IRC channel logs


back to list of logs

<dftxbs3e>Blackbeard, well.. deleted /var/lib/gdm etc. with a config from manual and worked. Thanks.
<dftxbs3e>the lightweight desktops set one
<dftxbs3e>though it doesnt change much from what I had
<dftxbs3e>Blackbeard, though I wonder.. GDM offers me GNOME even though it's not installed
<Blackbeard>dftxbs3e: that happens to me to. I do not know why
<Blackbeard>But I am glad it is working now
<allana>Hi guix. I just wanted to report back that I got docker working on my system. For some reason the overlay kernel module is not loading. I modified my operating-system's initrd-modules: (initrd-modules (cons "overlay" %base-initrd-modules))
<allana>now everyhting works
<hugo_dc>@allana any advice on getting docker runnig?, where can I start reading?
<allana>hugo_dc: I was reporting back on a specific problem that I was having today on my GuixSD operating system. Best place ot start is the guix manual. Installing docker-cli to your profile gives you the client-side docker application and there is a docker-service-type if you want to run docker enigne as a daemon on GuixSD.
<allana>hugo_dc: Have oyu tried using it yet? I'm not sure if I was running into a bug or if it was a series of correlated events. In th past few days I have switched from the kernel provided in guix to a custom one
<hugo_dc>@allana, thanks, I've just installed Guix on my secondary computer, I installed docker and I can see the dockerd binary but I couldn't make it work, but yes, I assumed I still have to give a good read to the manual
<Blackbeard>allana: that seems like something worth to be on the cookbook
<allana>Blackbeard: Looking into the cookbook now. I have not familiarized myself with it at all.
<dftxbs3e>Blackbeard, I think it's because GNU Guix sets owners of folders to buggy values
<dftxbs3e>non-existent uids
<allana>hugo_dc: There actually is not much to the manual concerning docker, but I think that it provides a default example that I have used for the past year or so. I have had a problem for the last month, and as I said a moment ago, I'm not sure if it is a current bug or a different problem. In my personal story, I have switched from running GuixSD on a VM to running on physical hardware. Throughout this process docker has not been working
<allana>for me, but it could have something to do with all of the modifications that I have been making.
<allana>Anyway, thanks to everyone for helping me over the past couple of days.
***roptat_ is now known as roptat
<nothingmuch>this is obviously PEBCAK, but i tried to add a package (just guix import pypi compiledb, appended to gnu/packages/build-tools.scm) and ./pre-inst-env guix package -s python-compiledb doesn't show anything
<nckx>mroh: Thanks!
<nckx>nothingmuch: Did you wrap it in define-public, and does the NAME field match your search query?
<nothingmuch>nckx: duh, i knew it was something stupid!
<mroh>nckx: pretty trivial fix, but good for me to get used to the workflow ;)
<nothingmuch>sorry it's been a while since i played with this
<nothingmuch>thank you!
<nckx>nothingmuch: Happy to help.
<nckx>mroh: That was my evil (and obvious) plan. I expect many more patches now.
<dftxbs3e>Now I have another issue: when I type WIN-D in i3, it doesnt show the top bar and I can't type in the program I want to start, it just hangs and shows a loading cursor
<lfam>Make sure you have dmenu installed
<dftxbs3e>It's required by i3?
<dftxbs3e>Didnt know
<lfam>It's a separate program that makes the menubar for the typical i3 setup
<dftxbs3e>alright, thanks for the info
<dftxbs3e>I got a kernel panic on ext4
<Veera>What's the channel for Outreachy Guix
<nckx>Veera: Hi! There isn't currently one, AFAIK. I created ##guix-outreachy for the purpose but it's currently empty. You should talk to g_bor[m] here (probably asleep), or any other mentors.
<dftxbs3e>Veera, hey! welcome, I don't think there's a channel :-/
<nckx>dftxbs3e: Eekaboo ☹ Bad?
<dftxbs3e>nckx, I don't think I can have a log because it didnt let me save it :-/
<dftxbs3e>disk access was just broken
<dftxbs3e>it said null pointer reference in remove extent functions after rmdir
<nckx>Yeah, that sounded… stupid. What I meant to ask was: any corruption or just a confused kernel?
<dftxbs3e>nckx, unsure.. buggy kernel probably? Is there a way to get 4.19 LTS kernels on GNU Guix?
<nckx>(kernel linux-libre-4.19)
<dftxbs3e>nckx, ahh! amazing!
<nckx>It doesn't magically work for any version but we do ship every supported LTS.
<dftxbs3e>okay. what could make it not work?
<dftxbs3e>I'll try 5.4 maybe, it's the latest LTS :P
<nckx>I mean that it has to be a explicitly packaged version (guix package -A ^linux-libre$), you can't just write (kernel linux-libre-3.12) and expect it to work.
<dftxbs3e>nckx, ohhh sure!
<nckx>People expected that in the past.
*nckx tired, → zzz, good night all o/
<Veera>nckx: Hi!
<dftxbs3e>nckx, good night!! :D
<nckx>I hope someone can answer Veera's questions or show them where to go.
<dftxbs3e>I'm getting a "VM: Bad Instruction 71a8" error
<dftxbs3e>GNU Guile probably, when doing 'guix system reconfigure /etc/config.scm'
<Veera>Does Guix supports audio output
<Veera>Plan to run Guix in Qemu
<Veera>Does Qemu audio devices supported by Guix
<dftxbs3e>Veera, GNU GuixSD has audio yes
<dftxbs3e>It should work within QEMU, linux-libre should support it
<Veera>How much disk space is needed for MATE/XFCE based desktop setup
<dftxbs3e>Veera, I'd say 5-8GB or so
<mroh>Veera: I have a xfce qemu img w/o anything else and thats 3.8G
<Veera>dftxbs3e, Thanks
<Veera>mroh, How much space a development environment takes
<mroh>idk, maybe 5G or something.
<Veera>mroh, Thanks
<Veera>names #guix
<Veera>Does GuixSD supports LUKS
<vagrantc>yes, although it might be difficult to support luks2 until grub gets support
<nothingmuch>does this patchset look OK?
<vagrantc>nothingmuch: needs better description...
<nothingmuch>it's also broken, bashlex tries to create a file in site lib
<vagrantc>nothingmuch: be sure to run: guix lint python-bashlex
<nothingmuch>vagrantc: of the packages or the git commits?
<nothingmuch>ah, the packages
<vagrantc>one-liner descriptions are a bit short
<vagrantc>i swear i already packaged shutilwhich ... but i don't see it ... maybe i ended up not needing it
<nothingmuch>there's whichcraft which seems identical
<vagrantc>hrm ... substitute server seems stalled: ... missing any of the recent commits
<nothingmuch>oof, seems like bashlex unconditionally attempts to write to site lib... ;_;
<vagrantc>got some patching ahead of you :)
<nothingmuch>it failed, and seemed to work and i don't see where it's required
<nothingmuch>so i think i'll just remove the outputdir param
<nothingmuch>vagrantc: fwiw lint didn't say anything apart from it not being archived on software heritage
<vagrantc>thought it had a check for short descriptions...
<bandali>Veera, vagrantc, good news: support for luks2 in grub has finally been implemented:
<bandali>now we just have to wait for a new grub release :-)
<bandali>but we do have luks support right now; just not luks2 just yet
<vagrantc>bandali: yay!
<vagrantc>that seems worth at least having a grub-next or something :)
<bandali>hint hint ^, if anyone's up for it ;-)
<vagrantc>i mean, i don't know how hard it would be to actually maintain in parallel ... but ...
<nothingmuch>vagrantc: i updated the gist if you don't mind having another look
<nothingmuch>although i think it'd be preferred to run it once before installing
<vagrantc>nothingmuch: you lost one of the packages in your patch series, the first two patches are the same
<nothingmuch>gah, copypaste fail
<vagrantc>happens to the best of us
<vagrantc>nothingmuch: i'm not the most savvy guixer, but i can catch the basics :)
<nothingmuch>hmm, yeah looks like the nix packaging of compiledb does just that
<nothingmuch>vagrantc: nor am i, and i really appreciate a second pair of eyes, so thank you!
<jackhill>Veera: yes, Guix does support audo output (pending hardware support, naturally).
<jackhill>Veera: oops, sorry, I think I just replied to an old message that you posted some time ago
<jackhill>anyways, hi welcome. I'm glad to see that others have answered some of your questions :)
<nothingmuch>when sending a patchset, how do i know which debbug # it's assigned? do i have to wait for it to show up in the mailing list archive?
<jackhill>nothingmuch: you should get a reply from debbugs with the number
<jackhill>you're welcome. Looke for a subject like: "bug#nnnn: Acknowledgement (original subject)"
<nothingmuch>is it normally quick? not seeing a bounce or response yet
<nothingmuch>it's been ~10 min
<fishyfriend>nothingmuch: i submitted a bug report, it was 4-5 hours before receiving the auto-response (i assume patches work the same way)
<nothingmuch>fishyfriend: thanks! in that case i think i'll go to bed and follow up tomorrow
<fishyfriend>sure thing
<marusich>Hello Guix!
<Gooberpatrol66>I packaged a program but it won't launch and I can't figure it out. Would it be useful to submit the package regardless?
<jackhill>nothingmuch, fishyfriend: I think if it's your first one it may take longer (anti-spam measure)
<jackhill>Gooberpatrol66: I might send it to guix-devel@ and ask for help. Or put a link to the package definition here
<bandali>hello marusich o/
<Gooberpatrol66>it's the cadence package
<Gooberpatrol66>returns exec: python: not found
<Gooberpatrol66>i tried to use wrap-script but it didn't change anything
<nothingmuch>jackhill: makes sense, thanks!
<nothingmuch>and i actually received it 15 min ago =)
<Gooberpatrol66>is there a reason guix-devel isn't on this page?
<jackhill>nothingmuch: yay
<apteryx>nckx: haha... the VM's host name is 'gnu', but it asserts against expected 'liberigilo'. Shepherd 0.7.0 broke the hostname service, perhaps.
<jackhill>Gooberpatrol66: I don't know, but it is here: I wonder if it just hasn't been updated.
<jackhill>Gooberpatrol66: I'm going to bed for th night, if you don't here back here, the mailing list might work better (especially this time of day).
<apteryx>Gooberpatrol66: try adding python-wrapper as an input
<apteryx>just adding python adds a 'python3', but not a 'python' binary
<apteryx>ah, you already have it... hmm
<apteryx>hehe, you added it like a minute ago :-)
<apteryx>Gooberpatrol66: I don't think so; it looks like an omission.
<apteryx>I'm not sure exactly how that content ends up there though
<Gooberpatrol66>hmm nope
<guix-vits>Hi Guix.
<Veera>Is Danny Milosavljevic here
<Veera>Is Gábor Boskovits here
<bandali>not sure about danny m, but Gábor is g_bor[m]
<navik>rekado_: thanks for the feedback yday - it was me booting different versions and mismanaging guix cmd's in general that caused the issues.
<navik>now I'm curious why it seems I cannot use `(device (uuid "long-uuid-string"))` in a menuentry of the bootloader. Using the BIOS grub `(bootloader grub-bootloader)`
<navik>I had problems with the device having a short AAAAA-BBBBB UUID, which was corrected by reformatting the device. But now I get a backtrace from `ice-9/boot-9.scm:1515:18: object is not an exception of the right type [...]`
<navik>changing to `(device "/dev/sdaX")` is of course a quickfix on my side.
<guix-vits>navik: can you paste the config in ?
<navik>warts and all
<navik>guix-vits: it now works for dualboot with that config - just a bit of a tear in my eyes that I couldn't use the commented `uuid` device-line.
<navik>next step is to wipe everything and encrypt the partitions, redo everything the same way.
<navik>Oh, and try another BIOS, see if it can handle the dualboot.
<navik>But this is a nice milestone for today :)
<net-safe__>Gábor Boskovits or Danny Milosavljevic is here
***net-safe__ is now known as Veera
<janneke>Veera: please have some patience, g_bor[m] will read your messages
<Veera>Gábor Boskovits or Danny Milosavljevic is here
<janneke>Veera: `dannym' can be reached better by email
<Veera>janneke: I sent him a mail yesterday same time but he has not responded
<Veera>janneke: I am an Outreachy intern and he is the mentor for the chosen project
<Veera>janneke: Sorry Outreachy applicant
<janneke>Veera: Sure, welcome!
<janneke>Veera: good to reach out to us here
<Veera>janneke: thanks
<Veera>janneke: Outreachy page says first contact the mentor
<Veera>janneke: Gabor is the coordinator
<guix-vits>Veera: yes it's a first challenge for applicants :P
<guix-vits>find the mentor in woods of IRC
<janneke>Veera: yes, we'll make sure you get in contact :)
<Veera>janneke: I want to know how to start
<rekado_>Hi Veera
<rekado_>Gabor is usually online a little later.
<rekado_>It’s still a bit early.
<rekado_>Veera: which of the projects have you applied for?
<Veera>rekado_: Integration of desktop environments into GNU Guix
<rekado_>Veera: thanks, I’ll take a look at the project page
<rekado_>Veera: have you used Guix before?
<Veera>rekado_: No I have used Redhat/Fedora/Debian and others
<Veera>rekado_: Guix no
<Veera>rekado_: Have built myself Linuxfromscratch and Beyond LFS
<Veera>rekado_: Do not know Guile/Scheme particularly lambda exp
<rekado_>Veera: since the project involves testing of desktop environments in Guix you will probably need to build virtual machines running Guix System. Guix makes this rather easy.
<rekado_>one of the first tasks will be to submit a patch for a new package in Guix
<rekado_>so you’ll need Guix first.
<rekado_>I suggest you get started by installing Guix with the installer script.
<rekado_>You can download “GNU Guix 1.0.1 Binary” here:
<rekado_>oh wait
<rekado_>the installer script…
<rekado_>I suggest you fetch this script:
<rekado_>run it as the root user to install Guix on your GNU/Linux distribution.
<Veera>rekado_: Okay
<rekado_>at the end this will provide you with the “guix” command that you can use to build a virtual machine, for example, or to build packages.
<Veera>rekado_: But I want to use the iso image
<Veera>I am quite familiar with Qemu
<rekado_>you can do that or even download the ready-made qemu image
<rekado_>but note that you will probably have to build virtual machine images for testing
<rekado_>it may be easier to do this from an environment that is not virtualized
<rekado_>you can use Guix either as an operating system (called Guix System), or you can use it as a tool to install packages and build operating systems.
<Veera>In morning asked got that I need 5GB for Desktop and 3GB devel setup
<rekado_>personally, I think that might not be enough to be comfortable.
<rekado_>my recommendation is to either use Guix System as your host OS, or to use Guix on your host OS, but not to start with a virtual machine.
<Veera>I am thinking of 30GB setup
<rekado_>welcome back :)
<rekado_>my recommendation is to either use Guix System as your host OS, or to use Guix on your host OS, but not to start with a virtual machine.
<rekado_>it’s more flexible this way.
<rekado_>30G should be enough
<drainful>Guix intern would be a dream job; I'll have to check out the application next time.
<rekado_>(I’m most comfortable with 50+G myself.)
<Veera>VM setup instruction says to go for 50GB qcow image. Is that much needed
<Veera>Or I have to use and test
<rekado_>drainful: I think you still can. GSoC applications are still open:
<rekado_>Veera: when building packages you need a bit of space. Much of that will become garbage at some point which can be automatically removed. But it’s comfortable to have more space when developing.
<rekado_>drainful: the Outreachy applications, however, are closed already.
<rekado_>drainful: for Guix the list of project proposals is here:
<drainful>Oh that is great news!
<drainful>I had only seen the Outreachy program
<rekado_>Veera: for your project I think the easiest setup is to install Guix with the installer script, then use the “guix system” command to build virtual machines for testing the various desktop environments.
<Veera>janneke: Gabor TZ is UTC+1
<rekado_>Veera: you can easily remove Guix if necessary by deleting just a few directories, so it won’t “pollute” your host system.
<Veera>janneke: When will be Gabor available in his localtime
<Veera>rekado_: Okay
<rekado_>Veera: usually Gabor will be around at this time. Maybe something got in the way.
<Veera>rekado_: The project page specifically says install the OS not just the Guix pkg mngr
<rekado_>Oh, it does.
<Veera>rekado_: May be it is possible as you say but I do not know
<rekado_>That is unfortunate. Maybe Danny overlooked that possibility.
<Blackbeard>Veera: installing Guix is fairly easy
<rekado_>with “guix system” you can create virtual machines running Guixx System as needed, so I think that qualifies.
<Blackbeard>Is there a reason you can't dual boot with it?
<Blackbeard>You shouldn't have much trouble given you already did LFS this will feel like a piece of cake :)
<Veera>Perhaps hardware issues
<rekado_>Veera: I need to go now. If you just want to get started *now* you can install Guix the way I suggested first. You can always “upgrade” later if needed. Good luck!
<Veera>Using qemu and kvm and virtio drivers is fast and enough I think
<Veera>Running it fullscreen does the job
<rekado_>especially for the first task (submitting a patch) it doesn’t matter much.
<janneke>Veera: yes, you're still here :)
<Veera>janneke: yes
*janneke saw Veera's "test" message
<guix-vits>Veera: "/msg sneek sneek: botsnack is better" -- got smile is connected.
<janneke>sneek: botsnack
<janneke>guix-vits: yes, that's a nice trick
<leoprikler>huh, so you really have to mention sneek
<guix-vits>leoprikler: even in "/query sneek" one should mention sneek, because sneek!
<leoprikler>/query sneek?
<guix-vits>yes, /query sneek; botsnack is 0; sneek: botsnack is
<Veera>Is Gabor or Dannym here
<guix-vits>g_bor[m]: are you there?
<Veera>guix-vits: Even Ricardo Wurmus and Ludovic Courtès are listed as coordinators
<Veera>guix-vits: do they irc
<rekado_>I’m Ricardo.
<rekado_>I’m not mentoring this time and I coordinated Guix in Outreachy in the past rounds.
<rekado_>but you can always ask questions here.
<rekado_>have you tried installing Guix yet?
<Veera>rekado_: ah!
<Veera>rekado_: have started on it.
<rekado_>which method did you decide on?
<Veera>rekado_: qemu kvm iso method
<rekado_>let us know if you run into any problems
<guix-vits>Veera: Ludovic is civodul
<Veera>rekado_: oh yes
<Veera>guix-vits: ok
<kbw2222>i have an issue that's bewildering me. I'm trying to create a custom installation image, so i do `guix build --source guix` to get install.scm from the source tree. I copy and edit install.scm in my local directory
<kbw2222>but when i run guix system disk-image on it, i get guix system: error: failed to load 'install.scm': No such file or directory
<kbw2222>What's confusing me is that i don't get this error when I try to build it using the original file in the guix source tree. And when i use guix download to add my file to the store, or even add the original file back to the store, it doesn't work either and gives the same error
<kbw2222>So i don't get how it fails for an identical file on a identical filesystem... anyone encountered this issue?
<guix-vits>kbw2222: why not `guix system disk-image ...`?
<kbw2222>What do you mean? That is the command I am trying to run and failing at
<guix-vits>kbw2222: my bad.
<janneke>kbw2222: you work on a copy of gnu/system/install.scm?
<janneke>as that's a module that guix needs to find, it would need to be in GUIX_PACKAGE_PATH; with the correct prefix
<janneke>adding `.' to GUIX_PACKAGE_PATH is often tricky; guix will look for any scm files there; better to place custome modules in its own directory or tree and point GUIX_PACKAGE_PATH there
<kbw2222>janneke: thanks, that seems to have worked. I thought the guix system would just look at the file given as argument, did not know i had to help guix find it.
<janneke>kbw2222: yw, yes this may need some documenting somewhere
<janneke>wow, automake taking already > 30min to run its tests -- terrible
<civodul>hey janneke!
<civodul>the automake test suite takes ages
<janneke>civodul: hey -- actually, i'm pretty happy to "complain" about a packages' build speed on the hurd :)
<civodul>oh that's on the Hurd? woohoo!
<guix-vits>janneke: how much RAM hurd can make use of currently?
<janneke>civodul: hope you had a nice holiday -- i sent a hurd progress update to guix-devel ;)
<janneke>making some progres to build -f hurd-bare.tmpl now
<janneke>guix-vits: haven't looked into that yet i'm giving it 4GiB
<guix-vits>cool. i'll try to test it on my new SBC, when it arrives.
<jonsger>janneke, guix-vits: I think its 4GiB because of 32bit but I'm not sure
<guix-vits>jonsger: but it should fly on 64-bit ARM (not even sure which words to use for search query)?
<janneke>guix-vits: i haven't heard of arm work for Hurd since 2013 or so :(
<janneke>getting the hurd to boot in a guix vm would be an amazing next step
<Veera>g_bor are you there?
<guix-vits>g_bor[m]: ^
<nckx>g_bor[m]: ☝ Veera: I know they've answered mail today.
<Veera>nckx: mail did not reached me yet
<nckx>Veera: I just meant mail in general 🙂
<Veera>nckx: i am awaiting mail from DannyM
<Veera>nckx: ah
<guix-vits>Veera: Emacs' built-in IRC client has bash-alike Tab-key auto-completion: g_ + Tab = g_bor[m]
<nckx>Veera: I don't think I've ever seen dannym on IRC. Have you sent them mail? (Please also CC guix-devel at when you do, if possible.)
<guix-vits>/msg Veera V
<guix-vits>Possible completions are:
<guix-vits>Veera: vfP56jSe:
<guix-vits>viaken: vshyba:
<nckx>guix-vits: Do you know if Matrix (that's what the [m] stands for, right?) is clever enough to highlight <nick> and not just <nick>[m]?
<MtotheM>Have anyone every successfully built an x11 desktop container?
<guix-vits>nckx: no, i didn't.
<guix-vits>i was thought that [m] mean "moder" :)
<nckx>Oh, no.
<lfam>MtotheM: It depends on what you mean by "successful". It's not trivial to find all the filesystem paths that need to be shared with the container
<nckx>It's some IRC ‘bridge’ to another hip protocol. I think it's Matrix.
<nckx>Or Mattermost 🤷
<lfam>MtotheM: If you have a specific use case it's totally possible. If you want a "full desktop" it will easier to use a VM
<lfam>Did you find any of the older discussions about it, MtotheM?
<VeeraK>nckx: I have sent to Danny M but not to guix-devel at
<nckx>VeeraK: I'm sure they'll respond. Probably just taking a day or weekend off.
***VeeraK is now known as Veera
<MtotheM>I like the idea of having a bunch of desktops installed so I can try out new features and so forth. without all of the cross contamination that happens when you put them all in the same environment, but you are right. vm's are better for this. @nckx
<Veera>nckx: is mailing guix-devel at by non members allowed
<rekado_>Veera: yes.
<nckx>Veera: Allowed and encouraged! You might need moderator approval the first time you post, I'm not sure, but that shouldn't take long at this hour.
<Veera>nckx: oh!
<roptat>hi guix!
<roptat>Veera, although your first post will be moderated, so it may take time to arrive, subsequent posts will go to the list directly, even if you're not a member
<roptat>I'm trying again to create a maven-build-system, and I think I'm close. I have all the dependencies building with our current ant-build-system, and the maven-build-system is able to build a sample application with our maven, but only with binary versions of the dependencies I built (from central)
<roptat>The issue I have is that our ant-build-system does not install pom files, and does not create a maven repository layout, so I can't use the results. I'm trying to change that, so the ant-build-system will install jar and pom.xml files to lib/m
<nckx>roptat: I'm not actually sure that the conversation lists are moderated on first post. The debbugs lists, yes.
<roptat>oh, ok
<roptat>I managed to write something to override the dependency versions in the pom.xml files too, which is good (otherwise we would have to rebuild the java world, I mean almost every possible version of all the dependencies we have)
<roptat>however, I still am undecided as to how to proceed now, especially for packages that have mutual dependencies
<roptat>actually, I think I'm overthinking, it doesn't really matter because these loops are created because of test dependencies
<rekado_>roptat: couldn’t the maven-build-system take care of setting up the repository layout?
<roptat>it must at least have access to a pom file for each jar file
<rekado_>they don’t necessarily exist.
<roptat>if they don't, they can't be used in a maven build
<rekado_>can it generate them? Maybe from a property that can be added to each java package we have?
<janneke>hmm, what to do with prctl
<roptat>maybe, we could generate something from the groupid, artifactid and version
<roptat>but then we would lose some information on dependencies
<roptat>ok, retrying, but this time I'm ignoring build/test dependencies (it's annoying that everytime I change the ant-build-system, I get to rebuild icedtea-8)
<nckx>Veera: Your mail arrived (all 3 copies 😉).
<dftxbs3e>nckx, maybe it's the moderation buffer that's counter-intuitive?
<lfam>The mailing list moderation situation is terrible
<lfam>It affects the people who we want to have the smoothest experience of any contributor: the new people
<dftxbs3e>lfam, I think that one could make a user-friendly frontend to the mailing list
<dftxbs3e>like sourcehut or else
<lfam>It would need some kind of real anti-spam / anti-abuse system or it would require the same kind of manual moderation
<lfam>And there are a lot of people attracted to Guix and GNU who are vehemently against any of that technology
<dftxbs3e>lfam, same moderation but at least in that web frontend you can display a message to explain what moderation is
<lfam>(The anti-spam tech, that is)
<lfam>Yes, it's true. I still think it would be a bad experience to submit a comment on a web site and get a message that says "wait a day"
<lfam>It's simply unthinkable for younger generations
<dftxbs3e>lfam, then find a way to explain it in a way that makes it "thinkable" for anyone ever :P
<lfam>Contemporary web services don't even have this kind of spam issue
<nckx>dftxbs3e: Possible. I mentioned the possibility for delays up front, though.
<dftxbs3e>nckx, the issue with mail is that once they are sent you can't go back.
<nckx>lfam: Le whut?
<lfam>Even though email addresses receive thousands of spam emails a day, users don't see them
<lfam>Github issue pages don't get filled up with spam
<lfam>The spam is being sent but it is never displayed
<Veera>nckx: Sorry for that! First message I sent without subscribing to list
<dftxbs3e>lfam, it's because there's a captcha on registration
<lfam>Yes, and it's super effective!
<dftxbs3e>and I do receive spam mail from github. I set my email visible because I want people to reach to me when they enter my sponsorship program
<Veera>nckx: I waited it to appear in html archive but it never showed
<dftxbs3e>Veera, it's normal. there's a moderation queue and it doesnt show there until your message is approved
<Veera>nckx: In many list a subscription is needed to post
<dftxbs3e>I think the mailing list software should at least respond: "Your message is pending approval"
<lfam>Anyways, my point is just that it would be great to fix this issue but it would probably start some kind of internal culture war
<nckx>Adding to that delay: the archive isn't updated live. So people got your messages before the archive displayed them.
<Veera>nckx: So I subscribed and reposted the msg but again it did not arrived in inbox
<dftxbs3e>lfam, I don't think you need captcha to solve that user-friendliness issue at all.
<Veera>Sorry guys for 3 msgs.
<lfam>dftxbs3e: I'm sure you can get unwanted email addresses, but the workflow of "make pull request, see it appear" is reassuring to new users in a way that "did my email go through?" is not
<dftxbs3e>Veera, so we can conclude that you lacked feedback on your actions here and thus went to retry?
<lfam>I mean, unwanted email messages
<Veera>dftxbs3e: Yeah
<dftxbs3e>lfam, I think that Veera's issue here could be solved by having the mail software explicitly tell users when their emails are being moderated
<dftxbs3e>instead of being left in the dark
<lfam>Yeah that could be a great improvement for users. Usually antispam systems don't do this because it helps spammers adapt their approach
<nckx>dftxbs3e: I've been looking for such an option in Mailman for the past 5 minutes, but can't find any. Do you know any?
<Veera>dftxbs3e: There is a huge delay in receiving the mail I don't know why
<dftxbs3e>nckx, I never administered a mailing list so I wouldnt know. I would think this is a very basic feature
<nckx>Meet Mailman.
<dftxbs3e>lfam, you can shadow ban spammers on a case by case basis :P
<lfam>No, the volume of spam is way to ohigh
<lfam>Too high
<lfam>Email addresses receive thousands of spams per day
<nckx>dftxbs3e: We really can't sanely manage a blacklist of spammers. I can't anyway.
<nckx>We don't have ze tools.
<lfam>If you use gmail or fastmail and look in your spam folder, you only see a few. But those messages are the ones that the antispam system wasn't sure about. There are waaaaay more that are totally rejected
<dftxbs3e>so: it's an issue to give feedback to spammers as much as it is to give feedback to new users who are not familiar with mailing lists. What to do?
<dftxbs3e>What's more important?
<lfam>That is indeed the question
<dftxbs3e>I don't think guix-devel is specially a victim of targeted spam, so giving feedback wouldnt alter spammer's behavior just yet.
<lfam>It's hard to solve without machine learning tools such as captcha
<lfam>Why do you think that? Are you handling the moderation?
<dftxbs3e>lfam, I don't. But I think there's a lot of mailing lists to spam, and guix-devel is only one of them. Maybe with the recent rms drama there could be a vague of spam but that's all.
<Veera>nckx: I waited for 15 minutes for the archive to show up for first msg
<lfam>Every publicly listed email address is flooded with spam constantly
<dftxbs3e>lfam, not *targeted* spam
<dftxbs3e>targeted means that there's a person who's willing to spam a specific mailing list and thus will do everything to bypass its anti-spam measures
<nckx>lfam: I'm genuinely sorry if that's your experience but it's not universal.
<roptat>actually, my email is public, and I don't get that much spam (maybe one per day)
<nckx>I get a spam or so a day.
<lfam>It's not my experience because fastmail handles it
<lfam>Alright, that's good
<roptat>and I don't have an antispam on my server ^^'
<nckx>Same, because Guix doesn't have rspamd 😛
<dftxbs3e>same story here.
<dftxbs3e>email public but I don't receive that much spam
<roptat>I mean I have a blacklist of a few entries for some domains that have spammed me like crazy in the past, but it's very short
<dftxbs3e>not even 1 per week
<nckx>The ‘you'll be killed by the spam’ is up there with ‘running your own mail server is really hard you should never try’ meme that's in… well, certain interests to promote. Neither are really true, but I'm sure there are exceptions.
<lfam>In 2007 my university switched their email hosting to gmail. While transferring my mails from the school's internal email system to gmail, I witnessed a bug where gmail displayed every message they had ever received for my email address. These messages predated my having applied to the school, back to 2003, which is before gmail was even a public service. There were hundreds of thousands of spam messages sent to this one address, which
<lfam>had never existed. gmail had kept everything
<lfam>Maybe the situation has improved since then, due to ISPs becoming more restrictive with their customers
<nckx>I wonder if even spammers just assume email == Gmail and leave the rest of us reasonably alone. I wouldn't mind.
*lfam g2g
<nckx>Grr, Mailman calls the same thing (moderation) ‘hold’ in some places and ‘rejection’ elsewhere. I think. No wonder I couldn't find anything. So member_moderation_notice might do the right thing. But does it apply only to members, as the name implies? Who knows!
<nckx>cat ¯\_(ツ)_/¯ > mailman-user-manual.txt
<dftxbs3e>nckx, isnt sympa mailing list software better?
<nckx>I wouldn't know. Our mailing lists are currently hosted by the FSF/
<nckx>So we use what they offer (meh) or spread ourselves ever thinner setting up and maintaining our own listserv (meh). First meh currently wins.
<nckx>OK, I think I've set up notifications for held messages.
<nckx>…ff course this will conveniently notify trolls that they are being c€n$0reD. They could find out by checking the archives, of course, but sending them a helpful message is something else.
<dftxbs3e>nckx, isnt everyone new being moderated anyways?
<nckx>dftxbs3e: No, not at all! Only first-time posters.
<dftxbs3e>nckx, that's what I meant by new :P
<nckx>Sorry, I've been staring at HTML 4 too long and missed the ‘new’.
<dftxbs3e>Let's see what this change does to new users!
<nckx>There's no option to set the message, so Imma moderate myself and post to the list.
<dftxbs3e>nckx, I'm still personally puzzled with submitting patches, at least, I don't yet feel confident doing so.
<dftxbs3e>Also, is there a tool that helps people be compliant with the commit format and all?
<nckx>There's no linter for commit messages. You just have to learn (if you use emacs there are yasnippets for common actions), and until then you should find everyone very helpful and friendly. This isn't LKML.
<nckx>dftxbs3e: What's your greatest obstacle?
<nckx>It's not supposed to be puzzling, just ‘git send-email -1’ in the simplest case.
<dftxbs3e>nckx, rewriting my whole commit history to be GNU Guix compliant. It's also just a mental lack of confidence with the process that pushes me to delay it until I get more and more things working.
<dftxbs3e>When would one need multiple patches? Can a single patch modify/delete/add multiple files with git format-patch?
<nckx>Rewriting's non-negotiable I'm afraid.
<dftxbs3e>Sure, rewriting. Now, what's an easy way to go and modify commit messages way back?\
<NieDzejkob>Hmm, `guix size mplayer' returns "no available substitute information for '/gnu/store/9lvnm7zhf8n0rnl37b0c5wvz8dhd1cas-mplayer-1.4'", but `guix weather mplayer' says "100% substitutes available"
<dftxbs3e>like: git commit --amend but more than one commit backwards
<NieDzejkob>git rebase -i
<nckx>I use ‘git rebase -i HEAD~<number of commits in the past>’, then mark the commits with r[eword].
<nckx>dftxbs3e: ☝
<dftxbs3e>NieDzejkob, nckx: I see. I'll try that.
<nckx>If you use magit this is slightly faster & pleasant but it works with any editor.
<dftxbs3e>I'll install magit inside emacs :P - I've only recently been getting up on emacs
<dftxbs3e>nckx, what's your GNU GuixSD configuration? Especially emacs and DE/WM configuration?
<nckx>dftxbs3e: In general we prefer smaller commits that do one thing and do it well. All of Guix must be in a working state between commits.
<nckx>dftxbs3e: Relatively vanilla emacs configuration, emacs-guix, magit, i3, termite…
<dftxbs3e>nckx, how do you start i3?
<nckx>dftxbs3e: Slim auto-logs me in and runs my .xsession, which runs i3 through dbus-launch and some other crap.
<nckx>2 LUKS passphrases is enuf.
<dftxbs3e>nckx, I see.
<nckx>I hope this is sent to non-members (should be) and clear enough for the average newbie (because I don't see a way to change even the ‘reason’):
<nckx>Good enough? Good enough.
<dftxbs3e>nckx, Good enough.
<dftxbs3e>Thanks for that :-)
<nckx>Me, sysadminning
<nothingmuch>i was surprised to find lldb is not in guix, given clang stuff is, i was wondering if it just didn't happen yet or if there was a reason
<nothingmuch>nix has it and the definition seems pretty straight forward
<NieDzejkob>nothingmuch: it's just missing
<NieDzejkob>You can try packaging it, we'll gladly help if you run into an issue
<nothingmuch>NieDzejkob: yep, already my plan, i just wanted to make sure i'm not missing something before starting =)
<NieDzejkob>QEMU's build failed on CI, I'm trying to build it locally
<rekado_>I wonder if the mailman will ever be upgraded to the newest version with the pretty web interface.
<NieDzejkob>mroh: your patch is "bare" without commit metadata. Do you want the commit as "Michael Rohleder" or "Mike Rohleder"?
<mroh>NieDzejkob: ups, I am sorry. please take "Michael Rohleder"
<NieDzejkob>Ok. The command you want to use next time is git format-patch ;)
<NieDzejkob>(or git send-email, or both...)
<rekado_>huh? “Wrong type (expecting exact integer): #<condition &invoke-error ..."
<rekado_>since when does ‘raise’ require an ‘exact integer’ instead of a condition?
<mroh>NieDzejkob: thanks!
<NieDzejkob>Hmm, looks like I'll have to package some new Rust compiler versions...
<rndd>hi everyone! I have question about incorrect symbols in web browsers. usually it is about japanese or chinese symbols, sometimes it's about something like "->". how i can fix i?
<Gooberpatrol66>when i try to open something in thunar, i get "Failed to execute child process "gio-launch-desktop" (No such file or directory).
<Gooberpatrol66>Does anyone else have this issue?
<NieDzejkob>rndd: you need to install a font, then run fc-cache -fv
<Blackbeard>Gooberpatrol66: Ohhh somebody had a similar issue recently
<Blackbeard>Gooberpatrol66: what happens if you use a terminal to open the file
<NieDzejkob>rndd: font-google-noto is one font that should have almost any symbol in the unicode^H^H^H^Hverse, but it's like a gigabyte
<NieDzejkob>1494.1 MiB, to be precise
<NieDzejkob>alternatively you can try font-dejavu and font-liberation
<Blackbeard>Gooberpatrol66: check this
***sneek_ is now known as sneek
<Gooberpatrol66>terminal works afaik
<Blackbeard>Gooberpatrol66: did you check the link?
<Blackbeard>Gooberpatrol66: do you think you are facing the same issue?
<rndd>NieDzejkob: thanks, that works awesome
<NieDzejkob>rndd: Glad to hear that! Which font(s) did you go for?
***spk121_ is now known as spk121
<apteryx>using latest master and an offload machine: guix offload: error: corrupt input while restoring '/gnu/store/3k5fafyc5p0hfr8m34wqb3vw4zh646df-linux-libre-5.4.24-guix.tar.xz' from #<unknown channel (freed) 7ff7afaef080
<rndd>NieDzejkob: i dont know anything about fonts, so decided to just use font-google-noto
<nothingmuch>i'm helping a friend with a project, and i've been using guix environment for my own needs, i tried to search the web for examples of how to put that under version control but i'm not finding anything.
<nothingmuch>what's a sensible approach? at this point it's just a list of package names
<nothingmuch>is that what manifests are for?
<nothingmuch>looking at $GUIX_ENVIRONMENT/manifest it seems to me like it depends on the store and leaves the guix revision implicit
<rekado_>nothingmuch: yes, manifest are the right thing to use here. It can be as simple as (specifications->manifest '("this" "that" "something" "else"))
<rekado_>the ‘manifest’ file that is part of a profile (and an environment) is something else.
<rekado_>(it’s an ‘output’, not an ‘input’)
<jsoo>hi everyone. i am very new at irc.
<sirgazil>jsoo: Hello :)
<NieDzejkob>jsoo: Hi! If I'm understanding your nickname right, I looked at your channel the other day and noticed that your rustfmt package is out of date because they don't publish to anymore
<jsoo>NieDzejkob: Yes very possibly rustfmt is out of date. Last I checked, certain rustfmts required nightly and I was not in the mood to figure it out. I just wanted to try writing some rust.
<jsoo>sirgazil: hi!
<sirgazil>People using window managers, do you know something better than dmenu for launching applications? It seems dmenu only accepts ASCII as input and it doesn't seem to have command history...
<jsoo>NieDzejkob: I am quite ignorant of which rustfmt will work with which rust version. I can update the package to a newer one, if someone can point me to the right version.
<jsoo>sirgazil: I used rofi for a while but just switched back to dmenu that i patched.
<jsoo>sirgazil: I definitely can recommend rofi. It's a good program
<sirgazil>Thanks, I'll take a look.
<rekado_>my only complaint about rofi was that it can take a long time to start. I used it with stumpwm.
<rekado_>long enough to be noticeable.
<jsoo>i can vouch for that. i think the delay tended to be under 1 second for me