<quiliro>guix-daemon --build-users-group=guixbuild <rekado>how did you install things? Are you using GuixSD? <rekado>what installation method are you using? <reepca>quiliro: that part there just means "that's how it could be run". Later in the binary install process it shows how to actually set it up to automatically run. <quiliro>i cannot run the daemon, and set it to automatically start on boot. <quiliro>guix-daemon --build-users-group=guixbuild <reepca>that command is in step 4, isn't it? <rekado>“5. Run the daemon, and set it to automatically start on boot.” <rekado>what follows is installing the systemd unit and starting the service <quiliro>so now i have to do step 5 which i do not know <rekado>“If your host distro uses the systemd init system, this can be achieved with these commands:” <rekado>“If your host distro uses the Upstart init system:” <reepca>Perhaps he didn't know which init system debian uses <rekado>“Otherwise, you can still start the daemon manually with:” <quiliro>oh ...sorry ...my bad habit of sequentiality and not holisticality <rekado>Debian probably uses systemd these days <rekado>so the first series of commands should be executed here <ng0>I have a motivation to learn assembler now. I am going to write a shell. <ng0>among other things.. but the editor I use now is nice, and it's so small <quiliro>Failed to execute operation: No such file or directory <ng0>yes, this is a "bug".. you need to use the absolute filename <ng0>it's not covered in the documentation though.. I just assumed that people search the mailinglist <reepca>When I tried "systemctl enable guix-daemon" I got the message "too many levels of symbolic links" <ng0>uh... let me search the mail archive <ng0>no, it is written in the email not the links. the links are just context. <quiliro>systemctl enable /root/.guix-profile/lib/systemd/system/guix-daemon.service <quiliro># guix archive --authorize < ~root/.guix-profile/share/guix/hydra.gnu.org.pub <quiliro>warning: failed to install locale: Invalid argument <reepca>if there wasn't any other output, then it seems safe to say it succeeded. <quiliro>oh...i thought that was only on other users, not on root <quiliro>how can i configure the nginx server to mirror all guixsd packages? <reepca>now you use the binary install of guix to set up the right environment for building the wip-uefi branch <reepca>ah yes, see section 7.1.7 of the manual for information on building the installer image. <quiliro>but making the installer image for uefi is different from the manual <reepca>Ah. Then I'm not sure how that's supposed to happen. Maybe mbakke left a note about how that differs in the repository... <reepca>Oh there it is, back in the chat a bit - "./pre-inst-env guix system disk-image --image-size=1200M <quiliro>use "./pre-inst-env guix system disk-image --image-size=1200M gnu/system/install.scm" to build the installation image <quiliro>and then dd it onto a USB drive as usual <quiliro>but i don't quite see how to do that <quiliro>but i also have to make the mirror so i can make the offline install <mbakke>basically, just "guix environment guix", and then "./bootstrap && ./configure --localstatedir=/var && make" <mbakke>add make -j10 to make it build a little faster :) <mbakke>quiliro: you don't have to be root <mbakke>I recommend doing it with a normal user account <mbakke>quiliro: you should do a "guix pull" for the user first, though, to get guile2.2 guix <mbakke>before using "guix environment guix" and configuring the git repository <quiliro>how can i make a mirror for offline install <mbakke>quiliro: you can include more packages in the installation image by adding them to "gnu/system/install.scm" <mbakke>you can copy it somewhere, modify it, and generate a uefi disk image from it with "./pre-inst-env guix system disk-image my-install.scm" <mbakke>anyway, did you run the bootstrap and make steps? <mbakke>also, make sure you are on the "wip-uefi" branch by running "git checkout wip-uefi" <quiliro>mbakke: i would like to have an installation image and a mirror to install the packages from <ashtioee>hello there i have a question re derivation flags: does guix have a meta.priority like nix? and what is the default value? (zero in nix) <quiliro>mbakke: could you tell me what to do after "guix environment guix"? <quiliro>it is not possible to have a temporary connection for installing guixsd...it is a long download process <reepca>"./bootstrap && ./configure --localstatedir=/var && make" (assuming you have the wip-uefi branch checked out. If you don't, or if you're not sure, run "git checkout wip-uefi") <quiliro>it is imperative to have a way to download all derivations or sources <reepca>In the git repository you cloned <quiliro>is taking too long and i get cut off the connection en a while <quiliro>mbakke: ./bootstrap: 5: exec: autoreconf: not found <reepca>is that after "guix environment guix" succeeded? <quiliro>reepca: so it will have to be another day <quiliro>i do not have connection in a few minutes <quiliro>reepca: will it work after "guix environment guix"? <reepca>it should, if not follow the instructions in 8.1 of the manual <reepca>also make sure you run ./bootstrap and the other commands in the same session you ran guix environment guix in <mbakke>reepca: thanks for covering, lol. looking forward to their feedback :) <Lambda>Hello, I need some assistance booting guix ***Lambda is now known as Guest10610
<Guest10610>I did, dd if=guix of=usb bs=4096, and it won't boot, the screen flashes black when I attempt to boot it from my USB and that's all that happens <brendyn>ACTION changes his keybourd layout back... <brendyn>Are you using the 0.12 GuixSD image? <brendyn>I don't really understand much how UEFI works but Guix might not work with it at the moment <brendyn>for me, I ran sudo dd if=guixsd-usb-install-0.12.0.x86_64-linux of=/dev/sde <brendyn>sha256sum guixsd-usb-install-0.12.0.x86_64-linux => 3ad03a796d091c536f4d3c5658efeb75fd65f8f4a6233c615dca0a8cced81005 <brendyn>You're sure you don't need the i686 version? <Guest10610>This is what I get for forgetting to wipe the previous dd <brendyn>I wouldn't have thought that would matter? <brendyn>Maybe mine worked because I overwrote a ubuntu image ha <Guest10610>Haha Ubuntu broke on me that's why I'm going for guix <DoublePlusGood23>Couple of questions. 1. is it possible to use GDM over SLiM, it seems to be in the repos. 2. is there a reason why the new GNOME Night Light isn't available? <brendyn>DoublePlusGood23: I think so yes. I asked about display managers before and the Boss said he'd be happy with GDM <DoublePlusGood23>brendyn: I think it's the reason why I can't logout of GNOME, though I could be missunderstanding <brendyn>I've never used a display manager before anyway <brendyn>I'd like to start using gnome or kde stuff, but still have a tiling window manager <DoublePlusGood23>brendyn: tiling wm's are great to. I used i3wm for about a year, but I kept running into edge cases. <brendyn>DoublePlusGood23: I've heard some people run them within gnome so they can use gnomes toolbar, suspend, networkmanager, etc.. ***andreh1 is now known as andreh
<rekado>“systemctl enable guix-daemon” should be enough. An absolute file name does not need to be given when systemd has updated its registry <efraim>mbakke: I got the following error from your branch on my x86_64 machine: ERROR: In procedure #<syntax-transformer bootloader-configuration-bootloader>: <efraim>ERROR: Wrong type to apply: #<syntax-transformer bootloader-configuration-bootloader> <rekado>the java bootstrap is coming along nicely, but the test errors in kaffe really are problematic <rekado>27 regression tests fail and it’s hard for me to understand them. <rekado>analyzing this requires reading Java… :) <rekado>looks like they are all related, though. Something to do with serializing arrays. <rekado>I ignored the failures and tried to just use jikes+kaffe+classpath 0.93 to build ant 1.8.1, but it fails setting attributes. <rekado>that’s a big deal because ant reads lots of xml files where each tag contains attributes to set. <slyfox>when building packages with --no-substitutes is there an easy way to disable tests in all packages when derivations are built? <civodul>slyfox: no, that cannot be done without changing the derivation <slyfox>aha, and '#:tests?' is really a keyword for a particular build system, so i could flip the default locally as guix/build-system/gnu.scm: s/(tests? #t)/(tests? #f)/ <civodul>you could do that, but you'd end up with different store file names than everyone else <b11111000000>is there a general guix way for auto mount usb devices configuration? If I don't use The Gnome? <davidl>I just let guix system init --fallback run for over 2 days and then the process hanged and now the command hangs right away. I don't know how to get this installed. <brendyyn>davidl: I've been going for 3 days. compiling webkit at the moment <davidl>brendyyn, yeah webkit really takes forever. I did that too 3 months ago. <brendyyn>Sleep_Walker: Yo, any updates on LVM support? <rekado>I’m happy: I just built a bootstrap ant (1.6.0) without the OpenJDK <rekado>I used the unmaintained sablevm with jikes <rekado>kaffe had lots of problems that could not be resolved, unfortunately <rekado>jamvm requires a later version of GNU classpath (which cannot be built). <rekado>sablevm comes with it’s own version of GNU classpath, and it just works <rekado>it hangs trying to compile ant 1.8.x, but 1.6 is totally fine. <rekado>will try with 1.7.x and then move on to building ecj. <brendyyn>Sleep_Walker: Cool. I think I'll just go for plain old ext4 for now <civodul>rekado: in other news, i just pushed "proot-static" <civodul>combined with 'guix pack', it provides an easy way to run Guix stuff in hostile environments :-) <efraim>Does proot build on anything other than x86_64? <rekado>what is your opinion on better documenting environment variables <rekado>I see that R, for example, needs CURL_CA_BUNDLE in order to find certificates. <rekado>Should I just add this to the X.509 Certificates section? <civodul>rekado: but for purely computational code, it shouldn't be a problem, i suppose <civodul>rekado: re CURL_CA_BUNDLE yes, feel free to add it to "X.509 Certificates" <rekado>woohoo: ecj-3.2 has built! Now I wonder if that’s already enough for classpath 0.99. <roelj>rekado: Wow, really cool! I'm preparing myself for reviewing the second batch of Java patches. Unfortunately, it seems I only have time on monday. <rekado>roelj: thank you! The second batch is much more regular. Almost nothing had to be patched. <rekado>when this is done I should write a blog post and publish it on bootstrappable.org <roelj>Definitely. This is another leap for GNU Guix I think. <roelj>Why did we change '_' to 'G_' in the guix code? <roelj>It broke GWL.. and I'd like to avoid finding it broken again.. :) <rekado>roelj: I think it was to avoid a conflict with match syntax. <civodul>roelj: you shouldn't use it though, because it looks up translations in the "guix" gettext domain <civodul>or at least, you shouldn't use it yet ;-) <brendyyn>Some code is just so mysterious I start to come up with conspiracy theories about it. <brendyyn>I'm trying to package Helio Worksation, but the header files for JUCE seem to be generated with this proprietary program, and all the C++ code is dual licensed proprietary & GPL3 and there is no actually build script or anything <brendyyn>I was wondering what motivated them to suddenly make this program a libre version, and where it even came from <brendyyn>So far the makefile doesn't even work for me <brendyyn>But I talked got word from someone in #fsf that uses JUCE and it seems legit <rekado>sneek: later tell brendyyn JUCE is quite hard to package. I gave it a try some months ago, but it isn’t straight-forward. <sneek>Welcome back brendyyn, you have 1 message. <sneek>brendyyn, rekado says: JUCE is quite hard to package. I gave it a try some months ago, but it isn’t straight-forward. <rekado>weird, ERC doesn’t give me name completion for your name. <brendyyn>rekado: What do we need to do to package it? <rekado>brendyyn: it’s been too long since I looked at it. I just remember that it took me longer than I had patience. <Petter>Claws-Mail isn't very good. I didn't exit it cleanly, and now it doesn't find my configured e-mail accounts... <Petter>And when I initially configured it, I deleted the first account. And then as well it didn't find my accounts any more. <efraim>it seems python-graphene is more than just bumping the version number <davidl>I tried to install using GPT and it went well except for grub which says this GPT partition contains no BIOS Boot partition. And apparently you are supposed to make a separate partition and write "(parted) set N bios_grub on". Is it possible to do this in the Guix declaration? <jlicht>I am getting some segfaults when giving input to a simple guile repl. Something regarding SIGABRT and (probably) readline. <jlicht>Do other people also have their `guile-2.2` located in `/gnu/store/3kqj6adsk53h0zr4fhbh9qmaq2rl0qci-profile/bin/guile`? <civodul>hmm "herd stop cuirass" doesn't seem to stop the process <m-o`>hi civodul, maybe it's because of guile-2.2 switch ? <civodul>davidl: the GuixSD declaration doesn't say anything about your partition table; you really need to set those flags when you partition your drive <m-o`>i've been doing a lot of herd restart cuirass without problem <civodul>m-o`: maybe it's the new snapshot that has this problem <jlicht>after doing the guix pull thing twice ;-) <civodul>jlicht: people are unlikely to have the same profile as you <civodul>can you reproduce it in "guix environment" or something? <jlicht>it is on x86_64, but I will have to do some more research regarding the ¨guix environment¨ <jlicht>hmm, in `guix environment --pure --container --ad-hoc guile -- guile`, I do not have this issue <jlicht>could it have something to do with guile 2.2 trying to load some 2.0-related files? <davidl>civodul: yeah I realized it was a stupid question and did that. but thanks. <balduin>how can I cross-compile guix for arm64? My goal is to run guix as package manager on top of debian on a Raspberry PI 3 <civodul>jlicht: it could be "something" in your environment, maybe something in GUILE_LOAD_PATH <civodul>you can also try: env -i $(type -P guile) <civodul>if it doesn't, then it's something related to your env vars <civodul>balduin: "guix build guix --target=aarch64-linux-gnu" might work <efraim>i have to go, i'll be back tomorrow evening-ish :( <civodul>because of cmake-build-system, which doesn't support cross-compilation <efraim>and I don't seem to have an already uploaded unofficial binary-install tarball <civodul>balduin: ATM i think your best bet on aarch64 is to build Guix from source (git master) <efraim>oh, i have one on my aarch64 board, I can upload it to flashner.co.il/~efraim <civodul>m-o`: re "herd stop cuirass", it's just the 'evaluate' process that kept going; 'cuirass' itself was stopped as expected <efraim>give it a couple of minutes, it should be there <jlicht>civodul: Thanks! I found the culprit: export TERM=xterm-256color somehow lets *something* in readline and/or guile crash <jlicht>I had this export in my bash profile so my emacs theme rendered properly on temrinals, but I guess I will have to use something else <civodul>jlicht: i can't seem to reproduce it, even with guile-readline loaded <civodul>could you try to find a way to reproduce it? <jlicht>civodul: I´m looking for a way. Thanks for the help <jlicht>sneek: later tell civodul: The segfault is very hard to pinpoint, as it comes from libc, but I can reproduce it each time TERM is set to *anything* x-related. Both the xfce and st terminals let guile crash, but not the non-Xorg C-M-<f2> tty. <jlicht>Is there any practical reason why guix could not fetch its substitutes via e.g. bittorrent? <paroneayea>so, does anyone else experience this with icecat? <paroneayea>I've noticed this for a while and I've found it kind of disturbing? <paroneayea>I tried opening a new icecat profile, same thing <paroneayea>why does icecat require internet access when opened in the first place to continue working correctly? <paroneayea>it makes me nervous that maybe something strange is going on. <avoine>paroneayea: sounds like the offline mode is on <avoine>paroneayea: maybe it's caching the error page <paroneayea>avoine: it could be, but then it's for all webpages. <rekado>paroneayea: next time this happens check if “work offline” in the File menu was automatically enabled. <paroneayea>avoine: actually, it's only happening in this one profile. <paroneayea>maybe I should ditch this profile and recreate it. <catonano>paroneayea: what do yo mean with "profile" ? The Guix profile or an Icecat speciic thing ? <paroneayea>I created a totally new profile and this stopped happening <jackhill>Hi, If I'm looking at a package description, and see in the location for where a package is defined (e.g. gnu/packages/haskell.scm:1684:2). Can I look somewhere on disk to see the package description? <rekado>jackhill: you can use “guix edit package-name” <balduin>jilicht: it would be awesome to have bittorrent support in Guix <sneek>civodul, you have 1 message. <sneek>civodul, jlicht says: The segfault is very hard to pinpoint, as it comes from libc, but I can reproduce it each time TERM is set to *anything* x-related. Both the xfce and st terminals let guile crash, but not the non-Xorg C-M-<f2> tty. <civodul>m-o: did you see the trick in build-aux/build-self.scm to handle guile-ssh? <civodul>you could do something similar for guile-git <civodul>and reinstate the (guix git) commit that was reverted recently 