<mbakke>lfam: Updating Cython on staging sounds sensible. I'm hoping we can start it when Hydra calms down. <lfam>mbakke: Cool, I'm gonna push the update to the staging branch <lfam>qwertasd: After rebooting, is there anything in the X log? <qwertasd>Fatal server error: systemd-logind disappeared (stopped/restarted?) <lfam>It's not that easy to help without more info <lfam>Ideally the entire log and your config.scm. Other people may even see the discussion in the log and offer advice <qwertasd>I started xorg with the script in the tutorial <lfam>What did you put in '~/.xinitrc'? <lfam>I don't have a graphical GuixSD system to check, but I wonder if logind keeps any logs <lfam>They are typically in /var/log <qwertasd>Well maybe I have to use graphical login manager <qwertasd>How can I change theme of slim? Can I just download a new theme ***jonsger1 is now known as jonsger
<A87221>Hello, my full reconfigure fails with an error. Appears to be a seg fault. gstcheck.c:596:E:general:test_opus_encode_properties:0 <mange>Mmm, that's a pain. I assume your config.scm depends on gst-plugins-base, in some way? <A87221>Hello mange. What is gst-plugins-base? How can I determine? <OriansJ>A87221: if you run guix package -s gst-plugins-base ; you will see that it is the Plugins for the GStreamer multimedia library <mange>A87221: Can you show us your config.scm? <A87221>How may I exclude the package, OriansJ? <A87221>mange, I am trying to build an LXDE environment. I have a minimal env now, trying to run reconfigure. <A87221>OriansJ, I ran the cmd. How may I exclude the gstreamer mm library? <mange>Can you paste your config.scm using the paste link in the topic? It looks like LXDE doesn't depend on gst-plugins-base, but something else might. <A87221>mange, I need a few moments to scp the file to this box. <A87221>lfam helped me yesterday to edit the config file. <mange>It looks like gst-plugins-base is being brought in by colord, which is in %desktop-services. It might also be brought in by other things. <mange>So, to work around the bug (that gst-plugins-base doesn't build), we'll have to do something about that. <mange>Yep, but I have to do something else for a few minutes. I can help you out soon. <OriansJ>A87221: http:paste.debian.net/1034447 here is my minimal setup, you might find it useful (It does however encrypt the entire drive and requires you to set the volume's label to root) <mange>A87221: This is my attempt to exclude colord from your config: https://paste.debian.net/1034448/ I haven't yet verified that gst-plugins-base is gone, but I figured you may be interested in the type of change to make. <mange>This is a bit hacky, because colord-service-type isn't actually exposed, so we need to use @@ to get access to it. <Kurt86>Can someone help me solve this problem? I receive this error: grub-install: error: /gnu/store/bdzxdpdw25k8v6lz54clz42bilx47srj-grub-2.02/lib/grub/x86_64-efi/modinfo.sh doesn't exist.Please specify --target or --directory. <A87221>mange, are your only changes in lines 60-62? <mange>I also changed line 3, to add (srfi srfi-1). That's what let me use remove in line 60. I also added another closing paren on line 63, to balance them. <OriansJ>A87221: it strips out all the services and packages that are not explicitly listed (hence why it has to explicitly list bash and the service required to have TTL1 and TTL2 have consoles for login) <mange>Interesting that you've listed so many things as system packages. Any reason to not install things like git and icecat as user packages, rather than at the system level? <OriansJ>mange: Short version; If there was an ability to define user level packages in the system definition I would have but as I want to be able to just login and know all the tools I require are installed and setup correctly <A87221>mange, I am going to launch the reconfig. // oriansj, thanks for sharing. I can learn from it. <A87221>I don't exactly know what I am doing, of course. (o: <OriansJ>30 seconds after guix allows one to setup user or group packages as part of a system definition, I'll be upgrading my configuration again to strip system level down to the bone <OriansJ>A87221: good, it means you'll learn alot. <A87221>mange, build is running, will take some time. What does "srfi" provide, besides "remove"? <mange>Excellent! Sorry that this was so hard, but I'm glad we managed to get something working for you. <A87221>The compile error was a blocker for me. I am a coder, but I don't know guile and my apt-get-fu is useless here. <mange>Yeah, and it's only a problem on i686-linux, from what I understand. I had no problems building that system locally, but I'm on a x86_64 machine. <A87221>I tend to use i686 in VMs and x64 on metal. <Kurt86>I'm installing guixsd and it reaches a part where I get this error: grub-install: error: /gnu/store/bdzxdpdw25k8v6lz54clz42bilx47srj-grub-2.02/lib/grub/x86_64-efi/modinfo.sh doesn't exist.Please specify --target or --directory. Can someone help me solve it or give me a hand please? <Kurt86>I'm installing on a 250GB drive and I want to use it completely for guixsd. <mange>Hey Kurt86, sorry. I've seen that mentioned on the mailing list before, and from memory it was an efi issue, but let me try to find the thread. <atw>I heard that I can write a build phase like (lambda* (#:key version #:allow-other-keys) ...) but I can't find an example of that in the guix/build directory. Is it possible, or will I have to do something like (let ((version ...)) (package ...)) <atw>I'm doing this because the thing I'm trying to package has a path I'm trying to construct like (string-append "source/Agda-" version "/src/data/lib/prim/Agda/") <mange>atw: The way (package ...) expands will make it so you can use version within it, I think. So (package (version "...") (... (string-append "source/Agda-" version "src/data/lib/print/Agda/") ...) ...) should work. <mange>If you look at most packages, the version is included in the source filename with exactly that pattern. <Kurt86>Mange /sys/firmware/efi exist for you? what is the full path ? thanks in advadced <mange>I'm not on GuixSD at the moment, but it does exist on my GuixSD machine at home. That is the full path. <atw>mange: Thanks! I found my mistake: I previously had (let ((version "...")) (package (version version) ...)) which I wrongly refactored to (package (version version) ...) ☺ <mange>I use Emacs, and before that I used Vim. I haven't seriously used anything else in a long time. <A87221>emacs, vim... you embody the debate! // How to search the package repo for the string "editor"? <OriansJ>They both have their place and their time, use the right tool for the right job <A87221>OriansJ, I jest. But I do like Geany and have useful config files for it. Perhaps I should package it for guixsd to "give back" to the community. <mange>A87221: That would be great! If you need any help with packaging then please ask. :) <A87221>Ahem... I need help, point me in the right direction, sensei. <atw>mange: compilation could start but the execution of my build phase fails because version appears to be bound to a proc: the backtrace includes (string-append "source/Agda-" #<procedure version ()> "…") <mange>atw: Can you show us a paste of your package? That sounds like you have a procedure named version that's messing things up somehow. <mange>I don't really know about Geany and how to build it, so you may have to do some exploration. :) <atw>mange: I may have gotten over that hurdle. I unquoted version. Makes sense that I had to do that because I'm inside a build phase like (package (arguments `(#:phases (modify-phases %standard-phases ...)))) <Digit>potential for reproduceable builds?[/curious-cat] <Digit>oops, thought i was getting a bit verbose for irc again. n_n <roptat>regarding reproducibility, we have ways to check: guix build has a --check and a --rounds option, we have guix challenge... <roptat>we also copy patches from debian sometimes to fix the reproducibility issues <roptat>we also try to work with upstream when we find this kind of problem <kmicu>Hi Digit, Guix and Nix papers explain know-how. Any reproducibility problems solved by Debian, Guix, Nix, etc are beneficial to all of those projects. <felipebalbi><sarcasm> debugging random, hard-to-reproduce kernel bugs in fun <civodul>ACTION just run 'guix pull' in 1m30s <civodul>that's still more than 'apt-get update' but it's getting more acceptable <rk4_>how long does that normally take you? :) <civodul>rk4_: the best case used to be ~4m i think <civodul>it's kinda fun because i wasn't think of this when i made the optimization above <g_bor>sneek: later tell mbakke: I'm on vacation until Sunday. Could you do the wip-java8 merge to staging? <civodul>g_bor: close your IRC client and enjoy your vacations! :-) <snape>welcome back, matrix users :p <OriansJ>thank you snape, we appreciate your support <rekado_>tried to reconfigure berlin.guixsd.org, got an error in /gnu/store/83j0pcbiwr4m68skqry6bp34fxmalx5f-boot-builder <snape>rekado_: please make a backup of /var/run/cuirass/cuirass.db before reconfiguring :-) <rekado_>it contains a procedure, i.e. “#<procedure 36a6120 at gnu/services/shepherd.scm:70:4 (state)>” <davidl>What type of laptop do guix-devs use, that doesn't overheat from compiling all the time? <davidl>I need to put icepacks on my X200 everytime I guix pull, which is starting to get old. <OriansJ>davidl: well, I live in Michigan; so I let nature do the cooling especially in the winter. <OriansJ>One can also underclock their processor to drop the wattage consumed (it just slows it down a litte) and thus reduce the cooling required <civodul>rekado_: did you run 'guix pull' or are you on a local tree? <davidl>OriansJ: wich I could also use nature =) I am using --cores=1 and sometimes cpulimit but I need to drop to very low cpu percent to make sure the laptop doesn't overheat. <civodul>davidl: i have a recent laptop, but these days i often get substitutes for 'guix pull' <davidl>civodul: are you using berlin.guixsd.org ? <davidl>civodul: I have been considering puri.sm laptop with an nvme drive, but those are really expensive. <OriansJ>davidl: the best upgrade for an x200 is an SSD followed by More memory. As it is a core2duo chip, they can easily be undervolted to 15W without losing more than 20% performance. <OriansJ>Although in virtualization, Core2Duo got completely hammered by K8s; despite having higher floating point and integer benchmarks. Not that most people buying Chips at that time really cared much about virtualization then <davidl>OriansJ: thanks for the tip. I actually just replaced to an SSD which helped with overheating to some extent. <davidl>OriansJ: not sure how to make it undervolted. <OriansJ>which is max for the x200; so not much room there; undervolting is generally done in the bios or by direct circuit changes (something you probably want to avoid) <OriansJ>and I don't think core/libre boot projects have added overclocking tools to the mix yet <rk4_>probably obvious: old laptop heat sinks tend to be caked with dust, thus have little cooling <snape>x200 tip: don't close the lid, because it prevents air cooling <davidl>rk4_: I replaced my fans and cooling paste last autumn, but didn't do much difference for compile times. Normal browsing etc isn't a problem. <brendyn>Does anyone know the difference between ffmpegthums and ffmpegthumnailer? <davidl>OriansJ: I am not sure how undervolting works - but I found the PHC-project, which is GPL; would it possibly not work due to core/libre-boot being installed? <JeremiahOrians[m>davidl: no, since the bios setting of voltages can be overridden by the kernel. so yeah it looks like the linux Processor Hardware Control program could be used to undervolt your CPU and drop your thermals <JeremiahOrians[m>just remember to drop clock speed with voltage and small adjustments only. All chips have a unique performance/power curve that later you'll want to map out to optimize your balance of power and performance <davidl>JeremiahOrians[m: ok, thanks for all the tips. I might try this. <civodul>rekado_: i cannot reproduce the issue on my laptop <civodul>that is, boot-builder doesn't have #<procedure ...> <efraim>python[2]-pyqt@4 is pretty much only used by pybitmessage, we can probably drop python-poppler-qt4, poppler-qt4 and python-pyqt4 <brendyn>For an input like libjpeg-turbo, should we call it "libjpeg-turbo" or "libjpeg" ***sturm[m] is now known as sturm
***sturm is now known as Guest14691
<civodul>rekado_: reconfigure complete! i used "env -i ~/.config/guix/current/bin/guix system reconfigure ..." <civodul>it's the story of a one second difference <Guest14691>I'm running some Python code in a virtualenv that uses cffi, which is failing with "OSError: cannot load library libgobject-2.0.so.0". The strace says it's searching in the Guix python, glibc, gcc and libffi packages but this library is actually in the "glib" package. Any tips? <civodul>Guest14691: if i recall correctly, cffi searches .so files in the compiler's search path, no? <civodul>perhaps you can work around it by setting LD_LIBRARY_PATH appropriately? <civodul>rekado_: so commit e4752118691e41ae8307649d1abfd4739b3e4bfa should fix the mtime issue that led 'guix system reconfigure' to fail on berlin <a_e>Hello! I am just trying out a new IRC client. <a_e>Now I could become a gsoc student! <ecbrown>pkill9: isn't the answer subjective? <pkill9>I'm wondering if I could run GuixSD on one of those core 2 duo thinkpads <pkill9>without it taking ages to manage the packages <ecbrown>it will probably depend on CPU, network, and disk speed <pkill9>I'm not including compiling packages, just the core parts of Guix <snape>pkill9: x200 is core 2 duo and *lots* of GuixSD users have it <civodul>pkill9: disk speed plays an important role <pkill9>i would ideally put in a very cheap SSD <pkill9>or possibly even just run it off a usb stick <pkill9>it would be a super cheap travel laptop, purely for accessing hostel websites and very light tasks like that <pkill9>the idea is i wouldn't care if it got stolen (though i'm not sure what thieves would want a laptop that looks straight outta the 90s for) <ecbrown>you'd have to keep it hidden from esp. openbsd hackers <ecbrown>ACTION is interested in whether GuixSD/Hurd is available as an ISO, for hobbyist purposes <snape>pkill9: but still, the SSD is a worthy investment <snape>in terms of performance and robustness <pkill9>yeah i recently upgraded to an SSD, i agree <ecbrown>my current system (i5) i got out of a dumpster on earth day, and a $10 refurbed 250GB disk <ecbrown>actually, i rescued three identical i5's <pkill9>where did you get it out of a dumpster? <ecbrown>i work in a big office building, and apparently i5's are considered "junk" <ecbrown>they pulled the hard disks and threw them out. i schlepped them out to my car, and now i use them for experimenting with OS's including GuixSD and Debian Hurd <pkill9>i would like to find old laptop in office dumpsters <pkill9>but if i don't work there it might be a bit weird <ecbrown>those were there, just not of interest as I wanted a high probabilty of hardware compatibility with libre OS <ecbrown>it was a great experiment, as now i want to incorporate Guix in my actual work. i'm just stunned at how Guix makes this stuff possible, and seems like i haven't slept for days due to excitement. <pkill9>Guix makes what possible ecbrown? <ecbrown>easily containerized, reproducible, sensibly scriptable environments <ecbrown>i had been using pkgsrc esp due to its unprivileged mode, but i think i can do similar. that's where i'm heading, at least <brendyn>could someone on GNOME please tell me what the value of their GIO_EXTRA_MODULES is ? <ecbrown>always seem to find myself on some ancient centos without root, and needing to step it up to something modern <civodul>looks like the db upgrade didn't really take place <snape>did you backup the database? <a_e>Yes, calibre is broken right now. It would be good to try an upgrade to the current version. <brendyn>buenouanq: Yes I've been trying to upgrade calibre but got stuck on an error <snape>civodul: did you update the config? :-) <snape>because if you didn't, there is no way it'll work <snape>for example, the (for-each (lambda (input) (db-add-input db (assq-ref spec #:name) input)) (assq-ref spec #:inputs)) <a_e>brendyn: Which version? What kind of error? <snape>is very likely to return the exact error you have, if there is no input <Guest14691>Is there a way to get `herd` to give you the output from a service it's trying ta start? My mysql-server is failing, but I don't know why. <snape>civodul: yes, it's normal, the db upgrade took place <snape>otherwise it'll use two specs: the one that got upgraded and the one of the current config.scm (which is buggy) <sdb>just filed bug#32222 Workrave is not found by xfce4 run-program dialog nor xfce4-appfinder after install & reboot <snape>Guest14691: no, and that's a bug <snape>the best way is to start it manually <snape>Guest14691: it seems mbakke[m] recently worked on mysql <Guest14691>snape: mysqld seems to be complaining that it can't downgrade, so perhaps we've reverted to an earlier version for some reason. Will try deleting /var/lib/mysql and system reconfiguring <Guest14691>(may not have been a recent change - I rarely use MySQL on this machine) <brendyn>a_e: ImportError: cannot import name QWebView <brendyn>Actually I just tried it with 3.28.0, got the same error but the build completed anyone somehhow <brendyn>a_e: I think it needs python-pyqt to be build with qtwebkit, but it doesn't have it as an input. I tried adding it but then python-pyqt failed to build <efraim>Debian has an alternate version of qtwebkit they use <brendyn>I'd really like to get calibre 2.30 in to guix. What should we do? <snape>from what I've seen calibre is quite a mess. For example it includes it's own python <brendyn>They're intending to maintain python2 after it is abandonded by upstream <mbakke[m]>sturm: Check the mysql error log and try running `mysql_upgrade` if instructed to do so. <brendyn>He said it's too much to convert millions of lines of code to python3 <brendyn>but he also doesn't particularly care to make packaging easy <mbakke[m]>Did the system service change to use MariaDB? <snape>mbakke[m]: mariadb is the default, it seems <Guest14691>mbakke: thanks, I just deleted /var/lib/mysql, ran "sudo mysql_install_db --user=mysql --ldata=/var/lib/mysql" and it seems be working well now. <Guest14691>mbakke: MariaDB has been the system services for quite a while I think <mbakke[m]>Odd that it said it could not downgrade then. <pkill9>are there any alternative ebook managers to calibre? <brendyn>efraim: is it different to our qtwebkit? <efraim>brendyn: it looks newer, but it looks like development stopped on it <Guest14691>yay, I have another one of my Pyton work projects running in Guix! Inching closer to spending the full work day in Guix. :) <civodul>Guest14691: cool, glad that it worked out! <civodul>if you have tips & tricks for Python devs, don't hesitate to share them on help-guix@gnu.org <bavier1>I get a "sendfile: broken pipe" from `guix download` on a local copy of the texlive source tarball <bavier1>`guix download file://...` seems to work on smaller files <bavier1>civodul: yes, probably, I thought something like that might be going on <bavier1>enabling some swap space seems to work around it <bavier1>uff da, but 'guix package -u' still insists on downloading it :( <bavier1>oh wait, it might be downloading a built substitute... <civodul>sneek: later tell snape could you take a look at the spec in guix-maintenance.git, under hydra/modules/sysadmin/services.scm? i'm failing to get it right <sneek>Welcome back snape, you have 1 message. <sneek>snape, civodul says: could you take a look at the spec in guix-maintenance.git, under hydra/modules/sysadmin/services.scm? i'm failing to get it right <mbakke>lfam: I tried to unbundle yassl too from mariadb and it seems to work. One less problem to worry about :-) <sneek>Welcome back mbakke, you have 1 message. <sneek>mbakke, g_bor says: I'm on vacation until Sunday. Could you do the wip-java8 merge to staging? <mbakke>Also, I'm trying to add a "lib" output now. <pkill9>will there be anything about Guix at freenode live in november? <ng0>so when you enter an environment with cups and have cups as a service running and set up before, all functional, AND you have no font-dejavu installed, you will have to waste some minutes to figure out why printing via lpr will complain about a missing font. is this our territory to document? <ng0>And installing dejavu, as suggested elsewhere, does not remove the error <ng0>D [20/Jul/2018:18:01:54 +0000] [Job 68] Set job-printer-state-message to "Ignored bad font \\"monospace\\"", current level=WARN <ng0>apparently enscript would be good to package <ng0>or maybe you have a different idea. I need to make harcopies of around 50 files <ng0>no postscript so far, just plain emails <ng0>where do I find this binary? <ng0>421 pages :O I think I had some encrypted ones in there <snape>sneek: later tell civodul: done :-) You can reconfigure <ng0>I got the first erlang package besides erlang and elixir :) <ng0>I just need to clean it up and check that it actually works <ng0>and then procrastinate over sending it for weeks because I might have no time earlier <ng0>the slow crawl to exploring elixir + phoenix hacking <ng0>I also have elixir-mode ande everything, so when I send patches again there will be many :) <enderby>umm, no matter what I do for 'guix package', I keep getting a message about adding "export GUIX_LOCPATH=$HOME/.guix-profile/lib/locale". I have both glibc-utf8-locales and glibc-locales installed. Lots of my applications are giving me "failed to sh: warning: setlocale: LC_ALL: cannot change locale" <enderby>can anyone help? I've been kind of rolling back and forth between generations and I can't figure out what to do <lfam>You should export GUIX_LOCPATH from ~/.bash_profile, assuming you use Bash. It won't take effect fully until you log out and back in, since ~/.bash_profile is only used when logging in. You can test it without logging out by starting a new login shell: `bash --login` <ng0>alternatively, source .guix-profile/etc/profile in the running shell <lfam>You'll only need one of those packages installed. glibc-utf8-locales is a small set of locales for testing. It happens to include en_US.UTF-8, so if that's what you need, you can get away with using glibc-utf8-locales, which is smaller than glibc-locales. <lfam>enderby`: Did you miss my other messages? <enderby`>lfam: and ng0, thanks. I think I may have gotten it working now ;-) got booted, but I checked the IRC logs for your responses <pkill9>is /etc/profile generated? Can I edit it to set global environment variables? if not, where should i specify them? <Thra11>I managed to fix some of my locale warnings by setting GUIX_LOCPATH, but I still get them occasionally. I think it might be that the build daemons don't have GUIX_LOCPATH set. Is that a possibility? <pkill9>yeah Thra11, are you running guix on a foreign distro or GuixSD? <Thra11>pkill9: foreign. I did the orignal install using the script <pkill9>ah ok, just rerun the daemon with GUIX_LOCPATH set <Thra11>pkill9: Should I point the daemon's GUIX_LOCPATH directly at the store rather than at the locales in my user's profile? <Thra11>Hmmm. looks like the systemd unit file already sets GUIX_LOCPATH to look in root's profile. I guess I just need to install the locales in root's profile <Thra11>If I've run `guix pull` as one user but not as another, does that mean they'll try to install slightly different versions of packages? <pkill9>since i use this machine as a single user, i symlink my ~/.config/guix to /root/.config/guix (and change permissions) <pkill9>i wish the system-installed guix would use the latest root guix-pull by default, and then if a user hasn't got a ~/.config/guix then it uses root's guix-pull <Thra11>If I wanted to guarantee that two users could share packages, presumably I could use `guix pull --commit=<something>` to force them to be in sync. <pkill9>yes, but i would just symlink to root for simplicity so i don't have to update for each user <pkill9>plus it will take time to compute <pkill9>to identify the substitute for the guix pull