<lfam>Yes, I get substitutes for at least one of those private packages <lfam>Namely, java-org-ow2-parent-pom-1.3 <lfam>According to `guix weather`, mesa has failed for i686 <cbaines>this is where we could really benefit from tracking build metadata <lfam>I wonder who is using i686... <cbaines>then it might be possible to easily determine why it builds on some machines and not others <lfam>I'm building i686 mesa locally and on berlin <lfam>This will probably take a while <lfam>Rust has also failed for i686 <cbaines>I think Rust not working on i686 is a long standing issue <cbaines>There's at least one issue open about it, I think upgrading mrustc was mentioned as maybe being helpful <lfam>If it has never worked, we should disable it on that platform <cbaines>mothacehe was looking at that, and did try something <cbaines>I'm not sure what effect it had thougth <lfam>I've sent reports for each platform to guix-devel <lfam>It would be great to get some help! Especially for the more niche platforms like armhf and i686 <raghavgururajan>Hmm. I am getting "c++: fatal error: Terminated signal terminated program cc1plus" during the build. Could it be caused by memory hit? <lfam>You probably ran out of memory <Ikosit>Is there a command-not-found utility for guix? <lfam>Ikosit: Would that give advice about which package to install? <Ikosit>Yes, that's what im searching for <lfam>I don't believe so... (could have missed it) <lfam>I might be able to help if there is a specific command you are missin ***iyzsong-- is now known as iyzsong-w
<Ikosit>Thanks, but that was more of a general question. I just dislike to search the internet to find the package of a command <lfam>jmarsden: It should be on there somewhere, in a copy of the Guix source code, but I don't remember off-hand where to find it <lfam>It would be nice if the "message of the day" would say where to find it <jmarsden>OK... that should work, thanks. It might be good if the file was added to the image as /etc/config.scm or similar, so it is easier to make changes to the system. <lfam>Yes, especially since the MOTD says it will be at /etc/config.scm <jmarsden>lfam: OK, many thanks. Glad it wasn't just me missing something "obvious". <lfam>It sounds like you tried the obvious :) We should improve it <lfam>I will poke around in the image and see what we can do <lfam>Thank you for pointing it out <lfam>jmarsden: It can be found at '/run/current-system/configuration.scm', or similar at '/run/booted-system/...' <lfam>The distinction only mattering between running `guix system reconfigure ...` and rebooting <lfam>The MOTD is not seen because the image provides a graphical system, so the console is not shown <lfam>I'll see about making sure a copy ends up at /etc/config.scm <PotentialUser-52>How can I create a rust package that inherits from a newer version without the #:cargo-inputs argument? <lfam>What do you mean by "without the #:cargo-inputs argument"? Like, you want to not inherit that part? <PotentialUser-52>Inputs were added for the new version, but are not needed for the older version. <lfam>Untested, but you could try adding (arguments '()) to the package you are writing. It would clear the arguments of the inherited package <lfam>I grepped, and this is done a few times in crates-io.scm <lfam>jmarsden: I'm curious, when getting started with the QEMU image, did you follow any set of instructions? <lfam>jmarsden: I wonder if the easiest thing would be to mention /run/current-system/configuration.scm in those instructions <lfam>Oh, I checked, and the manual section Running Guix in a Virtual Machine does mention it <lfam>The only further improvement would be to make a copy appear at /etc/config.scm <jmarsden>lfam: I didn't follow the instructions... I saw an online review suggesting the VM image as a good way to get started with GUIX and gave ita go. I'd seen mention of /etc/config.scm in the official docs(I think in the Reference Manual),so when I wanted to edit the system config, that is what I looked for. <lfam>jmarsden: Makes sense. We should change it so it does what you expected ***Guest48965 is now known as txgvnn
*apteryx merged staging into core-updates ***iyzsong-- is now known as iyzsong-w
<raghavgururajan>building of `/gnu/store/dv8r55j7k2yrh8byl5j2z73yzlvil332-tdesktop-2.5.1.drv' timed out after 3600 seconds of silence <lfam>raghavgururajan: You could try `guix build /gnu/store/dv8r55j7k2yrh8byl5j2z73yzlvil332-tdesktop-2.5.1.drv --max-silent-time=7200` or even longer <lfam>You should also set --timeout <efraim>lfam: certbot built just fine on aarch64 hardware on staging <sneek>Welcome back efraim, you have 1 message! <sneek>efraim, lfam says: Can you try building certbot from the staging branch on your aarch64 hardware? <leoprikler>how do I compile Guix with all of its optional dependencies? Whenever I run make, I end up getting unsupported bytecode in the next pre-inst-env <civodul>leoprikler: that's because Guile 3.0.5 bumped the bytecode version <leoprikler>I think I'm incorrectly setting up my environment <civodul>maybe you need to clean them another time? ;-) <civodul>it surely has to do with 3.0.5/3.0.4 anyway <civodul>3.0.5 can read bytecode created by earlier revisions, but not the other way around <efraim>i went with 'git clean -dfx' for this upgrade <efraim>probably have to clear $HOME/.cache/guile also *civodul wonders if someone went with "rm -rf ~" <civodul>but yeah, these bytecode incompatibility issues can be quite unsettling *efraim ran rm -rf .config yesterday by accident :/ <Ikosit>I want to package pulseeffects, which includes the c++ filesystem standard library in a way that is incompatible with gcc before version 8. meson-build-system/gnu-build-system use gcc 7. D: Should i fix the pulseeffects, or should i redefine the build-system? <civodul>Ikosit: hi! maybe you can just add "gcc-10" (say) as a native input <civodul>alternatively, you can call package-with-c-toolchain to properly force use of a newer toolchain for your package <Ikosit>Ah nice, i already tried to include gcc-toolchain, but that didn't work, it still chose gcc 7, but the second suggestion sounds promising <PotentialUser-76>Why do I never connect to shadowsock? Guix System seems to have a major problem with shadowsocks... <iyzsong-w>PotentialUser-76: the ERROR said missing password or port_password in your s.json file <iyzsong-w>if the s.json file work with other distro's shadowsocks (same version), then it maybe a bug of guix's shadowsocks <iyzsong-w>also it feel a bit rude to direct general questions to others... <PotentialUser-76>izysong-w. I do not know much about shadowsocks, I extracted this config from v2rayng! <PotentialUser-76>izysong-w, I do not know much about shadowsocks, I extracted this config from v2rayng! <PotentialUser-76>iyzsong-w, I do not know much about shadowsocks, I extracted this config from v2rayng! <leoprikler>Ikosit: when doing native inputs you should do ("gcc" ,gcc-VERSION), because internally we don't use toolchain ;) <ekaitz[m]>hi all yesterday I managed to create a user-service for redshift, as you guys suggested, following the post by efraim flashner. So first: thank you all. <ekaitz[m]>I have a final question for all this, what's the best way to launch the shepherd user instance? <leoprikler>I registered it in GNOME autostart, but launching it from .profile should also work. <leoprikler>You could also do it from those xinit scripts if you're using a lightweight WM like ratpoison. <PotentialUser-71>Since shadowsocks do not work on Guix, I urge Guix developers to fix this. Thanks <ekaitz[m]>leoprikler: I'm on i3, running it on i3's config is ok too? <ekaitz[m]><leoprikler "Yep, i3's startup files should a"> yay great thanks <civodul>PotentialUser-71: please report a bug by emailing all the necessary info to bug-guix@gnu.org so that Guix developers know where to start and don't forget about it <ekaitz[m]>leoprikler: I confirm it works if I launch shepherd using `exec` in i3's configuration! thank you very much! <zimoun>civodul: you have tried to update to guile-json 4.5.1, right? Well, definitively, the test suite needs a bit more love. :-) Anyway, does it make sense to downgrade guile-json? While the fix of everything. <ekaitz[m]>hi, I does anyone have an example of configuration for libinput? can I change libinput properties of a specific device in the system configuration? I'm reading xorg-configuration but I'm not sure if this is the best way to handle it <ekaitz[m]>at the moment I'm setting my config in the startup using `xinput --set-prop`, but having it in the system configuration would be great <ekaitz[m]>let me rephrase and do the question in a different way: how can I add a xorg.conf.d/ file? <leoprikler>assuming that configurations is read from etc you could try etc-service-type <ekaitz[m]><ekaitz[m] "let me rephrase and do the quest"> a package that produces the X11/xorg.conf.d/XX-NAME.conf file is a possible solution? <leoprikler>that could also work, but the question is where it is put <leoprikler>if you just need some lines in your xorg configuration, you can also use the extra-config field of xorg-configuration <ekaitz[m]><leoprikler "if you just need some lines in y"> that's what I thought, but I prefer to add a separate file with the configuration of the touchpad <ekaitz[m]>if i add it to extra-config, where is it wrote? in xorg.conf itself? <leoprikler>I don't think that pattern is too well supported in guix, but if Xorg supports embedding file names, you could write it to a file-like-object and then put that into your extra-config somehow <ekaitz[m]>i'm kind of surprised now because I don't have anything currently in /etc/X11 folder <ekaitz[m]>making a new file /etc/X11/xorg.conf.d/30-touchpad.conf is going to work? hmmm <leoprikler>guix doesn't necessarily put files where you'd expect them <leoprikler>conversely, putting files into such locations doesn't always do what you'd expect either <ekaitz[m]>the simplest way then might be to add a system-wide package that places a file in /X11/xorg.conf.d/XX-whatever.conf right? <abcdw>hi guix! Need a small hint. I was working on adding ability to use OBS plugins installed in the profile. For that I had to patch obs itself (add support for OBS_PLUGINS_PATH environment variable), set correct native-search-paths and properly build plugins. Is it a good idea to apply this patch to the package and remove it once the change accepted by upstream? <leoprikler>ekaitz[m]: again, not sure if that's how xorg rolls, normally you'd have to add such things to a service and xorg-configuration does not seem to support your use case <leoprikler>abcdw: Yes, but be prepared to answer questions w.r.t. your patch coming both from Guix and from upstream. <ekaitz[m]><leoprikler "ekaitz: again, not sure if that'"> understood! thanks for your help! I'll try to make some tricks here and there and see what happens! <ekaitz[m]>leoprikler: I tried to make it work with etc-service but it says etc/X11 exists (which is true) but it's empty, how can I make it fill the directory? is it possible? <civodul>mroh: re the Xfce upgrade, did you test it in a VM (or bare metal)? <mothacehe>cbaines: I just sent a few improvements to the PostgreSQL service :) <mothacehe>after that, I'd like to be able to create roles and databases from the service definition. I saw that NixOS was doing about the same thing. <civodul>mothacehe: congrats on the switch to postgres, that's quite something! <mothacehe>thanks, took me some time, but I think it was worth it :) <mothacehe>it seems to perform really well and the monitoring possibilities are way superior! <civodul>yesterday we noticed that build logs from Cuirass were interleaved: you see the log of the dependencies mixed with the log of the thing you're looking at <civodul>is that related to remote build support? <mothacehe>yes probably. I'm now streaming the output of "build-derivation" to Cuirass from all remote workers. <mothacehe>it allows to follow the build process in live from the web interface <mothacehe>it's also handy when the build failed because some dependency cannot be built. Previously there were no logs in that case. <civodul>mothacehe: do you think the "@ build-log" trace mechanism could be preserved somehow? <civodul>that'd allow recipients do "route" log excerpts to the right place <civodul>but maybe that's too complex and not worth it <mothacehe>yes it would be a nice follow-up, I'll keep that in mind <civodul>PotentialUser-63: hi! you can drop it in ~/.local/share/fonts and then run "fc-cache -fv" <PotentialUser-63>civodul: I did this, but after that, the browser font automatically changed to this font! <Ikosit>Is there a way to set environment variables for programs in their package description? <civodul>it may be that you have to restart applications <civodul>Ikosit: for search path environment variables, you can do so via 'native-search-paths' <PotentialUser-63>civodul: Then I tried to copy the fonts (/run/current-system/profile/share/fonts) to this folder to solve the problem, but they are in (Read-Only) mode, and they can not be changedm and now I have "Read-Only filesystem" error. <civodul>exactly, you cannot change this directory <civodul>if the font is packaged, you can also just run "guix install font-whatever-its-called" <civodul>PotentialUser-63: ~/.local/share/fonts works for me, so to me that is the solution :-) <PotentialUser-63>civodul: Oh! I understood... New font affects IceCat, and we have to change the IceCat font manually. It is good? Is it better to make changes to IceCat for this issue? <civodul>i don't know actually; i'd expect the new font to become accessible to IceCat, but not necessarily to be in use right away <Ikosit>civodul: Is search-paths used at runtime or build time? <leoprikler>Ikosit: native-search-paths are expanded in all environments, search-paths during cross-compilation only. <apteryx>PotentialUser-63: It's sometimes tempting, but please try not pushing your debugging requests onto people :-). In my experience, volunteers are more likely to offer help if you openly ask the question. <apteryx>PotentialUser-63: could it be that Icecat 'prefers' some fonts if found to others? <apteryx>in the advanced fonts preferences of Icecat, there's a box: 'Allow pages to choose their own fonts, instead of your selections above'. If untick, perhaps it'd cause your preferred font to be honored? *apteryx has it configured to Deja Vu <leoprikler>PotentialUser-63: Local fonts (~/.fonts, ~/.local/share/fonts) take precedence over what you have installed in your Guix profiles always. That's just how fontconfig seems to roll. <apteryx>PotentialUser-63: what is the problem exactly? I see glyphs? <apteryx>ah, I see, the text should show as English but it's showing as Old Persian after you 'installed' that font. <apteryx>Do you have deja vu installed in your user profile? IIRC fontconfig is supposed to fall-back to it, but maybe worth trying. *abcdw sent changes to obs upstream, maybe will just wait few days for merge. In that case it won't require to maintain a separate patch. <Ikosit>Am i just stupid or is %build-system really only bound in the `arguments` slot of `package`? <apteryx>Ikosit: that's how it is. Otherwise the records field themselves establish bindings (e.g., you can use 'version' in later fields) <nij>Hello folks :-). I'm trying to navigate in the source code of Guix using Geiser. Currently, I have an emacs buffer with guix/build-system/gnu.scm. I have added guix to geiser's load path by 'geiser-add-to-load-path. But while I 'geiser-eval-region-and-go, it gives an error: https://bpa.st/HDMQ . <nij>Question: How to correctly load a scheme file in a project into Geiser? <abcdw>nij: maybe you need to compile file first with C-c C-k ***Guest364` is now known as daviid
<nij>compiled @@ it says "no code for module (guix config)" <nij>I ran ./bootstrap, which successfully creates ./configure . <nij>But while running ./configure, it asks me to unstall GUILE-GIT. *civodul fixes phase return value handling in core-updates <civodul>now going for a loooong rebuild to try things out <civodul>also getting rid of repack for non-tarball sources <apteryx>civodul: are you including changes from 45773 in the mix? They'll conflict. <nij>Folks, what is your workflow like while looking into the source code of guix? <nij>Using the browser certainly isn't the best way. Frequently I need to quickly jump to the definition of some symbol, and jump back. <civodul>do we really want to support snippets for things that are neither tarballs nor directories? <jgart[m]><nij "Using the browser certainly isn'"> one really easy way to get started with something that is way better than searching in the browser/git web interface is by using ripgrep in a clone of the guix repository <jgart[m]>then when ripgrep shows you the line number and file name for the match you can open it with your editor by using the + option for jumping straight to that line where the match lies <nij>jgart[m]: That's usually how you search? I'm actually thinking of M-. M-, as in SLIME. <nij>Can I make this work even for a repl (geiser, in my case)? <jgart[m]>I just use ripgrep with vis editor. I haven't set up slime. I think ryanprior uses ripgrep integrated into emacs (spacemacs?) to search a code base <mroh>civodul: I did test the xfce update in vm's not bare metal. Thanks for testing, reviewing and pushing! <civodul>mroh: yw! it builds surprisingly quickly <nij>jgart[m]: That's encouraging to know! Maybe I should do that too.. <jgart[m]>I usually don't do recursive searches across files from a repl except for history but maybe I should be doing that at some point. <nij>and stop thinking about other ways :( <jgart[m]>well don't look at me as a role model, ask what civodul, rekado, ... do and others haha ;) <nij>Hopefully understanding guix inside out will teach me a lot. <nij>There are many things to learn though.... <jgart[m]>nij, do you have matrix? I'm hosting a guix packaging meetup in a few weeks to work on packages in a group setting <nij>! I didn't even know what matrix is. <jgart[m]>it's like the free software federated slack <jgart[m]>you can bridge to the matrix channel from xmpp also if you happen to be on xmpp <nij>Thanks for inviting. <nij>Things I have to learn sort of branch out again. <nij>I will take some other time to see how to set matrix up. <civodul>jgart[m]: you should let guix-devel know about the meeting, in case there are people from your area lurking <jgart[m]>civodul: it will be over the internet. I will let guix-devel know also. Thank you for making that suggestion <abcdw>nij: I use run-geiser, C-c C-k to compile current file and M-. M-, works for me. Not perfectly, but works. Sometimes it jumps to export declaration instead of later definition. <nij>abcdw: Have you tried that with the guix source code? <nij>It doesn't work for me right away.. <nij>I suspect I need more knowledge about autotool T_T <apteryx>abcdw: yeah, Geiser is not very helpful here as well... I mostly have to navigate using grep 'ic package-name' to find the packages. <apteryx>Otherwise guix-edit from emacs-guix can be of help too, I think (it was broken for a while, but I understand it's been fixed). <apteryx>civodul: it's big mostly for test code <apteryx>civodul: about snippets on single files; why not? it'd be inconsistent not to. <abcdw>nij: Yep, guix/.dir-locals.el populates load-path automatically and I only need to compile a file I'm working with. For other projects I do C-c C-e l to add project root to load-path. <civodul>i guess it depends on how frequent it is <nij>abcdw: Didn't your system complain that guix/config.scm is missing? <abcdw>nij: nop, it doesn't complain. I'll try to do all the steps on clean guix repo. <nij>jgart[m]: OH that's VERY helpful!! <nij>Hmm.. but it has its limit too. For example, while looking at <nij>(define gnu-build-system (build-system (name 'gnu) [...] <nij>I tried to find the source of 'build-system, but gd returns a huge list. I narrowed down by querying "define", but there's not place that defined 'build-system.. <nij>jgart[m]: thanks ;) I will check that dotfiles out. <abcdw>nij: Cloned latest guix to a separate directory, C-x C-f ~/tmp/guix M-x run-geiser guile C-x C-f gnu/packages/video.scm C-c C-k C-s cmake-build RET M-. M-, works fine <jgart[m]>nij: see line 45 of guix/build-system.scm <nij>abcdw: Followed your procedure. It says : <nij>no code for module (guix licenses) <abcdw>nij: Did it poped a message about .dir-locals? <nij>Yes. Should I say yes @@? <abcdw>After that it should save configuration to the custom section of your init.el <nij>ok.. lemme do it again. <nij>Same.. "no code for module (guix config)"" <nij>I failed upon hitting "C-c C-k" <nij>abcdw: Did you get the source from savannah? <nij>Instead, there's only guix/config.scm.in <abcdw>nij: config.scm not needed at "runtime" it's probably a configuration for build daemon. <nij>Right, but while you C-c C-k in video.scm, <nij>doesn't it try to fetch (guix config)? <nij>That's the error message I got :( <abcdw>Have you started a fresh emacs with new geiser repl? <nij>no! lemme try. see ya <kevinfish>Hi everyone. I'm trying to install a dual boot with windows 10 btrfs root GuixSD and I have a catch 22 I can't figure out how to resolve: When I go to generate the system it says that grub-install didn't work (I can't remember exactly why and I was in console mode so I couldn't copy/paste the error but the gist of it was that it was because I boo <kevinfish>ted in bios mode). I had to go into bios and disable my legacy boot and windows still boots but I can't boot the GuixSD USB ISO. It's an HP laptop and when I hit ESC during boot and select boot options it walks me thru the EFI boot files and I picked both the boot files and it just returns to the boot menu. I also tried hard resetting my lapto <kevinfish>p (unplugging it, pulling out the battery and holding the power button down for a while) plus booting windows into recovery mode and letting it try and fix the boot (which probably only did the windows one as that was the only one that showed up) and still nothing. How do I build an EFI boot if I can't boot GuixSD ISO as EFI? <nij>It complains about missing (guix config) again. <nij>I `,q` and moved back to the 0-th layer. Switched back to the buffer, and C-c C-k again. <nij>Then it complains about "Unbound variable: <derivation>" <nij>I `,q` again .. and so on. ***andre is now known as Guest86477
***Guest86477 is now known as mzbm1
<apteryx>nij sounds like you haven't configured your checkout (guix environment guix; ./configure --localstatedir=/var && make) <apteryx>Geiser also works best when the modules are already compiled (otherwise everything comes to grinding halt). <davidl>By the way I have a Nextcloud-desktop app that builds and kind of works in that you can login to an NC server and sync files and receive notifications via I think libnotify, however - the GUI windows don't appear. <apteryx>with the advent of poetry, flit and what not <mzbm>hello, is anyone here using jetbrains products on guix? <apteryx>it falls back to setuptools but otherwise implements a pep517 builder <PotentialUser-28>Is it possible to upgrade pip packages in Guix? I upgraded two pip packages; after the "pip list outdated" command, packages are still old! <apteryx>that's because Guix makes use of PYTHONPATH, which takes precendence over pip packages <mdevos>Is it possible for the for the channel gexp in ‘unattended-upgrade-configuration’ to import additional (non-standard) modules? ‘with-imported-modules’ doesn't seem to have any effect <apteryx>PotentialUser-28: you could work in a guix environment --pure which doesn't include these packages, or better, update their definitions in Guix :-) <apteryx>Python uses many locations (called sites) to find its libraries. PYTHONPATH affects sys.path directly, even before site.py gets called IIRC. So the libraries discovered via PYTHONPATH take precence over any other locations (including the locations used by pip to install packages) <joshuaBPMan>really off topic, has anyone looked into the zig programming language? Does anyone else see it as a viable C alternative? <mdevos>mdevos: I think I know what's goin on ... by default, module-closure filters out modules that aren't named (gnu ...) or (guix ...). I'll override that <rekado>apteryx: is there another issue in the bug tracker where GUIX_TEXMF is implemented? Or is this already the case? <jgart[m]>joshuaBPMan: I haven't checked out zig except for the wikipedia entry and maybe the homepage of the project. I'd like to at some point <joshuaBPMan>jgart: the youtube video the road to 1.0 is pretty interesting. <apteryx>rekado: it should be in this one! perhaps mumi is playing tricks on my patches again? :-) <apteryx>I don't know why but I seem to trigger that MUMI problem on almost every git send-email submission <rekado>apteryx: hmm, I guess I’ll have to investigate at some point, but … I really don’t have the stamina or focus for debugging debbugs these days. <apteryx>is there a way to teach Emacs' C-x l g to follow file renames in the git history? <apteryx>I was trying to move definitions to a new module, and the biggest challenge is keeping the copyright attributions intact <jgart[m]><joshuaBPMan "jgart: the youtube video the ro"> cool! I'll check it <mdevos>‘sudo /gnu/store/...-unattended-upgrade’ is busy grafting builds, I assume that means I've configured unattended upgrades correctly. Next step (for later): offloading, and preferably letting the other machine build in advance. Thank you whoever wrote the unattented upgrade code! <Rovanion>There's no golang package importer right? <mdevos>Is /etc/bashrc supposed to be world-unreadable, or did something weird happen on my system? <lfam>Rovanion: Please test and provide feedback :) <apteryx>rekado: I meant C-x v l (vc-print-log) or C-x v g (vc-annotate), two Emacs functions <lfam>mdevos: That doesn't sound right <mdevos>No idea what happend there, let's just ‘chmod o+r’ for now (‘stat’ indicates only 1 link, so should be safe). <jetomit>mdevos: I've had my user's umask (077) mess up permissions on reconfigure with sudo -E <mdevos>jetomit: that might have something to do with it, I've experimented with the umask lately, and ran the /gnu/store/*-unattended-upgrade manually with sudo <Rovanion>lfam: Is there an easy way to figure out the latest version of the patch/branch? I'm used to working with branches in git repos. I tried downloading the patch in the next to last email but it wont apply :/ <lfam>mdevos: Recommended not to do that :) <lfam>Rovanion: Somebody else tried it recently. I think it was in the past couple weeks. If you search the mailing lists for discussion (try searching for the bug number) you might find more recent versions of the patch <lfam>Rovanion: Or check out the master branch from the timeframe when the patch was originally submitted <lfam>It's a problem with a patch-based workflow <mdevos>lfam: There were some issues with the channel definitions, and I didn't want to wait until mcron scheduled the job. <cbaines>Rovanion, which patch are you having trouble applying? <mdevos>lfam: perhaps ‘guix system reconfigure’ and related should set the umask before manipulating things under /etc, activating services, etc.? Or is this done alreadysomewhere? *civodul just spent 1+ hr catching up on 2 weeks of guix-devel <lfam>mdevos: I'm not sure that what you did is related to your problem. I'm just saying it's not recommended <lfam>I don't know the answer mdevos <cbaines>Rovanion, ah, right, if it's an old patch, there's not much you can do about conflicts. <cbaines>given it doesn't apply, there would probably be conflicts when trying to rebase <mdevos>lfam: me neither. If a find out where things go wring, I'll open an issue, and perhaps submit a patch. <mdevos>I did a quick 'git grep -F umask’, and it seems ‘guix system reconfigure’ doesn't set an umask anywhere! So activation scripts will inherit the umask from the environment I think, which doesn't seem ideal from a reprodicibility POV. <mdevos>I'll submit an issue (and perhaps a patch) later, I'm done with hacking for today. <mdevos>the unattended upgrade just completed, and now /etc/bashrc is unreadable again. Seems like a good indicator of where the issue lies! (umasks) 