<nckx>Seems like IceCat on Sway segfaults more often than ever before. Which is to say, at all; it never used to until few weeks ago. <mbakke>zceejkr: can you file a bug report with information about the partitions (i.e. 'parted /dev/sdX print')? Sounds like it should be fixable. ***jess is now known as sandcat
<mbakke>db48x: were you able to try the same version of guix-daemon? If you still can't reproduce it, I guess we can go ahead with the last version of the SELinux patch + the JIT fix. ***sandcat is now known as jess
<vagrantc>i'm getting a lot more successful substitutions with "guix pull" lately <vagrantc>rather that pull building a lot of things all the time <apteryx>hmm... how can I combine patches that want different patch level at application time? <janneke>apteryx: prolly easier to change the patch level ***db48x` is now known as db48x
***apteryx_ is now known as apteryx
<nckx>raghavgururajan: Bayfront is a D16. <raghavgururajan>nckx: Ah! Do you know what cooling system is used? Like make and model? <nckx>If anyone (still) does, it's Andreas. <lispmacs>hi, I'm getting a gateway timeout error from ci.guix.gnu.org <nckx>lispmacs: ‘Probably a fluke, try again’ is the unsatisfying answer. <lispmacs>I was able to build from source using --fallback <nckx>raghavgururajan: If there's a good chance it will be used in future I'll keep it. <nckx>lispmacs: It was the source that was being downloaded anyway, so you lost no time. <nckx>Downloads fine here so --fallback was probably optional. <apteryx>janneke: the thing is one patch wanted -p1 and the others -p2 <apteryx>is modifying llvm-9 OK for master? It's the default llvm, but not many packages seem to depend on it <mbakke>apteryx: should be fine as long as LLVM 10 is not affected (which does have many dependents) <nckx>apteryx: I think they meant: edit the actual patch, or at least that's how I read it. ***iyzsong- is now known as iyzsong
<mbakke>db48x: I suppose what I need is a way to prevent the guix_store_content_t transition on the particular Guile executable guix-daemon needs for its own operation. <mbakke>But I don't see a way to do that that cannot be subverted by malicious build jobs. <mbakke>is there an IRC channel or mailing list where SELinux experts gather? :-) <mbakke>anyhow, if the same guix-daemon works on Fedora, I don't see a reason to withhold the patch <db48x>there's already a nice big hole in the rules <db48x> (filecon "@storedir@/.+-(guix-.+|profile)/bin/guix-daemon" <db48x> file (system_u object_r guix_daemon_exec_t (low low))) <db48x>any file in the store that happens to be named bin/guix-daemon will get relabeled to guix_daemon_exec_t eventually, and then it can do anything guix-daemon can do <db48x>actually, now that I think about it <db48x>what was the label on the guix-daemon process you were trying to run? <mbakke>db48x: I think you may be on to something <mbakke>-r-xr-xr-x. 2 root root system_u:object_r:guix_daemon.guix_store_content_t:s0 569 Jan 1 1970 /gnu/store/wvn92qff4pg4ipflg6169zdnmw8i6jkl-guix-daemon <mbakke>it does not have '.../bin/...', causing the label to miss <apteryx>mbakke: was looking at python-pysam; I fixed it on version-1.2.0 by updating + disabling a few tests. I see you pushed a patch from Roel having it used samtools-1.9 instead. Do you know which is preferable? <mbakke>db48x: indeed, adjusting the label to /gnu/store/.+guix-daemon and relabelling, my last two changes are seemingly no longer necessary! <mbakke>apteryx: I think you have to ask Roel, who pushed the patch :-) <db48x>why isn't it in a directory though? <db48x> (filecon "@storedir@/[a-z0-9]+-guix-daemon" <db48x> file (system_u object_r guix_daemon_exec_t (low low))) <mbakke>db48x: or change the existing guix-daemon rule? do we need both? <db48x>it was easier to write as a new rule <db48x>probably easier to read as well <mbakke>db48x: thanks a lot for your guidance on this :-) I got to go now, but will try and push it later or tomorrow evening <db48x>I'll rebase and send an updated patch ***amfl_ is now known as amfl
<db48x>mbakke: we could fix it by creating a second store just for guix-daemon packages <db48x>and store it in /gnu/guix-daemon/, for example <db48x>this would allow new guix-daemon executables to automatically be labeled correctly (no need to manually run restorecon after ever guix pull, for instance), and guix-daemon would only allow itself to be put there <apteryx>mbakke: ack, I thought I saw you were the one who pushed it, I must have been mistaken :-) thanks. <divoplade>Has anyone succeeded in installing jitsi meet here? <db48x>I've considered attempting it <db48x>not that this helps you much <divoplade>There's that videobridge that seems to build with maven <jonsger>what can I do against `/bin/sh: /gnu/store/hifvfda7vr5gx2jilssrc0a2282mskdh-profile/bin/guile: Datei oder Verzeichnis nicht gefunden` inside guix environment guix while running make? <divoplade>Re-run configure within the environment maybe? <jonsger>divoplade: ah yeah, forgot about that :P <g_bor[m]>I used the menu installer, and selected gnome and nothing fancy. <g_bor[m]>I just got an error from the nss cerst installer, that one of the files, Netlock_Arany_ClassGold_F??tan??s??tv??ny Nos such file or directory. <maav>g_bor[m]: does this happen during the installation? <db48x>guix-binary 1.2.0rc1 contains guix-daemon.cil.in instead of guix-daemon.cil :P <db48x>does anyone know how the release binaries are built? <lafrenierejm>Does anyone have experience enabling unstable features in a Rust library (crates-io.scm)? I'm trying `#:features` and not getting any luck… <isengaara>Today I managed to get guix running on my Talos II <isengaara>after to Guix Days, I'll plan a GNU Lilypond workshop via BigBlueButton <vagrantc>doh, i've been too busy working on packaging guix in debian to prepare something for guixdays... <Gooberpatrol66>Where's a good place to see examples of how to define shepherd services in config.scm? <dustyweb>non-urgent ping in case we are both around at the same time <dustyweb>it would be good for me to switch over my servers fully this weekend <jonsger>dustyweb: switch your servers to Guix System? <dustyweb>jonsger: they are already partly switched, yes <dustyweb>the last one being the mail service stuff *jonsger should go and write a service for radicale :) <jonsger>I'm trying to move away from Nextcloud. It feels a bit to big and not really handlebar... <divoplade>Anyway, roptat's config has a lot of good ideas <nckx>sneek: What is in your pocket? <sneek>I could be wrong, but in your pocket is a tasty cookie! <nckx>Anybody else having frequent IceCat segfaults on Sway? <jonsger>nckx: I'm not using Icecat. But maybe it helps to create icecat-wayland like it's the case for icedove <jonsger>I'm running firefox-wayland just fine on sway on HiDPI... <nckx>(I'll take crashes Wayland in general too.) <nckx>jonsger: Thanks for the tip. <nckx>about:support says ‘Window Protocol wayland’. Does that mean it's already as native as can be, or can it be nativer? <nckx>db48x: It's designed to be completely i3-compatible (down to the configuration file), but due to how Wayland works it's also a bit of a platform in itself. E.g., you have to ask Sway (not Wayland) to manage DPMS instead of X, Sway (not Wayland) manages your displays instead of X(randr), etc. <nckx>So far my understanding of it anyway. <db48x>have you gotten the crash in a debugger? can you record it with rr? <nckx>'s Why I avoid calling it ‘swaywm’. <nckx>Do I have to start the whole browser under rr or just enable core dumps & wait for it to shit the fan? <nckx>It says it's ‘lightweight’ so fine. <nckx>It also says it ‘makes debugging much more fun’, joy. <db48x>you shouldn't notice much slowdown <vagrantc>you even need to use sway to configure the keyboard layout <db48x>I've gotten pretty used to X, but I guess I'm going to need to learn something about Wayland eventually <nckx>SDDM (with insta-auto-login; unsightly beast) does that for me here but that might be a Guixism. <nckx>(service sddm-service-type (xorg-configuration (xorg-configuration (keyboard-layout keyboard-layout)))) <vagrantc>ah, i just log in from the console and run "exec sway" :) ***str1ngs_ is now known as str1ngs
<vagrantc>yeah, it takes too long to boot up to a GUI, sometimes i don't bother to run sway at all :) <nckx>I'm busy and important. I can't ‘type commands’ or ‘think about what I want to do’. <leoprikler>nckx: I know people, who think that way about passwords. <nckx>I've encountered ‘0’ and ‘x’ in the wild. <divoplade>I think many users don't understand command lines because they are in english (and, let's face it, very strange english). If you could use more natural language (like when you use voice recognition interfaces), it would be a lot less scary. <leoprikler>There's certainly a factor of that being to it, but I don't think it would be friendlier to my fellow German speakers if I had "katze DATEI..." <leoprikler>as soon as you have --help and l10n exists, things are a bit easier <leoprikler>I think binding Hilfe! to thefuck might help a bit, but it might also complicate some things. <leoprikler>I somehow like, that nixos has a thefuck rule and Guix does not. It means our users are less likely to shout profanities at the computer when stuff isn't quite working (or alternatively, that stuff works better on our side :D) <divoplade>Imagine that: you have a fully internationalized program, and you can run katze --hilfe. Then, you find a tutorial on the web, but it's in english and it says to run LANG=C cat --help. Because the locale is set to C, now the program is named cat and understands --help, but at startup you get a message: "You can get the same with katze --hilfe" <db48x>if the locale is set to C, how would it know to give any information localized into german? <divoplade>Because it would have remembered that you already ran this program in german <db48x>I knew this was starting to sound familiar <rndd>hi everyone! can i ask here about common lisp. i dunno where ask common lisp questions related to guix <dustyweb>so I could use some gexp help from people who are smarter than me about gexps ;) <dustyweb>I've defined the new <setuid-program> type, and I guess it's not a valid input itself to a gexp <dustyweb>is the right thing to do to "pre-unpack" this structure before handing to the gexp? <rndd>dont understand what's wrong <db48x>rndd: what is on line 6 of /gnu/store/iqicv56va1l6m4fdhqgpj3var2xv39dy-sbcl-claw-stable/lib/sbcl/claw.asd <divoplade>Is lisp good? pros: clean syntax, meta-programming; cons: the name of the function to create a pair. <leoprikler>dustyweb: what exactly is this <setuid-program> type? something like mixed-text-file? <dustyweb>leoprikler: let me see what mixed-text-file looks like again ;) <dustyweb>leoprikler: but you can see the details here: <db48x>rndd: ok, so common lisp doesn't have a false value like scheme does <db48x>so the error makes sense; it can't read this file <rndd>db48x: it is different as i see <db48x>rndd: now the question to ask is how it got there