<OriansJ>janneke: Good news, improvement to the stage0 lisp garbage collector has improved performance by a factor of 50 but I need a few more tests before I am happy with it. <janneke>I'm working through the tinycc tests/test2 suite, first 5 tests pass <CharlieBrown>I need to get Wi-Fi working. This is not working for me: <rekado_>ERROR: Wrong type to apply: "GNU with Linux-Libre 4.11 (beta)" <rekado_>and I noticed that the lightdm test suite fails in a non-deterministic way :( <rekado_>CharlieBrown: you can search the info file with “i” (for index) or with “s” (regular expression search) <buenouanq>so I've got slim in dvorak now, but how do I get all of x there? <buenouanq>the console-keymap-service only does the ttys right? <CharlieBrown>buenouanq: If it's an interface to loadkeys, which I would think it is, then yes: TTY only. <CharlieBrown>I would be surprised if there was a single interface to both loadkeys and setxkbmap. <buenouanq>on debian I set this (and my other custom key changes) in /etc/defaults/keyboard <buenouanq>It might be exactly how I changed slim, I just have to throw it another file <CharlieBrown>Mm... ntp-service... I really need to sync my clock. It says 17:45 RN. <CharlieBrown>I just installed tmux. Whenever I run it, it says "tmux: invalid LC_ALL, LC_CTYPE or LANG. <james-richardson>I'm installing guix on top of debian (I've done this several times before). This time I'm getting an error when I run guix pull, Error: no code for module (git) <buenouanq>so, what is the recommended way of just changing your keyboard layout? how do people on any distros normally do this? <buenouanq>I would love for it to be global though, so I didn't have to do this for each user. <CharlieBrown>I finally figured out how to edit files as root without changing my Emacs session. <CharlieBrown>buenouanq: I did setxkbmap in xinitrc, but that's probably not OK in GuixSD, because paroneayea said on Hacker Public Radio that xinitrc is not functional. <buenouanq>I'm actually not sure why what I did works in slim, but not in the DE. <CharlieBrown>OK, this is a serious problem: I can't edit my system declaration in Emacs! <CharlieBrown>Why would it be toggled in the first place? *I* didn't toggle it. <brendyn>I have 0.12.0.3490-fe24f but when I generated a install image and booted up, the guix version on it is just 0.12.0 <brendyn>infact many of the packages are older. serf 1.3.8 instead of 1.3.9 <paroneayea>CharlieBrown: you might have accidentally hit the toggle keybinding <buenouanq>anyone know how to enable `tap to click' under gnome on guixsd? <CharlieBrown>Hm. The manual says nothing about 'guix pull' being utterly broken. <brendyn>In order to get the new serf package i need to run guix pull but guix pull tries to build serf <buenouanq>I'm getting weird new guix pull problems I've never seen before too. <CharlieBrown>I'm just going to keep pulling it until my fingers get sore, or it's time for the evening meal. <lfam>CharlieBrown, buenouanq: I'm looking into it <lfam>You don't need to keep running `guix pull` until you see a commit that fixes the problem <CharlieBrown>buenouanq: I also get 'ERROR: no code for module (git)'. <lfam>I'll either fix it or revert the change in a few minutes <brendyn>ice-9 is just code inside guile that runs code <lfam>Weird. I can build a fresh checkout "by hand", but `guix pull` fails as you reported <DoublePlusGood23>rekado_: Odd. I passed it through the linter and it was fine, I actually cut up the description because the it said to. Is there a flag to remove the configure, build, check, and strip? The docs seemed to suggest deleting them was the proper way. <lfam>CharlieBrown, buenouanq: I've reverted the change that caused the issue. It should work now <lfam>The first 3 digits of an SSN correspond to one's birthplace. 038 is RI <OriansJ>AND DONE. Compacting added to stage0 lisp garbage collector with a nice performance boost that goes with it <platoxia>man I am having one hell of a time getting any irc clients to work form the terminal <OriansJ>only 53 lines needed for the compactor but 36 lines were needed to have the sweep short circuit a bunch of work <platoxia>I looked in the packages but couldn't find any emacs clients for irc...does the default have it? <OriansJ>platoxia: erc is included in the standard emacs <platoxia>I have been trying for 3 days with irssi but it won't connect <platoxia>Had to download ircii just to get a non-secure connection <platoxia>ok, emacs -nw; m-x erc works but not for port 6697...just like every other client I have tried <enderby>CharlieBrown: lol, accidentally typed an alias for ssh-ing to a server into the channel <CharlieBrown>How often should I do 'guix pull' before 'system reconfigure'? <CharlieBrown>I might be doing it several times in a row right now. Making small changes and then reconfiguring. <buenouanq>if it's only been a few minutes you needn't bother <buenouanq>good to get in the habit of, but don't go nuts <lfam>CharlieBrown: You only need to do it once, right before you reconfigure <lfam>The kernel is updated very often, usually more than once a week. <CharlieBrown>lfam: In how many hours should I do it again during my constant revision of the config? <lfam>I'm not sure what you mean. Can you clarify your question? <CharlieBrown>lfam: I will be editing my config some more if this reconfigure succeeds, and reconfiguring with the new edits. <lfam>I would run `guix pull` before the first reconfigure, and then test each configuration change without running `guix pull` in between. <lfam>On my machine, `guix pull` is just slow enough to be annoying. I wouldn't want to do it for each configuration change that I was testing <lfam>If I need to test some changes to Guix while reconfiguring, I'll use a Git checkout of the Guix source code instead of `guix pull` <buenouanq>CharlieBrown: we all do, it's a bandwidth/timeout problem with hydra <lfam>CharlieBrown: Once you build Guix from the Git checkout, it creates a script 'pre-inst-env' in the Guix source directory that you can use <platoxia>Have any of you been able to connect to freenode on port 6697? <buenouanq>scp runs and asks for the remote login, then stops with `bash: scp: command not found' even though it obviously found and was running it <platoxia>even my eww browser for emacs is giving me issues with anything using https...can't even get to the guix website. <lfam>platoxia: What kind of issues are you having? <platoxia>seems to be related to system wide ssl/tls/nss authentication <lfam>platoxia: Is the nss-certs package installed globally? <platoxia>I am using a very slightly modified bare bones config. <platoxia>I tried using nss-certs but it didn't fix anything <lfam>nss-certs is not part of %base-packages, so if you did not add it, then you won't have a system-wide cert store. <lfam>Depending on your needs, you can either install nss-certs per-user, or add it to the GuixSD (packages) field <buenouanq>is there any reason it couldn't or shouldn't be part of the %base-packages lfam? <buenouanq>seems sort of silly to me that it isn't, but there's prolly a good reason <lfam>It's not necessary to bring up a basic GNU / Linux system. %base-packages is really the *bare* minimum <lfam>platoxia: It's the section 'X.509 Certificates' from the Guix manual. You should have it locally with `info guix` <lfam>buenouanq: It fails to find scp on the remote end? What is the remote OS? <buenouanq>oh oh, it's going to be a --login shell issue <buenouanq>solved this internally with su aliasing su --login <lfam>Yeah... I worked around it by making openssh available globally on the remote GuixSD system <lfam>In the packages field of the system declaration. This ensures the login shell PATH is correct <buenouanq>doesn't seem the safest to open ssh to everyone though <buenouanq>but I'm no expert, and if that's the only/best way <lfam>It depends on the use case. I'm fine with that <lfam>You could also set PATH in your remote ~/.bashrc, but this will break `guix environment` and is generally bad practice <lfam>What do you mean by "open ssh to everyone"? On GuixSD, every user can install an SSH daemon if they want <lfam>But only privileged users can make the daemon listen on port 22 <buenouanq>is openssh still the recommended version of these I should be installing, or is one of the others we have better? <lfam>OpenSSH is definitely the most actively maintained SSH implementation we package. <lfam>I had a similar remote path issue using Mosh, and so I worked around it by installing the package in addition to running the openssh service <buenouanq>ok, it's time I actually learn how to figure out which modules things are in <buenouanq>openssh is unbound, so I need to include whichever package-module defines it <lfam>`guix package --show=openssh` will tell you the "location", which corresponds to the module name <buenouanq>so where is the root of this though? I have no /gnu/packages <lfam>When you run that command, the copy of Guix located at ~/.config/guix/latest is consulted <lfam>The module path will be (gnu packages ssh) <lfam>The ~/.config/guix/latest symlink is the thing that is updated by `guix pull` <buenouanq>can you now explain to me the intended differences in use between (use-modules ..., (use-service-modules ..., and (use-package-modules ... <buenouanq>and which things can take multiple arguments and which can't? <buenouanq>(use-service-modules databases desktop networking ssh web) <lfam>Those are service modules, so they are loaded by use-service-modules :) <lfam>They are found at 'gnu/services' <buenouanq>I can just keep adding here, can do the same inside (use-modules (gnu packages ...)) ? <lfam>I'm not sure what you are asking <buenouanq>(use-modules (gnu packages ssh) (gnu packages emacs)) <buenouanq>lfam: would wou have to add (leo packages whatever) or could you add whatever after openssh in your example? <lfam>buenouanq: (leo packages openssh) is a non-Guix module that I maintain with a customized OpenSSH package. <buenouanq>I'm pretty sure I know the answer though, and it's certainly not important. <lfam>I don't think so either (I'm not a very strong Schemer), but you could try it and find out :) <CalEmacs>Wait... should I put (gnu services networking) under (use-modules ...) or (use-service-modules)? <CalEmacs>(use-service-modules networking) vs. (use-modules (gnu services networking)) <CalEmacs>Adding NTP service: (use-modules (gnu services networking)) (services (cons* (ntp-service #:servers %NTP-SERVERS) %base-services)) <CalEmacs>Is %NTP-SERVERS defined, or is it a placeholder for something I have to define? <CalEmacs>ntpd is loaded, but why isn't the time updated? :-( <CalEmacs>But I still can't 'ssh cal@komputilo.local:2222' by enabling mDNS with (use-modules (gnu system nss) (gnu services networking)) (name-service-switch %mdns-host-lookup-nss)). <CalEmacs>Yeah. My clock's been way off all year now. <CalEmacs>I never bothered to learn how to do it in Parabola. But now, I know how to do it in GuixSD! <CalEmacs> -- Scheme Procedure: tor-service [CONFIG-FILE] [#:tor TOR] <CalEmacs>Isn't it against functional programming to rely on "[CONFIG-FILE]"? <CalEmacs>DoublePlusGood23: I'm using it to proxy with torsocks, and will eventually have a hidden service too <Apteryx>reepca: Congratulations for your GSoC! :) <bavier1>CalEmacs: The config-file is, I think, intended to be a 'plain-file' or such. I'd need to read the docs some more <CalEmacs>bavier1: Is a plain-file like a heredoc? <CalEmacs>bavier1: The doc said "CONFIG-FILE, a file-like object." <buenouanq>any ideas what the guixsd way of remapping a key under gnome should be? <buenouanq>have it enabled in the service on the remote machine, and am running it with -X <buenouanq>CalEmacs: you need a service on the host and to install the package under the user <CalEmacs>buenouanq: I have the service declared in my sys. config. I am trying to connect from my Android phone. Also, I'm trying to use the komputilo.local mDNS address. <brendyn>guix system disk-image --image-size=1G gnu/system/install.scm <brendyn>no longer works for me, can someone test it? <CalEmacs>What's a plain-file look like? I'm trying to get Tor working. ***CalEmacs is now known as CharlieBrown`
***CharlieBrown is now known as aoeuou
***CharlieBrown` is now known as CharlieBrown
<CharlieBrown>I don't understand the schema "tor-service [FILE-NAME] #:tor [TOR]". <CharlieBrown>Is ntpd provided by desktop? guix ystem: error: service 'ntpd' provided more than once. <buenouanq>yeh, you can't have both %base-services and %desktop-services <brendyn>CharlieBrown: I just had a look and it seems so. you can look in gnu/services/desktop.scm it has (ntp-service) <buenouanq>I have no idea why I can't get this to work. <buenouanq>but ssh -X tells me, `untrusted X11 forwarding setup failed: xauth key data not generated' <CharlieBrown>Now it complains that 'networking' is provided more than once, but when I remove both (gnu services networking) and networking, or just one of those two, it complains that dhcp-client-service is unbound variable. <buenouanq>(use-service-modules networking) will give you dhcp-client-service <buenouanq>but if you have %desktop-services it's included and can't be there twice <CharlieBrown>buenouanq: http://ix.io/sQr -- guix system: error: failed to load '/etc/leela-bare.scm: /etc/leela-bare.scm:52:19: In procedure #<procedure 3781600 ()>: /etc/leela-bare.scm:52:19: In procedure module-lookup: Unbound variable: dhcp-client-service <buenouanq>(use-service-modules ssh desktop) -> (use-service-modules ssh desktop networking) <buenouanq>and you can't have both dhcp-client-service and %desktop-services <buenouanq>you'll have to remove one, most likely the former <CharlieBrown>But it USED to be (use-service-modules networking ssh desktop). <buenouanq>right, and it complained that it was provided more than once because %desktop-services provides it <az`>esp me want to expand every %... variables in config now, because of verbosity <buenouanq>ntp is also part of %desktop-services too I think - did we just go over that? <CharlieBrown>buenouanq: How do I check if something is in something else? <buenouanq>it throws an error when you try to run it ;3 <az`>buenouanq: if I expand %desktop and others - may I have any troubles in future with that? <az`>CharlieBrown: you can inspect sources of any variable with <M-.> <buenouanq>I think that's a silly idea az` - You should let the globals like that that exist for a reason do their job. <az`>CharlieBrown: Yes, for me it works from <M-x>run-guile REPL <CharlieBrown>az`: My emacs isn't set up for guile stuff right now. Is there a guide on this? <platoxia>does %base-services add anything to %desktop-services or should I just use %desktop-services by itself? <platoxia>buenouanq: okay, that's what I needed to know, thanks <buenouanq>services can only be declared once and %desktop-services pulls in everything in %base-services <az`>platoxia: %desktop-services contain $base <CharlieBrown>buenouanq: I thought geiser was a package in Emacs. If I already installed it with Emacs, would guix not be aware of it, and I'd then be using two package managers by accident? <buenouanq>so, this whole `programs have their own package manager' thing <buenouanq>you should never install anything not through guix <buenouanq>why would you with all the problems that causes? <buenouanq>it make have been necessary for some thing at some point in the past, <az`>CharlieBrown: no, don't do it like this <CharlieBrown>But isn't GNU's own Emacs repo separate from GNU's own package manager? <buenouanq>if you find something you need, package it with guix <az`>yes and great side of guix is that, if something not in repo, you can "import" it! <CharlieBrown>I suppose so. After all, there's little difference between Emacs' package manager and that ghastly npm which needs to die a painful death. <CharlieBrown>So I was minding my own business, and then guix init shoved a login manager in my face. <az`>guix package -i setxkbmap <rekado_>CharlieBrown: if you’re using a not so recent Libreboot with a more recent Linux then the clock is regularly reset. <rekado_>CharlieBrown: it’s a bug in Libreboot <CharlieBrown>rekado_: That warning was a little random, but thanks for thinking about it. <rekado_>and you complained about your clock being reset <CharlieBrown>rekado_: Ahh, I see. Nah, my clock just skewed 10 minutes behind. I let it slip. <CharlieBrown>Thank you all for putting up with my whining. I'll add some money to the server farm donation cup once I get some. <rekado_>CharlieBrown: if this happens more often you can tell ntpd to adjust the time even if it is off by a lot <rekado_>with modify-services I match against ntp-service-type and override “(allow-large-adjustment? #t)” <CharlieBrown>rekado_: I think the problem was that I didn't know how to set up ntpd on Parabola. <Apteryx>I'm working in a git worktree, and guile can't find the modules anymore? Do I have to modify the load-path? <Apteryx>,use (guix build utils) returns: "ERROR: no code for module (guix build util)" <rekado_>well, yes, you need to tell geiser about your custom load path. <rekado_>(setq geiser-guile-load-path '("~/dev/guix")) <Apteryx>My main git tree is ~/src/guix but I'm working off a 'git worktree' at ~/src/guix-emacs-dvc <Apteryx>I'll try adding the later to my load-path. What happens if there are clashes? (same modules twice in the load path?) <rekado_>I suppose it will stop searching for another module when it found one with the correct name already <CharlieBrown>I don't have my .Xresources on me, and this white xterm is blinding me. <Apteryx>rekado_: Strange. Can't get it to work, even when my geiser-guile-load-path variable is set to only my worktree. Oh, maybe I have to re-run ./bootstrap, configure & make? It looks like a virgin git checkout. <Apteryx>You probably want to install some good fonts too... and configure antialiasing <CharlieBrown>Although one of the greens clash with the light blue, like in ERC> in Emacs ERC. <Apteryx>IIRC I just dropped a file somewhere and it did wonders. <CharlieBrown>I don't see any lack of anti-aliasing. Nothing looks jagged. <Apteryx>Looks like you found something to do ;) <Apteryx>I see! I think I just disabled highlighting in Weechat to get around that; but tweeking the color theme would be nicer indeed ;) <Apteryx>rekado_: I think I'll go sleep. As was typing (guix build util) buit it's (guix build utilS). eh. <CharlieBrown>I should probably sleep, too. I wanted my mic to work in Mumble, and for Tor to be up, but I should just be glad that I have a shiny new IceCat and a terminal with beautiful, clear fonts and nice colors. <buenouanq>I've uninstalled both gnome and xfce and am just using a plain xorg session <buenouanq>and it's still killing my second monitor as soon as I stop using it <buenouanq>have all been run and xset q shows that they've taken effect <buenouanq>what on earth would be shutting off my second monitor still? <buenouanq>I don't even know why this is possible, display cables should not be able to send commands. <rekado_>CharlieBrown: check the levels in alsamixer. <CharlieBrown>rekado_: There is no alsamixer in profile or in package listings. <CharlieBrown>I used that. I raised the volume a little by command line. It was tedious. <rekado_>CharlieBrown: it’s part of the alsa-utils package IIRC <Petter>Wow, Epiphany allows websites to set my system volume levels. I don't like this. <slyfox>tried to follow guixsd installation in a VM today for the first time. a few questions: <slyfox>installation uses package refinitions at a release time. AFAIU those are quite old. is it safe to 'guix pull' first? <slyfox>and second: many binary derivations are not available anymore. why? is it because nydra does not track release explicitly and garbage-collects it? <buenouanq>ok, so as I've encountered this every 10 minutes for the last few hours I'm starting to learn more about it <buenouanq>I can turn those on and they work and have a totally different behavior. <buenouanq>Something is actually turning off my monitor over the HDMI cable. <buenouanq>this isn't the same as screenblanking or a screen saver or dpms <buenouanq>I removed xfce and gnome and am just using a raw xorg right now to try and hunt this down. <Petter>Could it be a heat issue? (Just brainstorming) <buenouanq>I thought maybe too, but it's exactly every 10 minutes <CharlieBrown>Now I can troll that guy who hates Lisp, and say my init system runs on it. <slyfox>my monitor certainly disables after 1 hour after no activity. i control it with 'xset s 3600' <slyfox>(it used to be simething like 10 minutes, and was very irritating :) <slyfox>(i have also 'xset dpms 0 0 0' set) <CharlieBrown>Aw man... Mumble doesn't say "GuixSD" in the operating system field on my Mumble user info. <buenouanq>slyfox: I've tested xset dpms settings and they do something different. <buenouanq>this is instead like someone is hitting the powerbutton on the monitor, not just it going to sleep <slyfox>buenouanq: right, and 'xset s 3600'? <az`> icecat is very slow and use 5G with 3 tabs wtf <buenouanq>if I'm not back in 10 minutes, we have found the cause :3 <CharlieBrown>My copy of the Guix manual is different from the copy on gnu.org. Chapter 4 is "Programming interface" rather than "Emacs interface". :-( <Petter>The web based manual corresponds to the latest release. <Petter>buenouanq: Have you tried what slyfox suggested? <Petter>Nothing interesting in the link I pasted either? <CharlieBrown>Petter: My system is up to date/was just installed yesterday. It should be "the latest release", right? <buenouanq>everything is about xset and -dpms and I don't think it's either of those <buenouanq>I can set both to a very short time and see what they do right <buenouanq>and they work and it's something totally different <buenouanq>they turn off the display software side but it's still on <buenouanq>whatever this is is turning off the display like pressing the power button <buenouanq>I wasn't even aware this was possible, thus why I thought it might be an overheating thing. <Petter>buenouanq: Have you tried something like their Solution #1? <CharlieBrown>Installed emacs-magit-popup geiser emacs, but M-x guix still does not work. <buenouanq>seems so silly to have to install something to tell it to stop working <buenouanq>because if it's not installed, how can it be working <buenouanq>this whole experience is making me hate x more than I ever have <buenouanq>when I get home from my trip, I'm going to start dedicating time to a guile wayland window manager <slyfox>does problem stop happening if you are not starting X? <slyfox>would be interesting to check. because it could be monitor or video drivers <Petter>Didn't take 10 minutes now? Or do you use IRC on another computer? <Petter>Aha, I thought it would be 10 minutes of inactivity. <buenouanq>yeah, I tried things like while sleep 30 ; do xdotool keydown Shift_L keyup Shift_L ; done <Petter>Right, that conforms more to the story ;) <buenouanq>something I did yesterday fixed it, but it was lost on reboot <buenouanq>I thought it was a gnome problem - changed settings in xfce, went back to gnome and it worked <buenouanq>I have now removed both gnome and xfce and am just doing this from a raw xorg session <Petter>And next is to test without starting X? <buenouanq>but now will never know which of the countless things I tried actually fixed it ;~; <Petter>I'm guessing the xscreensaver trick. <buenouanq>yeah, cause the defaults for that are 10 minutes <buenouanq>I'm leaving on Monday till June and am not at all ready. <Petter>Where are you going? (If you don't mind me asking) <Petter>And it'll be great knowing your monitor will not shut down on you when you get back!! ;) <az`>where is right place to "export" environment variables? How to do this in scheme? <james-richardson>If I wanted to turn on ipv4 forwarding and load iptables rules, I would define a service and do the particulars as an activation-service-type? <civodul>or you could make it a shepherd service <az`>civodul: user context <az`>civodul: like SSL_CERT_DIR <az`>on other side GEM_PATH etc <az`>by default echo $SSL_CERT_DIR <az`>but in documentation we read: " export SSL_CERT_DIR="$HOME/.guix-profile/etc/ssl/certs" " <az`>where this exports want to be in guix ? <az`>I use Slim and ExWM via .xsession <az`>i.e. bash_profile not used <civodul>az`: that goes into .bash_profile or .profile, which is sourced by log-in shells <civodul>well, i have this in ~/.xsession: source "$HOME/.bash_profile" <az`>civodul: it is best aproach, yes <james-richardson>Does guix system container my-machne.scm work on foreign distros (e.g. debian)? <civodul>james-richardson: it should work, provided the distro supports user namespaces <civodul>otherwise 'guix system vm' should definitely work <civodul>LVM not sure, but mdadm support is here <civodul>raid-device-mapping corresponds to mdadm <slyfox>how to find out which module 'delete' should come from in guix? <civodul>slyfox: try ",apropos delete" at the Guile prompt <OriansJ>hopefully you are having fun today janneke, I certainly am (currently debugging assembly implementation of cell reclaiming in stage0 lisp) <janneke>i'm working on tinycc's `11_precedence' test <janneke>precedence is all handled by nyacc, of course...i'm currently looking wheter mescc's expressions need to be revisited <janneke>i know i cut some corners there, but making a change breaks everything... :-) <janneke>i may just get away with making an incremental change, duplicating some assembly and migrating later <janneke>the actual state of flags and accu, who tests, to jump Z or NZ -- those conventions...hairy stuff <OriansJ>Oh yeah, C is a massive ball of Hair in many ugly areas <janneke>yeah...i'm guessing everyting will fall in place once i see it...:-D <OriansJ>If not, we will figure something else out and have a lot of fun doing it :-) <janneke>Apropos...I'm postponing my reply to your last mail...you offered some inspiring urls and some help and expressed some hesitance to get into Mes's code base... <janneke>I'm talking about rewriting mes.c lisp interpreter into lisp compiler bit <janneke>For me it's even too early to think about code base, I'm wondering how to go about it, going from interpreter to compiler...and it would be fun if we could do that together? <janneke>possibly paroneayea has some input there too <OriansJ>Well, we already have multiple Lisp compilers to learn from. Heck we could even simply use the SICP Lisp Compiler design and tweak it slightly to have it produce x86 <janneke>Oh?! I thought all Lisp compilers I've seen do lisp->C ... schem48's PreScheme does that, I think Chicken Schehme does that and you also sent a url <janneke>Do you think that lisp->C->x86 is feasible, or do you just want to learn and write a direct backend? <OriansJ>janneke: A C compiler backend that generates its output based on as AST produces identical code when given an AST generated from Lisp <janneke>Wow, and Matt Wette is preparing for Nyacc 0.77 with all my wishes implemented! <OriansJ>Just simply take the work you have done from your C compiler backend and attach it to a lisp AST generator (parser) <janneke>OriansJ: hmm, it sounds simple enough <brendyn>I have serf 1.3.9 but when I build an install image, It ends up with serf 1.3.8. why is that? <OriansJ>Hence janneke, why I said you were much closer than I was to getting a working lisp compiler. Not to mention I still have to address the performance problems you have with Stage0 Lisp <janneke>OriansJ: yes, I see. I was hoping that we would be able to repurpose much of our work, esp. the C compiler or LISP->x86 compiler for Guile proper. I guess you have heard about Nash? If we can pull it off that parts of the bootstrapping codebase can be used more widely, we may get more help. To go via C there seems clumsy...but it's probably best to keep focussing on bootstrapping first. <paroneayea>janneke: I doubt I have time to pitch in but I will be interested in testing things <janneke>paroneayea: yeah, i don't have too high hopes on getting you to hack on Mes/stage0/bootstrapping...but now that OriansJ and I are talking rewriting Mes interpreter -> Scheme compiler...that sounds a lot like PreScheme <janneke>and you know and care about such things (and people, even) <slyfox>i suspect a lot of core-updates' modify-phases are lacking depends <OriansJ>paroneayea: However if you wish to help out stage0 there are still tons of small things that need to get done <OriansJ>Like find bugs and add them to the savanna tracker, improve documentation, make the guix packages for stage0 etc <janneke>i'm not happy with != ..., just waiting for inspriation there :-) <CharlieBrown>ERROR: In procedure mount: mount "/dev/log" on "/var/run/containers/tor//dev/log": Not a directory <OriansJ>FOUND IT!!! Finally cleared out the bug in Stage0's lisp interpreter's mark/sweep assembly implementation <OriansJ>Was COPY R2 R2 instead of COPY R0 R2 <CharlieBrown>alezost's config had just "(tor-service)", but with that line I get the above error. <janneke>OriansJ: i spent hours looking for a bug where I updated the assembly comments next to the hex, but forgot to update the hex itself <janneke>OriansJ: ...and it took weeks to get the last bug out of Mes' GC...so great! <OriansJ>janneke: ooh, those are the absolute worst bugs to find. Especially if you have hard coded displacements... <OriansJ>One trick I found extremely useful is when hand writing hex programs with assembly in comments is have a disassembler. <OriansJ>Then use meld to see the differences, it saves SOOO much time. <slyfox>heh, fun fact: guix has no 'meld' packaged <slyfox>yeah. i think i use it every day <CharlieBrown>My main uses are supported -- Mumble, Tor, full disk encryption, WeeChat. So, I switched. <OriansJ>slyfox: nor lightdm, light-locker or sway (wayland i3 drop-in replacement) <james-richardson>I'm install guixsd on an old laptop. Seems to work well, until I reboot into the newly installed system. The networking service will not start. It seems that the networking service wishes to bring up all the interfaces it finds. This becomes rather problematic with the built in intel wireless adapter. Is there a way to tell the networking service to ignore the wireless device? <slyfox>usually wicd is two things: wicd daemon and 'wicg-gtk' (or 'wicd-ncurses') UI <CharlieBrown>slyfox: Ah, yeah. I found those interfaces. I try them both as root and my user, and neither can connect to the daemon. Will it only start once I reboot, and not simply after a system reconfig? <slyfox>no idea. i never managed to install guixsd <slyfox>i suspect ou need to poke shepherd to start wicd after install <buenouanq>with %desktop-services, it should be pulled in and started automatically <CharlieBrown>I'd imagine it would conflict if it did it without rebooting. <buenouanq>always surprises me what can start/change without a reboot though <alezost>CharlieBrown: (tor-service) has always been working for me as is, but I have not reconfigured my system for awhile, perhaps something was changed lately, dunno <buenouanq>CharlieBrown: off to the right there is a triange from which you can select preferences <buenouanq>make the wireless interface the name of your device as output by ip a <alezost>CharlieBrown: btw you don't need to install emacs-magit-popup for "M-x guix"; "guix package -i emacs-guix" is enough <alezost>CharlieBrown: re your config - OK, I mean I can't help you as (tor-service) worked for me as I wrote <lfam>Ah, python core library updates... renovating the house of cards <efraim>hexchat fails to build the source on my aarch64 board, and linuxdcpp fails to download the source ***grillonBNC is now known as grillon
***methalo is now known as methalo_
<lfam>efraim: I can get the source of linuxdcpp, but I can't build the source of hexchat either <efraim>i get a 303 error from linuxdcpp <lfam>I'm not sure what's up with hexchat :/ <efraim>back to trying to build an aarch64 image, looks like i need to patch in a conditional on the qemu calls for aarch64 to add '-machine virt' since there's no default <efraim>closer, Permission Denied on KVM kernel module <lfam>efraim: I reverted the hexchat change <civodul>lfam: "guix build -S hexchat" gives "/gnu/store/19lkrck1844idbcfq6ajzr0akmr8rshj-hexchat-2.12.4.tar.xz-builder:1:2293: In procedure module-lookup: Unbound variable: inputs" <lfam>civodul: Yes, I just reverted the commit that introduced the problem. Reviewer failure :( <efraim>is there a list somewhere of modules to modprobe? <civodul>efraim: i'd do: find $LINUX_MODULE_DIRECTORY -name kvm-\\*.ko <efraim>i think lsmod only shows whats currently loaded <civodul>/gnu/store/pc5lf291l3cllrqn1r6wz02r1sda4gma-hurd-0.9/bin/addauth: ELF 32-bit LSB executable, Intel 80386, version 1 (SYSV), dynamically linked, interpreter /gnu/store/f44p3szhq74h957mmfd878azvp81x4cc-glibc-cross-i586-pc-gnu-2.23/lib/ld.so.1, for GNU/Hurd 0.0.0, not stripped <efraim>civodul: I came up with a plan for gccgo, looks like I should try copying the libvtv patch <lfam>./pre-inst-env guix build --keep-going $(./pre-inst-env guix refresh -l python-mock python2-mock | cut -f2- -d:) <lfam>(Besides a branch on Hydra, of course) <rekado>would it be possible in the future to easily setup a dual-boot Hurd system from GuixSD? <rekado>currently, installing a GNU+Hurd system is kinda hard. <rekado>CharlieBrown: tor works for me with just (tor-service) <CharlieBrown>TIL both alezost and rekado can use Tor with (tor-service), but I cannot. Hm... <rekado>you wrote that the service doesn’t start, right? <CharlieBrown>If I knew the command to print the messages, I'd paste it. <lfam>CharlieBrown: Have you added the tor-service to you GuixSD configuration file and reconfigured? <CharlieBrown>lfam: I have added (tor-service) to the service list in my GuixSD config file, done 'system reconfigure', and rebooted. Neither of the last two got it to work. <lfam>CharlieBrown: Are you root? <efraim>hmm, `sudo modprobe kvm<tab>' on my x86_64 machine shows kvm-intel and kvm-amd <rekado>CharlieBrown: does “sudo herd status” list “tor”? <civodul>rekado: dual-boot would be hard because libc on GNU/Hurd has a different ABI <civodul>so everything in your profile would need to be rebuilt, *or* magically adjusted <rekado>I meant the simplest of all dual boot scenarios <rekado>like: I can select either in GRUB <rekado>but I don’t need to reuse the same profile or home directory <rekado>CharlieBrown: that means it’s running <rekado>CharlieBrown: you just have to make your browser use it as a proxy <civodul>rekado: and yes it could be the same root FS <civodul>efraim: you don't have to use KVM though, dunno if it's even supposed to work on aarch64 <civodul>just remove "-enable-kvm" from the qemu flags <rekado>civodul: Justus told me that he dd’d the image to his disk to install GNU+Hurd; it would be great if one could just boot GuixSD and say “guix system … gnu+hurd” and thus avoid the trouble of getting the system installed. <efraim>civodul: thanks for the hint, i'll comment that out too for the moment <lfam>efraim, civodul: QEMU docs suggest that KVM does exist for aarch64 <efraim>i know, its supposed to be available on this board, and I think I have the config enabled for it, but it's still not turning it on <Petter>efraim: Looks like this is the place to look for available modules, /gnu/store/wcdcqp00al40g2v5naz5qlzxy07dfmqh-linux-libre-4.10.12/lib/modules/4.10.12-gnu/ <efraim>apparently I can have nice graphics or kvm, and it won't switch to kvm <Petter>`man modprobe` says: "modprobe looks in the module directory in /lib/module/`uname -r` for all the modules.." <lfam>efraim: Have you tried the Firefly yet? <civodul>rekado: i'm looking at automating the release process a bit more <efraim>lfam: i plugged it in, default boots to android, haven't plugged in a mouse yet <efraim>looks like I don't have a kvm module on this board <civodul>lfam: do you think that's something we could add to the build farm? <lfam>That board is a "real" aarch64 machine, with fast networking and storage interface, and support for 16 GB DDR4 <lfam>I should buy one and find out :) <civodul>it would be nice to provide aarch64 binaries anyway <civodul>maybe one of the makers could "sponsor" us by sending one or two boards <lfam>To me, the most important thing with these ARM machines is upstream support for the mainline kernel. I'm not interested in the vendor kernels which are unmaintained from day one, but they are the norm rather than the exception <CharlieBrown>rekado: Along with doing torsocks curl, I set WeeChat to use socks proxy. It won't connect. <efraim>The board supports mainline Linux or Linux 4.4.x, mainline U-Boot or U-Boot 2015.11, UEFI (Linaro UEFI tree), <rekado>civodul: I’ve allocated some time next week at the office to help with preparing the release. <lfam>efraim: The Marvell ARMADA 8040? <rekado>CharlieBrown: does “netstat -tunpl” show you that tor is listening? <lfam>The next question would be, are those CPUs actually fast? Or is it just a storage / networking machine? <civodul>lfam: right, if it's not faster than qemu on x86_64, maybe it's not worth it <lfam>civodul: Right, although if it's totally compatible with linux-libre, it could be valuable in the long run to support development of freedom-respecting hardware <lfam>Although our funds would be a drop in the bucket <Petter>CharlieBrown: Yours show the same as mine, and I'm using Tor. <lfam>Besides, emulating aarch64 on x86_64 may not be that fast. I don't know... <lfam>rekado: Are you still using NFS in your Guix installation at the institute? <rekado>lfam: yes, and I get painful reminders every few days <rekado>the institute wants to get of most NFS servers <rekado>Guix is not the only service that suffers from poor performance <rekado>I’m almost certain that it’s really just the fault of these terrible Oracle appliances <lfam>It's okay if no, I'll get around to it <lfam>Mostly, I want to check if all the substitutions are still valid. This is a major drawback of substitutions, in my opinion. At least with patches, you get an error if the patch has no effect <lfam>But also, I don't have an NFS installation to test against <rekado>yeah, I agree. I wanted to change substitute* so that it returns #f or throws an error on failure, but I forgot about it. <civodul>lfam: there's an nfs system test, i think <lfam>civodul: True, I'll use that <civodul>i just checked and it mostly tests rpcbind in fact <rekado>lfam: I won’t be able to check this tomorrow and next week could be tough <lfam>rekado: I'll manually inspect the results of the substitution and make sure the NFS system test passes <rekado>ACTION will be recording drum tracks tomorrow <james-richardson>If I run 'guix system disk-image ...' image files are created in /gnu/store. Will these get GC'd? <Petter>Ooh, what kind of music are making rekado? <rekado>Petter: our music is a bit weird. I wish I could call it progressive rock, but it’s not really progressive nor is it rock. <lfam>james-richardson: Yes, they will be garbage-collected. <Petter>Going ones own way is the best way. <lfam>james-richardson: You can make a garbage collector root to prevent this: <lfam>sudo -i ln -s $(guix system disk-image ...) /var/guix/gcroots/my-disk-image <rekado>Petter: over 15 years ago I wrote a four part piece and I could never quite commit to a final version. I hope I’ll be sick of it (and will have finished recording and mixing) before this year ends. <janneke>...almost running vanilla Nyacc-0.78 <amz3>m-o`: sorry, I have issues with my mails, what's the bug you create against guile-bytestructures? <Petter>rekado: Hehe, I'd be interested in listening to it. <Petter>I saw music flowed in your veins at the previous FOSDEM. <rekado>Petter: I’ll be sure to share it when it’s ready <m-o`>amz3: I had issues running guile2.2 + guile-git in guile-bytestructures <m-o`>a missing include in some r7/xxx.exports.sld file <m-o`>I added then to guile-bytestructures build, it seems to fix the issue ***grillon is now known as grillonBNC
<rekado>I’m confused about inputs and native-inputs. ***cal is now known as Guest96303
<rekado>In commit c2e4f14ac8cd3e1ce7f46a192ad0c9acc084b210 I *had* to use (or (assoc-ref inputs …) (assoc-ref native-inputs …)) <Guest96303>The Emacs daemon stopped when I stopped my WM. O_O <rekado>although I’m *certain* that the patch is among the native-inputs. <rekado>now I have a similar case where I need to do this for the “mingw-w64” package. <rekado>“(assoc-ref inputs "xgcc-core")” does not work, but “(or (assoc-ref inputs "xgcc-core") (assoc-ref native-inputs "xgcc-core"))” does. <Petter>Guest96303: I'd expect it to if it was a child of your WM. <rekado>the context is a package that I’m building for Windows, i.e. gnu-build-system with argument “#:target "i686-w64-mingw32"” <amz3>m-o`: what's the bug number, I'd like to check the patch <rekado>this caused a rebuild of the mingw cross-compiler and it failed because it couldn’t find xgcc-core <Guest96303>Default Xfce is ugly as sin. "pi arc-theme arc-icon-theme". <rekado>do certain inputs move from “inputs” to “native-inputs” during cross-compilation? <james-richardson>Guest96303: seems like that is a known issue with emacs linked with gtk. <rekado>james-richardson: I only know of this when attaching to a remote Emacs via X forwarding. It would crash when disconnecting. <rekado>the problem here sounds more like it was a child of the WM process, as Petter wrote <Guest96303>james-richardson: Oops. I didn't read it. I've ignored it many times, because this hasn't happened to me before. Pardon. <james-richardson>Guest96303: no problem. I was there a few months ago. I just ended up starting eamcs normally and having a (server-start) in my init. ***james-richardson is now known as j-r
<Guest96303>Can I put it in my .emacs? I don't see an init for Emacs. ***Guest96303 is now known as CharlieBrown
<mekeor>is it possible to show the configuration.scm of a system-generation? <CharlieBrown>I like how I don't have te remember the command to add my user to a group. I just add it to the list in my config. 
