IRC channel logs

2017-07-14.log

back to list of logs

<joshuaBP`>so I'm trying to delete some packages that I don't need.... guix package -r artanis
<joshuaBP`>BUT guix is downloading files and updating subsitutes...why? that doesn't make sense.
<lfam>joshuaBP`: Which packages is it downloading?
<lfam>There are a few issues related to grafting and substitution currently
<joshuaBP`>zlib, linux headers, gcc, libpng, freetype, ...
<lfam>However, even if those bugs were fixed, sometimes removing a package requires some other packages. For example, if you've recently done `guix pull`, then you may need updated versions of a package in order to run the profile hooks when removing a package
<ng0>har. finally, I have a working latex output :)
<lfam>Here's one of the related discussions: https://lists.gnu.org/archive/html/guix-devel/2017-07/msg00025.html
<lfam>Another, fixed in the latest version of Guix: https://lists.gnu.org/archive/html/guix-devel/2017-07/msg00179.html
<lfam>And another: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=22990
<jonsger>if I add a new package which require a patch, should I pack both in one commit?
<lfam>jonsger: Yes, otherwise the package will be broken until the second commit is applied.
<jonsger>oke
<lfam>Bug about building disk and vm-images on core-updates: <https://bugs.gnu.org/27684> Help wanted!
<ng0>I feel like I'm using one or two too much, but this works for pdf, info and html :) guix environment --fallback --ad-hoc texinfo texlive-tex-plain texlive-tex-texinfo texlive-bin texlive-latex-base texlive-minimal -- makeinfo --pdf -v handbook.texi
<quiliro>rekado: are you available to continue helping me package python-qscintilla?
<cbaines>I've got an odd error from guix, I'm running guix system init, but it says: Throw to key encoding-error with args ("scm_to_stringn" "cannot convert narrow string to output locale" 22 #f #f)
<cbaines>any ideas?
<rekado>Hi Guix
<rekado>ACTION reads the chat logs now
<rekado>sneek later tell jsierles Yes, I have built julia 0.6 at least once before pushing the update. Haven’t tried recently, though.
<sneek>Will do.
<rekado>sneek later tell ng0 The big texlive package is called “texlive”. It still exists, unchanged.
<sneek>Got it.
<rekado>reepca, ng0 TeX Live is a bit more complicated than that. We can’t just propagate things. We have to generate configuration files for the specified subset as well. That’s what texlive-union does.
<ZombieChicken>Wasn't there a big update for guix that popped up on the mailing list fairly recently?
<rekado>like what?
<ZombieChicken>SOmethinghting about guile 2.2 and Guix having an issue with substitutes or something?
<rekado>that was a bug, yes
<rekado>this broke caching, and that’s now fixed.
<ZombieChicken>need to run guix pull;guix package -u;guix system reconfigure as root to fix it, right?
<rekado>right
<ZombieChicken>okay, thanks
<ZombieChicken>just wanted to be sure I got that right
<rekado>sneek: later tell quiliro Have you tried doing as the error message suggests? It says to specify the location of qmake by configuring with the “--qmake” flag, giving the location of the “qmake” executable as an argument.
<sneek>Got it.
<rekado>sneek: later tell quiliro But first you should check that “qmake” is actually available at build time. What package provides it? Is that package part of the native-inputs? You can use “find /gnu/store -name qmake” to find the name of the package providing “qmake”.
<sneek>Okay.
<rekado>sneek: botsnacks for you!
<rekado>sneek: maybe just one botsnack then?
<sneek>:)
<ZombieChicken>Where did the name "sneek" come from?
<ZombieChicken>or does anyone know?
<rekado>sneek: later tell quiliro “qmake” is part of the “qtbase” package, which is in the inherited native-inputs already. So you may just need to tell “configure.py” about it.
<sneek>Will do.
<TwistedFate>Hello everyone
<rekado>Does anyone know how Python / PyQt5 finds plugins?
<rekado>PyQt5 has a file PyQt5/uic/widget-plugins/qscintilla.py, which specifies how to load Qscintilla, but I cannot even import PyQt5.Qsci
<rekado>strace shows that it does not even look at the lib directory containing the qscintilla library.
<rekado>argh, nevermind.
<rekado>I forgot to add “--ad-hoc”
<rekado>hmm, things work in an environment, but they don’t work in the installed package.
<rekado>the difference is the PYTHONPATH
<rekado>if I run .openmolar-real inside of “guix environment openmolar --ad-hoc openmolar” it starts up fine.
<roelj>My build is failing at the 'reset-gzip-timestamps' with: "ERROR: In procedure open-fdes: Permission denied". What should I do about that?
<roelj>Change the permissions before this phase to make it writeable and after this phase making it read-only again?
<jlicht>hello guix
<mekeor>hello jlicht :)
<jlicht>o/
<Jackneillll>hey
<Jackneillll>is this rolling release distro i presume?
<adfeno>Jackneillll: sort of :)
<Jackneillll>please explain
<Jackneillll>the point of releases is to ensure stability, with this pkg manager you can even in rolling release
<Jackneillll>so why sort of?
<user_trisquel>Hi.
<user_trisquel>Can I route everything through Tor in Guix?
<alezost>Jackneillll: yes, it is rolling: after you do "guix pull", you get the latest "package recipes"
<user_trisquel>*GuixSD
<mekeor>user_trisquel: AFAIK no.
<mekeor>user_trisquel: well, you *can* but you'd have to first write the corresponding guix/shepherd-service yourself
<Jackneillll>alezost, where can i see/get to know what has benn changed compared to nixos?
<mekeor>user_trisquel: but there are services for tor and also there is a service to create a tor hidden service
<mekeor>user_trisquel: but there is no service for what you asked for, AFAIK
<adfeno>Jackneillll: There was atalk explaining the differences, it was during LibrePlanet 2016....
<jlicht>Jackneillll: AFAIK, only the nix daemon is still in use. And I believe reepca is working on replacing that as well
<mekeor>user_trisquel: ... not yet. but feel free to help implement it.
<mekeor>user_trisquel: you might also be interested in joining the infotropique project in that case. it's still mostly just an idea at this moment though: http://www.infotropique.org/
<adfeno>Jackneillll: .... The difference is that almost every object in Guix can be accessed by the recipes themselves, no need to do manual parsing with Awk, Sed, Head, Tail, Cat, etc.
<adfeno>s/The difference/One of the differences/
<Jackneillll>adfeno, and everything is 'guix' no multiple binaries?
<alezost>Jackneillll: I would say: the essence is the same (like all packages are placed in "store", user profiles, etc.), but the implementation completely different (only the daemon is mostly the same) - we use Guile all the client side; yes the one "guix" command to rule all actions
<Jackneillll>right
<alezost>Jackneillll: btw "guix" is not a binary; it's also a guile script
<Jackneillll>by completey different you mean you use guile (just a programming lang change), or semantic differences too / how things handled etc?
<adfeno>Jackneillll: Reference to that talk I mentioned: https://media.libreplanet.org/u/libreplanet/m/solving-the-deployment-crisis-with-gnu-guix-f8fd/
<Jackneillll>thanks
<solene>hello, is it possible to install virtualbox on guixsd ?
<adfeno>I recall that there are other videos too, not only from LibrePlanet, but I can't find them now, and I don't now if they are shareable (for the ones in LibrePlanet, I know that all of them are shareable because each video page shows the license).
<solene>or something that supports hardware acceleration ?
<adfeno>solene: Perhaps QEMU with KVM?
<alezost>Jackneillll: I don't know what "semantic differences" means; Nix uses some kind of bash/haskell custom language; we use a general purpose programming language. I don't really know the internals of Nix
<Jackneillll>shouldnt guix environment be renamed to env?
<Jackneillll>why the long name?
<Jackneillll>alezost, nix also patches elf files with patchelf
<Jackneillll>things lik ethat
<solene>adfeno: I'm going to try qemu but if possible I would have used virtualbox instead :)
<jlicht>Jackneillll: people in the guix project are not too keen on abbreviations ;-)
<Jackneillll>right
<roelj>How do I print the log of a build again?
<reepca>unless that abbreviation happens to be "drv", in which case we use that all over the place in the code.
<Jackneillll>drv?
<Jackneillll>derive?
<reepca>"derivation"
<jlicht>hmm good point reepca
<reepca>it's always a local variable name, so it's pretty unambiguous
<adfeno>solene: Hm... I guess... But if you do decide to do it, then we can only provide you information on using Guix, not VirtualBox.
<mekeor>Jackneillll: in some guix-mailinglist, there have been discussions to implement a user-configuration for Guix itself, which could also include user-specific aliases for commands (like "env" for "environemnt") (and also it could include user-specific services)
<adfeno>Because the latter is, as far as I know, non-(free/libre), and Guix support resources shouldn't foster use of non-(free/libre) software.
<adfeno>^ solene ( continuation of my last message)
<Jackneillll>do you guys use vscode? cant find it in packages. so even if tehres an .scm for it i need to build it? and wont be updated auto in that case. considering migrating from arch, but it has pkgs for everything at least in AUR, and guix doesnt as far as i can see barely anything
<jlicht>Jackneillll: probably not (not at my GuixSD machine right now), but the nice thing about guix is that it tries to erase the barriers between distro-hacker and user
<jlicht>Jackneillll: So it should be (in some sense) easy-ish to write your own guix packages, and possibly send them to the guix-patches mailing list :D
<Jackneillll>jlicht, nixos uses github, shouldnt you guys consider something more modern than a maling list? would be easier to create a PR for a guix-packages repo to add whatever pkg.scm file
<Jackneillll>mailing list seems like a barrier to me that my lazyness wont overcome
<Jackneillll>:D
<jlicht>Jackneillll: for me at first as well, but it has its advantages ;-). Seeing as guix is free software and free software promoting, a service like GitHub would not be confirming to those principles
<jlicht>The nice thing about the ML is that it allows each contributor to have their own workflow. In theory, you could even write a automated system that applies each patch series to its own git branch if you want :D
<Jackneillll>jlicht, right
<adfeno>Jackneillll: To understand what jlicht Said, see: [[https://www.gnu.org/software/repo-criteria-evaluation.html]]
<adfeno>Notice that in that reference, github is rated F (0).
<Jackneillll>adfeno, gitea?
<Jackneillll>gogs.io?
<jlicht>adfeno: Ah yes, I always think of 4 things and then share 3 of those things. Thanks for noticing
<Jackneillll>i think they should be evaulated at least
<adfeno>GNU Savannah, which provides most of the services that GitHub provides (including Git, mailing list, trackers for bugs/issues/task, custom webpages, and so on) is rated greater.
<adfeno>Guix usesthe services provided by GNU Savannah.
<Jackneillll>adfeno, though i wouldnt mind an UI that doesnt look shit, but yes i see your point
<jlicht>Jackneillll: I recall a discussion long long ago (maybe early 2016?) where this issue was discussed, and I think the passive consensus was that any Freedom respecting solution would be okay, as long as it /also/ allows a mailing list workflow. I could be mistaken in my understanding of this disccusion though
<Jackneillll>:)
<Jackneillll>i see
<adfeno>jlicht: Also free/libre JS should be made clearly marked to the visitor/guest of the website...
<adfeno>jlicht: We are also haveing this same discussion but more oriented to general comunity now in libreplanet-discuss, Mike Gerwitz also posted his answer to the same mailing list, but in his blog.
<adfeno>... and I kinda share his points too :)
<solene>adfeno: as far as I know, virtualbox is free with virtualbox-ose, the non-free parts can be installed separately
<jlicht>adfeno: Oh, would you be so kind as to share a link? On mobile right now :-)
<joshuaBPMan_>hello, I have about 20G on my root partition on GuixSD. Apparently I've used up 88% of it and have failed to run sudo guix system reconfigure /etc/config.scm twice now because I ran out of space. I've run guix package -d ; guix gc; Is there something else that I'm missing? Is there some other command that'll remove some space?
<adfeno>jlicht: See [[https://mikegerwitz.com/2017/06/Don-t-force-me-to-use-your-tools-on-the-Web]]
<adfeno>It also references to that mailing list discussion.
<jlicht>adfeno: much appreciated
<adfeno>You're welcome
<adfeno>:)
<reepca>joshuaBPMan_: you could try to delete some generations if you've got many.
<joshuaBPMan_>that's what I did with guix package -d
<joshuaBPMan_>:)
<jlicht>joshuaBPMan_: You might want to delete some system generations as well. Hang on, there should be a mailing list thread about this...
<reepca>... reading is hard.
<joshuaBPMan_>hahaha. So it is. :)
<joshuaBPMan_>I'm wondering if my log files have gotten quite large.
<joshuaBPMan_>I don't believe I've shrunk any of them.
<joshuaBPMan_>right now I'm trying to run du -h -max-depth=2 / to figure out which parts of my root partition are so large.
<joshuaBPMan_>jlicht: I did that.
<adfeno>solene: Strange....
<joshuaBPMan_>I only have 1 generation.
<adfeno>I'll search in Parabola, Guix repository, and also in Trisquel to see if they do provide virtualbox-ose
<jlicht>joshuaBPMan_: both profile and system generations?
<joshuaBPMan_>I ran guix package -d
<joshuaBPMan_>I did not think to run that command as root.
<joshuaBPMan_>let's give that a whirl.
<reepca>joshuaBPMan_: system profiles are what you get when you reconfigure, package profiles are what you get when you guix package -i blah
<adfeno>joshuaBPMan_: Profile generations can be delete with guix package -d, but for each user.
<joshuaBPMan_>ok. Yup. That seems to be it. I have several system profiles (aka sudo guix package -l)
<roelj>joshuaBPMan_: Try guix system list-generations
<roelj>Those are the system generations
<joshuaBPMan_>roelj: crap. I have a ton of those.
<joshuaBPMan_>How do I delete those bad boys?
<jlicht>joshuaBPMan_: https://lists.gnu.org/archive/html/guix-devel/2017-07/msg00006.html details some info on how to remove those safely
<jlicht>but be warned that you will not be able to boot into a generation anymore (seems kind of self-explanatory, but yours truly once deleted the last properly working system generation this way...)
<joshuaBPMan_>jlicht: thanks.
<joshuaBPMan_>I'll keep that in mind.
<adfeno>If I were to use GuixSD (I'm currently using Guix in Trisquel, not GuixSD), I would always keep a previous system generation that is known to work as a backup, just like I do when upgrading kernels in my copy of Trisquel.
<joshuaBPMan_>thanks for the advice adfeno
<adfeno>joshuaBPMan_: Just be careful as to not delete the system generation that you are currently on :)
<joshuaBPMan_>adfeno: Sure. I've deleted 1-10, but I've still got 11-15 left.
<adfeno>:)
<joshuaBPMan_>guix package --show=guile
<joshuaBPMan_>sorry thought I was in the terminal
<jlicht>is it possible include source files in a `guix pack`-generated tarball?
<lfam>ACTION compiles a list of core-updates package build failures
<lfam>Act now! Or there will be no more trivial build failures of Perl modules on core-updates for you to fix ;)
<taylan>funny anecdote: after a Debian upgrade, X.Org's keyboard and mouse input broke. I tried out 'gnome-session --session gnome-wayland' and it works almost perfectly; only synaptics is a bit jittery. I can even use 'xkbcomp' to set my keyboard layout, somehow O_o
<lfam>Hm, that's quite bad for Debian!
<lfam>taylan: As a Debian user, I'm curious, did that happen on the stable branch?
<taylan>lfam: yes, strangely enough. maybe it's something with my config though.
<taylan>actually, I'm not 100% sure whether I didn't enable testing at some point. how do I check again? :P
<lfam>Now I'm worried! Although yesterday I took an X update from them, and everything is still working
<lfam>I think that if you go from stable to testing, there is no way to ever go back completely, but I'm mostly ignorant of how it works
<taylan>my sources.list: http://sprunge.us/DdXE (yes, non-free *covers face in shame*)
<lfam>Ah, you are on old-stable, not stable ;)
<taylan>I have no idea, I use Debian rather casually :P
<lfam>Feeling confused and worried about breaking my system is what attracted me to functional packaging
<joshuaBPMan_>well I just reconfigured, and couldn't start X. BUt no worries I just switched back to my old profile. Very cool.
<taylan>exactly! wish I could do *that* with Debian!
<taylan>if I find time these days I guess I can start trying out GuixSD
<taylan>ACTION disappears again for now
<lfam>joshuaBPMan_: Hm, that's not good! Can you file a bug?
<lfam>Woooow: https://git.savannah.gnu.org/cgit/guix.git/commit/?id=a2bda5e4e30166e5b278276705e61008eb6110d2
<lfam>Much thanks to mark_weaver
<joshuaBPMan_>lfam: I bet I just did something wrong with my config file.
<joshuaBPMan_>I'll take a look at it a bit later.
<joshuaBPMan_>also, the error that I saw on the console was /var/run/slim.auth doesn't exist.
<joshuaBPMan_>Then it said that I should remove the file /tmp/.X0-lock
<ng0>mekeor[m]: it is more than just an idea, we are just very intransparent at this point.. should be better once we have finished the discussion which is ongoing atm.
<sneek>Welcome back ng0, you have 1 message.
<sneek>ng0, rekado says: The big texlive package is called “texlive”. It still exists, unchanged.
<ng0>rekado_: thanks!
<rain1>cgan I read this discussion
<ng0>yesterday we decided the content of the discussion is not in a state for public yet..
<ng0>but I hope it'll be finished soon, it's just redefining and shaping ideas we already understand but in a way that others can understand it aswell at the end of it
<lfam>ACTION prepares expat update
<lfam>It would be helpful if an LWN subscriber could share the content of this article: https://lwn.net/Articles/727703/
<adfeno>Hm....
<ng0>what's the subject?
<adfeno>I wonder if these articles really have a license to allow sharing.
<adfeno>ng0: Rethinking the Stack Clash fix
<lfam>... or share a link to a relevant upstream Linux discussion
<lfam>Typically I'm happy to wait 1 week to read LWN, but it would be good stay on top of this subject
<cehteh>is guile emacs dead?
<cehteh>build fails .. and last commits are 2 years ago?
<joshuaBPMan_>cehteh: I think there is not a ton of work going into it...
<joshuaBPMan_>there might be a newer branch somewhere.
<joshuaBPMan_>I believe that it needs to be able to do some optimizing of the elisp code still.
<cehteh>i tried some time ago and it was horribly slow because everything was interpreted
<cehteh>much slower than normal emacs
<cehteh>but now installing the package tries to build it and fails .. "required library not found ... libjpeg"
<joshuaBPMan_>also, I ran a guix system reconfigure /etc/gnome-awesome.scm today and now I can't log into gnome. I'm not sure what went wrong... I can still log, but X has not started. I did see X start to load, but then it quit. I've pasted my code on a pastebin.
<cehteh>so even the package description seems stale/broken
<joshuaBPMan_> http://paste.lisp.org/display/350862
<joshuaBPMan_>If anyone feels like double checking that I didn't do anything stupid. That'd be great. But by no means do you have to.
<joshuaBPMan_>I looked at the info documentation for what a normal desktop config needs to look like.
<cehteh>do you have a xorg log somewhere?
<joshuaBPMan_>I'll try booting said config in a vm and see if it still fails.
<cehteh> /var/log/Xorg.0.log
<joshuaBPMan_>cehteh: yeah. I used to have it installed too, but it is a bit slower than normal emacs. I hadn't tried to install guile-emacs in a while. I didn't know it was broke.
<joshuaBPMan_>And I'll check out the log good idea.
<cehteh>when the session starts but aborts instantly it could be that your windowmanager is broken or crashes
<cehteh>ACTION uses i3 from .xsession
<cehteh>dunno if the wm invocation and stuff gets logged somewhere
<joshuaBPMan_>grep EE /var/log/Xorg.0.log says
<cehteh>but you can hack your xsession a bit
<joshuaBPMan_>(EE) noveau(0) copy failed to allocate class.
<joshuaBPMan_>grep WW mentions open ACPI failed.
<cehteh>nvidia card?
<cehteh>i have some warnings and errors too in the log (running on kvm)
<cehteh>but still works
<joshuaBPMan_>hmmm. Yeah I have an old macbook 7.1
<joshuaBPMan_>I'll try running it in a vm and see if the issue persists there.
<cehteh>can you start xorg manually?
<cehteh>root
<cehteh>r0s3bud
<cehteh>top
<joshuaBPMan_>how do you start xorg manually?
<cehteh>top
<joshuaBPMan_>top?
<joshuaBPMan_>that starts xorg?
<eric23>startx isn't
<cehteh>Petter: qkillall virt-manager
<joshuaBPMan_>I tried running startx...I think
<Petter>o.O
<cehteh>wtf
<joshuaBPMan_>I suppore I could try xinit
<joshuaBPMan_>let me reboot and try that.
<cehteh>i did 'Xorg :1' in the (remote) kvm i connected with virt manager and my local X/vt crashed
<cehteh>mhm maybe i did that in the wrong term :D
<joshuaBPMan_>so I cannot manually start X in my latest system generation. xinit and startx did not work.
<joshuaBPMan_>they were unknown commands. Perhaps there is a way to start them with the shepherd.
<cehteh>mhm something i borked on my installation too
<ng0>afaik startx on its own never worked, at least from what people wrote on the mailinglist as it has to be started as root. Bzut I might be wrong
<joshuaBPMan_>i tried as root too. It still didn't work. It claimed it didn't know what the command was.
<cehteh>what you install as system isnt visible to root
<cehteh>but i havent get the details in guix yet either
<catonano>good evening, guix
<jdz1200>Trying to enable SSH access during install mode (usb stick image). Docs say to set root password (which is unset by default) using passwd, but [asswd doesn't appear to be in my install image.
<jdz1200>Is this expected behavior?
<jdz1200>Re: guixsd-usb-install-0.13.0.x86_64-linux
<lfam>jdz1200: It's a bug that we fixed since then. Use `guix package -i shadow` to install it
<jdz1200>Thanks lfam!
<lfam>You're welcome!
<amz3``>is guix running on guile 2.2 already?
<qx3>hello. i'm planning to try guixsd tomorrow. one question: does guixsd support an encrypted swap partition? i plan to use a btrfs root partition, so i cannot use a swap file
<amz3``>well guix on ubuntu stopped working for some reason
<amz3``>guix pull works tho
<amz3``>I mean guix pull is raising a few errors but seems like it not fully broken
<qx3>at the moment i use a 512-byte random key file for swap, and i need to configure /etc/crypttab
<amz3``>does guix plan to support gentoo-like use flags at some point?
<amz3``>ng0: what do you think of use flags in guix?
<amz3``>ACTION reading npm story
<ng0>the short answer: not really necessary, or could only be considered in some parts. I don't really have time right now to discuss or chat
<ng0>GUIX_PACKAGE_PATH does an amazin job combined with (inherit) etc
<amz3``>ok
<jlicht>amz3``: You can have a look at stuff like `package-with-python2`for an example of programmatically rewriting packages ;-)
<amz3``>oh, I know a little about that pratice, but it's not really use friendly
<amz3``>user*
<jlicht>not at all
<amz3``>it's just that gentoo was for the me the door into gnu/linux land many years ago and helped me grow solid skills in gnu/linux, I wondering whether guix/sd can do the same
<amz3``>and the primary reason I used gentoo was 1) because it was looking like the real thing and not a toy (compared to ubuntu and mandrake) and 2) it was damn fast compared to ubuntu
<jlicht>sneek: later tell lfam: It is totally okay to share the link, I just checked: https://lwn.net/SubscriberLink/727703/0a81571775f94212/
<sneek>Got it.
<jlicht>for anyone interested, this is a link to the lwn article ´Rethinking the Stack Clash fix`
<amz3``>nowdays, luckily I don't need my work station to be fast, I mean that my laptop is fast enough, but it might not be the case of every body
<reepca>silly question - which package is the "which" command in?
<reepca>ah never mind, seems to have its own package
<jdz1200>I added this [1] to the bare bones config and received error: "In procedure module-lookup: Unbound variable: grub-efi" 1) https://pastebin.com/raw/uNTPF8ag ... I tried adding grub-efi to packages, but no dice.
<jdz1200>What am I doing wrong? :)
<jdz1200>(I also added the bit about efi filesystem, but that seems unrelated to the error)