<ng0>kei or albin packaged red eclipse? what are the technical requirements? the website hides it pretty well.. <ng0>ya, an nvidia 7300GS is definitely a bottleneck for such games these days.. <lfam>What is the status of core-updates? Can I do the Expat "ungrafting" or is the branch frozen? <lfam>mark_weaver: Can you say if core-updates is frozen or if I can do the Expat "ungraft"? <kori>yeah it panicked again, I can't figure out what's wrong and how to debug it <lfam>kori: I saw your panic report in the log. I'm not really sure where to begin. Are you that the filesystem label of /dev/sdb1 is "Guix" (you can check with the `e2label` program) <kori>the error is 0x00000000... iirc <lfam>So you fixed the filesystem label but there is some other problem? <kori>yes, there's some other problem <kori>I copied over the lightweight-desktop config file <lfam>It would help to see the full panic <kori>yes. the question is... how? <lfam>That's a good way to test things, but the problem might not manifest there <lfam>It's worth a try anyways <kori>it did in fact kernel panic <kori>if I could get a bigger log that would be better <lfam>kori: What version of Guix are you using? Are you working from a Git checkout of Guix? A `guix pull`-ed installation? Et cetera... <kori>lfam: downloaded the iso today <kori>followed the steps to a T <lfam>We don't offer an ISO... <lfam>Are you sure you selected the right architecture? <lfam>Can you paste the command line you used to start QEMU? <kori>sudo qemu-system-x86_64 -snapshot -boot c -hda /dev/sdb <kori>sudo qemu-system-x86_64 -boot c -hda /dev/sdb <lfam>My experience with QEMU is fairly limited. Is it really possible to ask QEMU to use one of the host's partitions (-hda /dev/sdb)? I typically create a QEMU image file and pass that -hda <ng0>you can also just expand the guix image file <lfam>Right, it seems like you would want to pass the the Guix image to -hda, and a blank image file to -hdb, or something like that <lfam>In your command line, I don't see the Guix image being passed to anything <ng0>if it creates ponies and star dust, I'm tired and off to bed, but iirc I recall that this was possible last time I tried with qemu. <kori>I'm just using qemu to boot the image <kori>so I could get an img of the kernel panic <lfam>kori: The command you pasted doesn't seem to name the image, so I'm confused about how it relates <lfam>Unfortunately I don't understand the panic trace <kori>lfam: guix is installed on /dev/sdb1 <kori>and when I try to boot from it (via hardware) it hardware panics as well <kori>I believe its not finding init <kori>and I'm not sure how to fix that <lfam>I see. I didn't know QEMU could be used in this way <lfam>I'm building a QEMU VM from the configuration file you pasted <lfam>If I can do it quickly. Depends on this network's speed <kori>nothing has changed except the label thing <lfam>Which one should I try? The new one? <lfam>I can't download all the ghc-* dependencies of xmonad. It <kori>just skip the optional package <lfam>I'll try the latest configuration you posted, and replace xmonad with a smaller DE <lfam>BTW, I'm doing it like this: `guix system vm ~/tmp/config.scm`. That command is provided by the Guix package manager (I'm on Debian) and will create an immutable QEMU VM from the configuration file. <kori>lfam: you can just remove it altogether <kori>I just want the system to boot haha <lfam>Out of curiosity, what hardware are you using? <lfam>Like, what processor model? <lfam>kori: It builds. In the meantime, you have another system on /dev/sda? <kori>lfam: aye I'm on it right now <lfam>I wonder if GRUB is working right. Did you pass --no-grub to `guix system init`? <kori>I am using grub for guix <lfam>And there was already a GRUB for the other system? <kori>the other system uses syslinux <lfam>So, the panic happens after GRUB? <lfam>kori: Your configuration (with xmonad replace by xfce) works with `guix system vm`. Can you send a detailed report to help-guix@gnu.org? <lfam>Oh, actually it does not work. It failed later than I thought it would fail <kori>lfam: yeah the panic happens after grub <lfam>You set your home directory to /usr/kori. This won't work because there is no /usr on GuixSD by default <kori>lfam: ...if I create it, will it stop panicking <lfam>I don't know. It fails in a different way for me. Try using /home/kori instead <kori>there's only one way we can know <lfam>I doubt this is the root of your problem, but it is a problem <lfam>The kernel does init for me. It fails near the end of the boot <lfam>Probably there is some problem with GuixSD coexisting with your existing system <lfam>I still think you should mail help-guix@gnu.org <kori>and yeah it still panicked <lfam>You are more likely to get a good answer there. The people who can give you the best help are not reading IRC 24/7 ***fkz is now known as Guest86992
<ng0_>whoever asked on the list: I picked up the gnunet.scm again .. I have not decided on what I do with the gentoo ebuilds which are donefrom my pov, but I'll fix the guix pckages now. <ng0_>how are most gnome packages build? with the glib-or-gtk-build-system ? <iyzsong>ng0_: yep, if it (most likely) requires gsettings schemas. <ng0_>not sure.. it's libunique-3.0.2 I want to package <ng0_>I guess it requires gsettings <iyzsong>right, libraries don't need it. glib-or-gtk only add the wrapers (XDG_DATA_DIRS) for binaries. <ng0_>I thought gnome.scm had some special system for itself <boegel>seems like the answer is no there... <boegel>does Guix work without the daemon? or, in other words, how crippled will my Guix install be if I configure it with --disable-daemon? <janneke>boegel: without daemon you won't be able to build packages yourself <boegel>janneke: ok... and the daemon *must* run as root? <boegel>janneke: why is that required exactly? and you seem to suggest that may change soon? <janneke>boegel: i remember hearing about an effort running the daemon non-root, could have been hurd-specific <ng0_>I think the answer to the first question could also be if you want it, you can try to port it. <boegel>ng0: I was mainly wondering to be honest, since Nix does support it <ng0>there are differences between Nix and Guix' goals :) <ng0>where is the "clear" binary provided from? <ng0>for clearing the screen I mean <boegel>janneke: ok, I see, no isolation without root, makes sense, thanks! <boegel>since I'm especially interesting in the 'jails' Guix provides, I guess I'll need root <janneke>isolated builds is another great *ix feature <boegel>it's the main reason I'm looking into Guix/Nix <boegel>I guess they both use the same mechanism for that? <boegel>I'm still not sure why both projects are around, seems like a huge waste of effort... <boegel>(says a guy who runs yet another build tool :P) <ng0>boegel: Nix has a different roadmap and is driven by different ideas than Guix <boegel>ng0: is there a good overview of the differences somewhere? I know Guile is one <janneke>Guix was created by a Nix hacker who acknowledged Greenspun's 10th rule <ng0>easiest would be in the roadmap files (if Nix has one), else it was explained in recorded talks which are visible on the gnu.org/s/guix website, or someone has time to get into details.. I don't . Plus I think other people are more competent in explaining it. For my part I know Nix has no GNUnet distribution at the end of the roadmap, Guix has. <boegel>I saw an article or blog post pass by that Guix is able to make progress a lot faster than Nix because of the approach being used <janneke>random-nick: isn't that why we're here? <janneke>Nix is also not written in C or Fortran, but C++... <catonano>What does "progress" mean, exactly ? How does the guix approach differ from the nix One and how is that difference helping, exactly ? <janneke>catonano: progress, or even the need for it, is in the eye of the beholder...what would you want to create? <catonano>I have my own experiences with both guix and nix so I have some ideas and hunches. But because I feel like an outsider I'd like to know other people reasons. So Reading about progress due to an approach that makes my ears stir <ng0>I don't know much about Nix, but in Guix it looks much easier for me and some other people to get started with the package/update distribution via GNUnet task on the roadmap as the language is clear and easy. that's one of main selling points for me.. you don't have to learn something new for nothing, you can use it outside of the system, unlike for example ebuild/eclass/PKGBUILD syntax <ng0>so a distro which enables you to learn a language, that's awesome. <janneke>catonano: sorry, missed that remark from boegel <catonano>boegel: can you please post a pointer to that article ? <boegel>catonano: if I can find it again... <phant0mas>janneke: have you tried build anything else with your ming branch? <phant0mas>can you try ./pre-inst-env guix build -e '(@@(gnu packages commencement) gnu-make-boot0)' <phant0mas>janneke: gnu-make-boot0 is the beggining of the guix bootstrap-process <phant0mas>to start building a toolchain which has no references to those binaries <janneke>fwiw, here is the TEST script that I used on all commits, it passes on everything except the first addition of the readline patch when it's cross building readline <phant0mas>which branch can I use from your gitlab repo so I can try it myself? <janneke>v9 is the branch rebased on master (of 2weeks ago) which does not work <janneke>v9-core is the branch rebased on core-updates, which works <janneke>if on v9 you revert 28dc10a4, then it also works <janneke>i'm trying gnu-make-boot0 as soon as my laptop stops melting from my current experiment ;-) <janneke>phant0mas: oh that was easy, gnu-make-boot0 simply downloads from hydra <janneke>/gnu/store/v37bkvhr7xhsjn54i67wmwkmlx3dyw77-make-boot0-4.1-debug <janneke>/gnu/store/9q1687didb5izajfqszkz7vzrhxyz4sb-make-boot0-4.1 <phant0mas>wait a bit so I can try `./pre-inst-env guix build --target=i686-w64-mingw32 readline' in your branch now <ng0>./bootstrap; ./configure --localstatedir=/var; make; ./pre-inst-env guix ẁhateveriwantrun` is the right way to do in a git checkout for testing? I've been away a while, and on gentoo it behaves a bit differently it seems <ng0>that's what my notes say, but I thought i ask to check <ng0>the output looked different from what i remembered <janneke>they keep changing the output, tssk, tssk <ng0>but I guess I have to debug the guix on gentoo a bit more.. i have tested the recent updates I sent in in a different way, and only applied it to git afterwards, but this one now can't connect to the socket <ng0>maybe the socket is just elsewhere <ng0>right.. no socket. is the socket default on guixsd then? on my guixsd system I see /var/run/guix/daemon-socket/socket <ng0>maybe one part of the problem I have with the initscript. <ng0>though it is defined.. strange. hrm <janneke>ng0: fwiw, on my guixsd it's /var/guix/daemon-socket <ng0>pidfile != socket, right? <ng0>libmicrohttpd update to 0.9.50 sent. <ng0>I really have to fix that openrc script.. shutdown of 10-15 minutes is not funny <ng0>that is a screenshot of a comment <ng0>i could awk http with https on the haskell.scm if this isn't due to the importer <ng0>the grep just looked long <catonano_>I had run into this scrap of text too ;-) The point about scheme being a real programming language and the uglyness of the guix own language deeply resonates within me. I had a really hard time reading that language. <ng0>I'll submit a patch for haskell.scm today which replaces where valid http with https <slim404>has anybody installed go language on guix sd? <slim404>the standard way is to put it in /usr/local <ng0>so this is going to be the longest commit message I ever wrote so far.. <stack>hi! I was wondering if the default packaging method of things in guix offers some sane config defaults for processes normally run as daemons, for example I've installed postgres, and in profile/etc/ there is no default config as normally happens with other distro, the example configurations are in share/something though <stack>I was experimenting with guix in a way that offers me different develop profiles that I can save and eventyally archive, this withouth going crazy with docker explosions when building containers or having the """slownes""" of a virtual machine <taylan>stack: guix itself (not guixsd) doesn't do much with program configuration <taylan>in guixsd, one generally writes a "service" description for daemons, and there one can create a configuration DSL, or just make it take files, and I guess offer default configs... <stack>taylan: mh ok, I was opting with guix and not *sd because I would like a handy installation with luks and a stable underlying system , now I'm using debian <ng0>I really dislike writing so many affected package names.. may I skip this by just leaving that I replaced occurences of ... in file? <stack>taylan: searching I in fact found postgres-service and wondered why I was unable to find it with guix package -s <stack>also, the documentation I found points to broken link to download the binary version <taylan>yeah, it's a dmd service, nothing guix itself can use AFAIK. (I don't use guixsd either yet, so not much experience there) <taylan>stack: are you sure it wasn't a URL template where you need to replace some word with something architecture-specific? <stack>I spent 0.5s and arrived eventyally to the correct file changing the url in the ftp <taylan>ng0: can you reword your question? I don't understand it. <stack>so what is the suggested way to deal with daemon configuration here? <taylan>stack: most programs will allow you to pass an option to specify the config file. you could just use the default config in $profile/share/..., or copy it somewhere and change some things and use that <ng0>I can understand that I should include what changed, but this is getting really l9omng for a commit message <taylan>for instance, using the nginx installed from guix, I have a systemd service file containing: ExecStart=/bin/sh -c 'nginx -c "$HOME_CONFIG"/nginx/nginx.conf -p ~/srv/http' <taylan>ng0: oh I see. hm, I don't know, maybe ask the mailing list, or civodul when he's around <stack>taylan: can you couple systemd services with a given guix profile? <ng0>I'll submitt the patch, and I can spent some hours or a smart script with diff to adjust ther patch message <stack>taylan: I imagine I will have a bunch of daemons for a given profile, and the same daemons for a different profile and maybe with a different damon version <taylan>ng0: my gut feeling says it's OK to leave out the individual changes in such a case, so I'd just send the patch without them all and ask if it's OK <taylan>in fact, as I'm getting less sleepy (just got up), I'm starting to think it's pretty common-sensical to leave them out in this case :P <ng0>2904 lines patch / email... you're welcome ;D <ng0>I have not runt this (obviously) but the haskell page puts https:// for download links, and some in the file were already https:// links <taylan>stack: if systemd files allow setting variables, I guess one could start a file with profile=/blah/blah/profile1 and such <taylan>then use that in absolute executable paths <taylan>or is that insufficient for what you have in mind? <stack>taylan: I need to experiment with something real to check if that is sufficient <stack>given I not tought of systemd user services, this would still help <taylan>I'm not actually sure if service files allow variable settings though <ng0>damn. the last patch went out from another mail account, changed it yetserday at gnu.org .. this still arrives on the list, right? <ng0>haskel.scm came from the new one, xorg.scm came from the old one (which is still functional though) <ng0>ACTION is sending in more url patches <ng0>next, and last patch for today, will change around 141 files in gnu/packages/ to add https to gnu.org url where it wasn't already in place. <roelj>ng0: Nice to see you're taking care of it :) <ng0>reasons why I should learn more emacs.. detect a line, replace regexp and then add stuff into header after last line which has elements of author email etc <ng0>everything but the header part I have. <kristofer>is there a guide to setting up mediagoblin with guix? <ng0>is mediagoblin already fully finished, into guix? <stack>is there a web interface to search packages like packages.debian.org does? <stack>random-nick: that's enough, thanks <janneke>Cannot open mailbox /var/mail/janneke: No such file or directory <janneke>Cannot open mailer: No such file or directory <kori>pressed enter too soon :P <kori>I still can't figure out the segfaulting problem <mark_weaver>kori: the relevant error will be higher up, but it is most likely due to a failure to find the root partition <ng0>last patch adds also tls for elpa, I think I'll have a look at the elpa importer after this break and add tls there. They now use a Lets Encrypt issues certificate on elpa.gnu.org <kori>mark_weaver: that was the theory yesterday <mark_weaver>kori: the code that enumerates all partitions looking for the specified label may not be finding the right device <mark_weaver>kori: you might try specifying the device name directly, instead of the label <kori>mark_weaver: via (device), yeah? <kori>I'll try installing again <mark_weaver>it's also possible you may need to load more kernel modules before trying to mount the root partition <kori>mark_weaver: could the home dir possibly interfere w/ this <mark_weaver>kori: it wouldn't cause a problem at this early stage in boot, no <mark_weaver>whether it will be a problem later, I'm not sure, but I suspect it should be fine <kori>no, I'm installing it to hardware <kori>I was just using qemu to get a screenshot <mark_weaver>kori: what kind of partition map? what kind of storage device? <kori>mark_weaver: I'm pretty sure it's GPT, and it's a hdd, not encrypted <mark_weaver>well, I can think of two likely possibilities: (1) the module for your disk interface controller is not one of the modules included in our initramfs by default <mark_weaver>can you find out the name of the kernel module needed for your disk interface controller? <kori>mark_weaver: I doubt its specifically required <kori>everything works w/ make mrproper linux-libre <wingo>mark_weaver: i don't know what that is, so probably not :) <mark_weaver>wingo: the partition table type: msdos or GUID partition table <mark_weaver>wingo: do you remember running into problems with that on GuixSD, or did it "just work" ? <wingo>for me it has always seemed to just work <mark_weaver>kori: what does this mean? "everything works w/ make mrproper linux-libre" <kori>mark_weaver: from the "default" kernel config, I don't need to enable anything for my disk interface controller to work <mark_weaver>kori: what "default" kernel config you're referring to? when building manually from upstream linux-libre? most distributions have their own kernel configs, and GuixSD is no exception there. <kori>via getting the source and running make mrproper on the tree, which removes any sort of config that might have come with it <mark_weaver>kori: well, that's not the config that we use in GuixSD <kori>aye, there are a lot more modules there <mark_weaver>if you'd like me to try to help debug your problem, it would be helpful to know the kernel module needed for your disk controller <mark_weaver>I can find out if that module is included in our default initramfs <mark_weaver>and if not, I can tell you how to add it to your OS configuration <kori>mark_weaver: how can I figure that out? <mark_weaver>kori: okay, your SATA controller needs the ahci module, which is already included, so the problem must be elsewhere <mark_weaver>kori: if you haven't already done so, I would suggest trying to specify the root partition by device name instead of label <kori>mark_weaver: I've tried that, yeah <kori>well, I'm going to try it again <catonano>ng0: are you here ? I read now that you are picking up the gnunet work again. Good to know <ng0>but in parallel I'm still figuring something out for gnunet.. we now have a transition repository where we'll work on, this could also have openrc service files for packages we once had in the gentoo overlay <ng0>and also started discussing about a guix pull & distribution through gnunet-fs .. etc <ecraven>should lsh-authorize be in the default installation? I can't seem to find it :-/ <taylan>just as I'm done with higan, version 099 gets released. and big news: the "balanced" emulation core has been removed; only "accuracy" remains. I discovered my CPU is fast enough when I set the frequency governors to "performance", but on "powersave" the laptop is too slow. this is an i5-2520M (ThinkPad T420). I wonder whether I should leave Guix's higan package at 098 with the balanced core, <taylan>or update it to 099 at the possible expense of guix users with older hardware. <ng0>how are the source urls from elpa import constructed? is it up to passing the source url that it gets https:// or http:// ? I want to see where I can enforce that it is https://, but elpa.scm in the import folder doesn't look like it <ng0>taylan: what about making 099 the current, and 098 the inherit? <kori>mark_weaver: I just booted successfully <kori>However, this was after removing xmonad from my config <ng0>yes, the imports look to me as if they are not restricted to the protocol, hackage is the same <taylan>ng0: I guess I could do that as well <galex-713>taylan: so are there anything else than linux? else than hurd? <ng0>for imports I guess if it isn't broken don't fix it? however if the guix build system would prefer tls where possible to download it would be good <random-nick>galex-713: it's a Google Summer of Code project this year <ng0>it's facinating watching other gnu distributions discussing what guix and nix already solved <ng0>[gentoo-dev] the future of the sunrise project / fascilating user contributed ebuilds .. <ng0>at least one gentoo dev feels strong about nixos and takes distrowatch as a real source for showing that "lol nixos sucks" o.o <catonano_>wat's the situation of golang in Guix ? I see no packages containing the string "golang" in their names <wingo>gccgo is packaged, there are draft packages for go 1.4, 1.5, and 1.6 <wingo>see gnu.org/software/guix for a link to the repo <wingo>it is hosted on gnu.org, not github <yoosty>boegel: git://git.sv.gnu.org/guix.git <boegel>yoosty: so, the Guix on GitHub is just a wild mirror? <Acou_Bass>slight dumb question that im not sure if this was answered already but couldnt find mention of it... if i add the gnome-desktop-service in my desktop.scm, does this switch my login manager to GDM? If not, how do i go about this? <wingo>so you still have whatever dm guixsd uses by default <Acou_Bass>its in the gnome.scm but doesnt seem to be a service file for it i suppose <wingo>you select gnome at the greeter login using f1 <Acou_Bass>no worries, i doubt my laptop can run gnome very well anyway, so only gonna test it quickly =) <Acou_Bass>i tried nixOS recently by the ways... i was expecting pretty much the same as guixSD except less freedom-loving... was surprised to see how much different the two systems are <Acou_Bass>i mean obviously its the same sort of idea in terms of the purely functional packaging and the config file you build the entire OS from... but still <wingo>anything specific strike you? <Acou_Bass>the nixos equivelent of guix pull isnt a per-user thing for a start <Acou_Bass>like, on guix, i have to do it for my own user, and my dads user.... but on nix its all the same stuff so just pulls it all into one dir <Acou_Bass>slightly more obvious points are the systemd vs. shepherd, and the... whatever weird language .nix is supposed to be vs. scheme <Acou_Bass>i have no idea what .nix are actually written in... it sort of looked like lua but weirder <mark_weaver>Acou_Bass: if you prefer to have one user use the guix from another user, we can do that too <mark_weaver>Acou_Bass: for example, if you want user 'foo' to use the same version of guix as user 'bar', then make ~foo/.config/guix/latest be a symlink pointing to ~bar/.config/guix/latest <mark_weaver>note that "guix pull" sets that symlink to point to a freshly built guix in /gnu/store <mark_weaver>that symlink can also point at a built git checkout of guix. that's why several of us developers do, including me. on my system, both /root/.config/guix/latest and /home/mhw/.config/guix/latest point to /home/mhw/guix which contains a git checkout of guix 
