avenue though
<civodul>that'd be nice
<vagrantc>curious about just doing caching proxies and whatnot ... though obviously that falls down a bit when the thing being cached is down
<podiki[m]>speaking of, I must do the teaching thing, but I'll do some investigating and we can talk more. or maybe a guix devel posting to see interest and who wants to help
<vagrantc>but with a caching proxy, if the purging policies were working well, you could have a fixed size that you're willing to proxy
<vagrantc>e.g. you don't have to mirror terrabytes of stuff
<jpoiret>civodul: did you have a look at ?
<jpoiret>i've managed to isolate it a bit more. Seems that in the second case, the git-minimal used to fetch the scons source doesn't get grafted
<jpoiret>what's weird is that i can't reproduce it using only scons as the test package
<civodul>jpoiret: i haven't looked at it yet! i'll try to do that tomorrow
<jpoiret>i've been using pk left and right but i haven't found too much yet
<florhizome[m]>where would you put Cairo–dock? And where this
<unmatched-paren>florhizome[m]: the ``light'' program goes in gnu/packags/linux.scm, so maybe put ``clight'' there
<unmatched-paren>and cairo-dock should probably go in gnu/packages/wm.scm?
<nckhexen>florhizome[m]: Probably hardware.scm.
<nckhexen>For clight.
<nckhexen>For cairo-dock… maybe, and somewhat reluctantly, xdisorg?
<unmatched-paren>nckhexen: Looks like lemonbar is in wm.scm
<nckhexen>(Reluctantly because it suffers from the same ‘I use this on Linux, that means it must belong in linux.scm’ problem that makes linux.scm a dumpster of everything in the kitchen sink.)
<nckhexen>unmatched-paren: Mhm. So is all the i3 stuff.
<nckhexen>I see it has a -core etc. If it has a lot of plugins and whizbangs that will eventually be packaged, cairo-dock.scm could be its own module, as i3 in retrospect could have been. I am definitely overthinking this. Standard answer: look in gnu/packages, pick what looks sane and isn't one of the unholy garbage magnets of admin.scm/linux.scm/xorg.scm, and you're probably good. :)