IRC channel logs
back to list of logs
<pkill9>is there any difference between guix system build and guix build? <cbaines>pkill9, yes, but it's mostly in the name. One is for building systems, the other packages (or any derivation) <c4droid>Hi, I'm configuring my web service with let's encrypt, I followed documentation at Guix manual, after deploy system, the certificate not generated <PotentialUser-51>Hi all, is it okay to post questions about system configuration here? I'm running into some...unexpected behavior on my new machine. <unmatched-paren>Hmm, qutebrowser is refusing to start. It can't find qtwayland's plugin...
***PotentialUser-51 is now known as johnabs
<abralek>Is it ok to respond with XXXXemail@example.com in CC? will that email come to the ticket? <johnabs>I'm trying to get multiple entries into the (xorg-configuration (extra-config )) config setting, but if I add more than 1 Section entry, my display manager won't load on startup and the system hangs. I'm not sure why this happens, can I post what I'm trying to see if I can get some help? <johnabs>Also, just to make sure, are my messages going through? <johnabs>Thanks, I appreciate the confirmation <johnabs>So, is it valid to do something like this: <johnabs>(set-xorg-configuration (xorg-configuration (extra-config '("Section 1 <johnabs>Because that's basically what I'm doing it, but it only works when Section 1 is present only. If I add another, my display manager won't load. :( <johnabs>Oh, sorry about that, I saw it in the header, but didn't realize that's what it was for. I'll make a post and link it here with my system.scm, is that okay? <johnabs>Also, some advice on getting the substitutes added would be a huge help, I'm pretty sure I followed the manual exactly, but I get a "invalid field specifier" error, at lines 67-72. <davidl>rekado: Thanks for showing me that (jupyterlab package at guix-science). <davidl>Is there any foolproof way to limit the number of cpu's used for all guix processes? <PurpleSym>davidl: Do you mean something like the `--cores` option? <PurpleSym>(Which works for both, the daemon and `guix build`3) <davidl>I tried that earlier but it didn't always work. <davidl>Maybe I should just try again, unless you know of other methods? <davidl>Basically, I don't know if the daemon maybe forks processes that will start using other cores, and maybe that is the issue sometimes. <PurpleSym>By default the daemon only runs one build and passes the `--cores` option to the build system. <PurpleSym>Do you have an example where this does not work? <davidl>PurpleSym: no I don't have one handy. I just know that last time I tried to do this it didn't work. However, if it only applies to builds, then maybe the issues I had before were sometimes with a guix pull. <davidl>or calculating a derivation or similar. <davidl>(anything guix might do which isn't strictly a package build) <PurpleSym>Yeah, afaik it only applies to builds and not to anything guix itself is doing in parallel. But you could use `taskset` for that. <Luk6655>hi, is anyone here using guix shell --container? <Luk6655>It seems to fail for me (running guix on foreign distro) with guix shell: error: mount: mount "/home/luk" on "/tmp/guix-directory.7O87iN//home/luk": Invalid argument <Luk6655>I wonder what is the areguments to mount it is trying to use? <rekado>what’s the full command you issued? <Luk6655>guix shell --container --network --preserve='^DISPLAY$' emacs <rekado>by default it will mount the current work directory into the container <Luk6655>ah, perhaps that's it, I'm running it in my home folder <rekado>or run it in a sub-directory, yes <Luk6655>I need the app(emacs) to be able to see .emacs.d (in the $HOME) or a version of it <rekado>you can “--share” or “--expose” individual directories <Luk6655>so if I run it in a subfolder I'll need to find a way for it to think that's the $HOME <rekado>and you can map the directories wherever you want <rekado>--share=/home/luk/.emacs.d or --share=/home/luk/.emacs.d=/wherever/i/want <Luk6655>cool, yes I can see this in the manual now <rovanion>When I try to reconfigure my system today I get 'No space left on device /boot/efi' even though there is 80MB space left on it and it has never been a problem before. Any ideas? <cbaines>rovanion, what kind of system is this? <cbaines>I think I encoutered this error with a Honeycomb machine, but that is probably quite different since the hardware is different <civodul>rovanion: what does "df -h /boot/efi" say? <rovanion>/dev/nvme0n1p2 95M 16M 80M 17% /boot/efi <ulfvonbelow>has 'emacs-use-package' always depended on some "bind-key" package? <ulfvonbelow>on startup emacs is telling me that use-package can't be activated because it's missing <jpoiret>rovanion: is 80MB sufficient for GRUB? <civodul>my /boot/efi partition (on x86_64) is 20M <ulfvonbelow>hmm, this is truly puzzling, (require 'use-package) succeeds fine, and yet the first line in *Messages* is: Unable to activate package ‘use-package’. Required package ‘bind-key-2.4’ is unavailable <rekado>ulfvonbelow: one thing to check is whether you’re mixing Emacs packages managed with Guix with those downloaded by Emacs. <ulfvonbelow>package-list-packages shows csharp-mode and gnu-apl-mode as installed via emacs, with the rest either 'external' or 'built-in' *rekado adds a few cindex marks to the manual… <rovanion>jpoiret: Has been for more than a year up until now. <rekado>ulfvonbelow: the problem with mixing is that Emacs might load a package from your elpa cache when it should load the package from Guix instead. <florhizome[m]>So the manual states package committing patches should be separated. How do you normally do that when you have a bunch of packages in the same module, open a new branch for each? <rekado>florhizome[m]: I use ./etc/committer.scm <rekado>this is only about creating separate commits; you don’t need to give each of them a name (which is what a branch is) <ulfvonbelow>florhizome[m]: if you're wondering how to actually go about creating the commits, note that you can stage only changes to part of a file. I'm not sure how to do it with 'git add' directly since I use magit, though <rekado>but etc/committer.scm should be able to do that too <florhizome[m]>but if I have changes in the same file, they can’t be strictly independent, can they? <rekado>they don’t need to be independent. They just need to be separate commits.
***wielaard is now known as mjw
<dthompson>I am converting a package recipe to use g-expressions and I am wondering about a best practice: if I refer directly to package objects in the arguments field, and those packages are not needed by the build system otherwise, do I still include them in inputs? <dthompson>they aren't strictly necessary for the package to work as they are already captured in the gexp. however, the various guix UIs that examine the input fields to show dependencies wouldn't be showing the full picture. <dthompson>I'm leaning towards including them in the inputs fields anyway to make the relationship more explicit. <tricon>dthompson: fwiw, i like your latter point. <Kabouik>I am trying to package cobib (python) for Guix but am getting this error: https://0x0.st/oJCR.txt I don't really understand it, seems the installation process is requiring a connection to check things? Is it something that I can fix in my Guix package, or should I contact the developer to ask for an offline installation process?
***jespada_ is now known as jespada
<pkill9>Kabouik: have you added python-future as a (native?) input? i htink it's trying to download 'future' automatically since it can't find it <pkill9>Kabouik: look at the ModuleNotFoundError lines and add the packages they mention <Kabouik>Oh right, thanks. That's the first time the guix import does not work out of the box for pypi programs <Kabouik>Seems to work, I needed pytest and pyte too, thanks pkill9! I'll submit the patch this week-end most likely, will test the built cobib a bit today at work <nashdidan[m]>Hi there. Does anyone know/rember how to make GDM/GNOME unlock the "Default <nashdidan[m]>keychain" when loging in to the desktop everytime? Better, is there a <pkill9>cobib's source page mentions papis <Kabouik>I think one of the differences is cobib allows you to store your files anywhere regardless of where is your database; but I guess also the UI changes (it will change in cobib 4.0, which is in development, but won't be curses-based so I prefer to package one version now in anticipation of the headache that the new version could be to package) <Kabouik>For me the main point is it is a console literature manager, either through CLI or TUI. I didn't want a GUI or Web UI. <Kabouik>In all honesty, I haven't really moved from Zotero yet, so I'm not even too used to the Bibtex/Biblatex/Latex workflow… I had `cobib` installed on my machine for quite some time and really liked the idea, but never really transitionned to it by lack of time, and also because cobib still has no support for downloading PDFs through a proxy (this is planned). But seeing that Zotero is not packaged for Guix, I figured it might be a good time to package <Kabouik> cobib and try it again, with a little more will this time. <codonell>civodul, Ping? Around? I'm reviewing old glibc patches in patchwork (I'm driving an SLI on patch review based on patch time in a non terminal state). <rekado>FYI: On Wednesday, October 12th, 2022, there will be maintenance work on the MDC firewall between 05:00 am and 09:00 am (UTC+2). <codonell>civodul, It applies. Works as expected. I tested it on build-many-glibcs (which auto sets OBJDUMP). <Kabouik>Anyone got success using `pass tomb`'s timer feature? I think it depends on systemd-run, but asking just in case. *civodul sends message about a release
***califax_ is now known as califax
<apteryx>rekado: how will that impact the Guix services from the outside? <rekado>apteryx: possible interruptions to *.guix.gnu.org <nashdidan[m]>Is anyone here using doomemacs and is able to use guix.el ? I’m running into problems when I install the package from doom, where a Guix repl fails to start. When installed using guix itself it does work. <lilyp>in which constellation does doom affect your guix repl? <lilyp>if anything, it should only bork emacs <rekado>bleh, I’m getting “In procedure load-thunk-from-memory: incompatible bytecode version” again… <rekado>completely rebuilt the branch with “make clean-go && make -j” in a fresh “guix shell --pure -D guix” <rekado>turns out I need to “rm guile && make” <civodul>we may be able to remove our custom 'guile' binary <rekado>I’ll add (properties `((upstream-name . "…"))) to all packages in tex.scm <rekado>this should make automatic checking easier <apteryx>rekado: perhaps this should be sent to guix-devel then, so that more interested parties can take note <apteryx>civodul: what is this custom 'guile' binary needed for, anyway?
***mark_ is now known as mjw
<pkal>Is it just me or does the default vi on a guix system always crash when entering a non-ASCII charachter (eg. umlaute). <PotentialUser-91>hi all, how to install new version of package? I used guix build --with-source=... but then i realized that i needed to use the expr, how make it? or there may be a more elegant solution? <pkal>lilyp: Then I wonder what is wrong. I have considered proposing a patch that would replace the current implementation with the heirloom impelmentation.
***maximed is now known as antipode
<antipode>Are you using an UTF-8 locale, and if so, did you set up GUIX_LOCPATH? <pkal>antipode: I don't have access to a machine with guix right now, so I can't test it <pkal>unmatched-paren: The last commit was made earlier this year. <civodul>apteryx: see comment in guile-launcher.c: it has 2 purposes, one is locales, the other is reducing the stat storm <fiesh>`guix gc` gives me a `guix gc: error: reading directory `/gnu/store/trash/mbnpbiqziy3ir7v6jq3926glqgchy224-slang-2.3.2/share/doc/slsh/html': Input/output error` -- but a btrfs scrub runs through fine... any hints? <fiesh>(this is my first btrfs installation ever, can't say I'm thrilled, brand new nvme...) <antipode>fiesh: What does "/gnu/store/trash/mbnpbiqziy3ir7v6jq3926glqgchy224-slang-2.3.2/share/doc/slsh/html'" say? <antipode>("btrfs scrub" can be run while the device is still mounted, but with as cost being less thorough) <fiesh>antipode: I can list slsh/, but not (the directory) slsh/html <antipode>fiesh: Oops, I meant What does "stat /gnu/store/trash/mbnpbiqziy3ir7v6jq3926glqgchy224-slang-2.3.2/share/doc/slsh/html" say. <fiesh>antipode: looks normal except for access and modify being the unix epoch <antipode>However, this does not answer my question. <antipode>(I was wondering if it somehow was a symlink to /dev/something or such) <antipode>fiesh: If you give the actual output of "stat", that might help with determining what went wrong. <antipode>Except for the Inode, Change and Birth, I get exactly the same. <antipode>fiesh: I'm wondering, are you using RAID5 or RAID6 with btrfs? <fiesh>I guess I'll go to single user mode and fsck it... <antipode>Maybe "dmesg" has information on I/O errors <fiesh>valid point! BTRFS error (device nvme0n1p2): bad tree block start, want 36029579264 have 12438384300292430983 <antipode>Maybe, just maybe, the disk was temporarily disconnected or such and the I/O error was cached. <fiesh>hmm it's a laptop and the device is installed in it, and it wasn't dropped or any other physical shock <antipode>fiesh: I don't know what that error means but sounds useful information <fiesh>so... how do you get to single user mode? <ae_chep>So I know this isn't a scheme channel but I heard people talk about how scheme is their go-to for scripting. Do you not need to execute other programs? Pipe their output to other programs? Is it not cumbersome to do so in scheme? Are there utilities that make this easy enough? <rekado>kaelyn: it looks good to me, but I was hoping for someone who uses weston to chime in <rekado>I’m just a little worried that wayland-protocols-next is used – will this lead to problems with other wayland tools that are still using wayland-protocols? <kaelyn>rekado: No worries! I had a feeling that may be the case. I'm not sure what the motivation for creating wayland-protocols-next was, but the newer weston didn't compile with the old wayland-protocols, and I wouldn't expect the updated version to cause communication issues with clients on the older version (otherwise it isn't a very good protocol for one intended as X11's successor)
***Miro69 is now known as Miro[fsk]
<rekado>kaelyn: thanks for having tested it. I’ll apply the patch later tonight. <kaelyn>Thank you! I also just tested running foot in the new weston to double-check, and it seems to work okay. <two[m]>how do i expose /dev to a container?
***Dynom_ is now known as Guest29
<johnabs>Hi all, I'm getting a funky error when I try to reconfig using multiple entries in the extra-config of the xorg-config section in my system.scm file. Basically, it breaks my SLiM (or whatever default dm starts up when you install guix from the graphical installer) and my computer just hangs on a black screen then I have to power it down. Entry 1 in <johnabs>the list works fine, but adding either of the other two entries breaks it. Can someone check this paste and see if I put the entries in correctly? As an aside there are some commented lines toward the bottom where I can't add a substitute server, it says I have an invalid field specifier, even though I followed the manual/channel guide. Here's a <efraim>you're not on the 5.19.12 kernel, are you? <rekado>what’s the invalid field specifier it mentions? <johnabs>I'm also on a 12th gen framework laptop if that helps <efraim>extra-config can be a list of strings or objects, so nothing obviously wrong there <johnabs>I put the code for the substitutes server right after the %desktop-services) command, as described in the manual. But the error message isn't really helpful. It basically just says this: (services ..... %default-authorized-guix-keys)))))): invalid field specifier <johnabs>But if I take the mirror out, it at least "compiles". But it still breaks xorg/SLiM/whatever it is when I add more to the conf <efraim>no, signing-key.pub has the right number of parentheses <efraim>I would try taking out the device section or the inputclass section and see if that fixes it. perhaps something is named wrong? <efraim>I wonder if xorg has a utility to test an xorg.conf to see if it's correct <ppw>is it possible to switch to a binary-only installation? <johnabs>Do you mean the second and third entries in the extra-config form? I need the first one because the default uxa on the accel method makes my system unusable q.q <johnabs>Also, is the (modify-services ...) form in the right place at least? I'm pretty sure it is <efraim>johnabs: it looks like you can swap %desktop-services for (modify-services ..., but you'll need 2 more close-parenthesis at the end <johnabs>Sorry, you kind of lost me here; do you mean remove the line %desktop-services) entirely <johnabs>At least for the substitute server, thanks! <efraim>looks like you're missing a bootloader and file-systems though <johnabs>Now, if I can just get my backlight and touchpad working. Oh, yeah I just removed those from the paste since they're working fine :3 <johnabs>trying a reboot, wish me luck lol, but i'm keeping that xorg-config with just the first entry so I suspect it should be fine. <johnabs>Well that part worked! So it's definitely the (extra-config) extra options that are causing that issue <johnabs>Hi, sorry I'm going to be chatting in from matrix now, thanks for your help efraim! I hope I can get this other issue figured out soon, but i'll be the matrix johnabs now :3 <ppw>unmatched-paren: can I ask why the decision to have everything built from source was made? <johnabs[m]>I might be able to answer that a bit, but basically it ensures the software you're running is actually the software on the servers, and no jiggery-pokery is going on. Also, it gives you greater control over the software if you want to configure the packages yourself, same reason why people run gentoo. Of course, if you don't want to build from source, you can always use the substitute servers and grab the precompiled binaries from there. <Kabouik>Is `swapon -a` supposed to work with swap files? I use a swap file because after discussion here this was the best/easiest way to enable hibernation on an encrypted Guix and btrfs file system <apteryx>Since I started using GDM, my ratpoison sometimes gets locked up, not answering to input anymore <apteryx>has anyone else noticed such problems? <Kabouik>Unfortunately swapon fails to restore my swapfile I think. I use swapoff in the first place because of pass & pass tomb, as using them while swap is active may pose security risks. <civodul>i said goodbye to the ratpoison a year or so ago :-) <civodul>Kabouik: i use a swap file on my encrypted root, but i would assume that this won't allow hibernation
***justHaunted is now known as _justHaunted
<civodul>feels spooky when you first hear about it, but really, it's good :-) <Kabouik>It should civodul. TopExpert helped me configuring my Guix system to make it work (using btrfs added a little bit of work).
***_justHaunted is now known as justHaunted
<apteryx>hm. I'm a bit worried of making emacs the process holding all my other processes, as I make Emacs hang recularly (with processes, large buffers, or what not) <Kabouik>Weirdly swapon -a doesn't return any error, but I notice that `free -m` shows only zeros for the swap after I swapoff -a/swapon -a, and hibernation stops working. I don't think I had this issue with a swap partition (but using a partition had downsides) <apteryx>using Geiser with Guix seems to cause it to use vast amounts of memory over time for example <Kabouik>I guess I just needed to point swapon to the exact path of my swapfile, somehow -a doesn't find it <Kabouik>I might need to add it to /etc/fstab but I have no idea how to do that on Guix <Kabouik>Oh, I might. Is it just (file-system (mount-point "/swap/swapfile") (type "swap))? <Kabouik>Actually I already have somethign similar in my config (see the last lines here: https://0x0.st/oyMs.scm), but swapon still needs me to pass the absolute path to the swapfile to work, presumably because the swapfile is not listed in /etc/fstab <Franciman>when using guix shell --container, is it possible to have some services started inside the container? <lechner>Hi, what is the proper remedy for configure:13347: error: possibly undefined macro: AC_LIB_LINKFLAGS_FROM_LIBS when using Guile Autoconf macros, please? Adding 'gettext' to native-inputs does not seem to help https://github.com/wingo/fibers/issues/17 <rekado>the PDF manuals for most languages cannot be built <rekado>/gnu/store/j2wxmjkgj10vgi6vcidch4jv6mipgsw6-texinfo-manual-source/guix.de.texi:21981: TeX capacity exceeded, sorry [input stack size=5000]. <ppw>thanks johnabs[m], for the explanation. <two[m]>what is "warning: SQLite database is busy"?