<Petter>I assume those that can answer you are not here right now. <Petter>I would suggest trying again later, or writing an e-mail. <emas>it's does not matter much tbh, but i thought that it will be easier to get answers from here instead of having to read reading some stupid long manuals <ZombieChicken>So there is no way to install GuixSD onto a LUKS-encrypted partition? I can't figure out how to get GRUB to install in this VM, and it's using a drive in /dev/mapper/disk ***zimmermann_ is now known as zimmermann
<pksadiq>is it safe to install guixSD 0.11 on bare metal? I wish to try it out (I don't care grub rescues, I care about data loss and bricking hardware pieces) <pksadiq>can't guix be installed on extended partitions? <pksadik>I'm currently trying to install guixSD. By 'device' in 'file-system' does it mean something like /dev/sda3 (the device which my root dir shall reside) <jmd>pksadik: I'm not sure that I understand your question. <pksadik>in the installation notes I see a text "my-root". What should I replace with it? <pksadik>Sorry, its a bit hard typing on phone. Wish I have emacs in the live guixSD boot <jmd>Wait while I pull up the installation instructions. <jmd>I don't think you can install GuixSD on a phone. <jmd>So you have partitioned the disk and created a file-system and mounted it under /mnt ? <iyzsong>pksadik: depends on the 'title' field, which can be 'label or 'device, your device field should be the label or path for the partition. See "File systems" in the manual. <jmd>and you have copied the template config.scm into /mnt/etc/config.scm ? <pksadik>i copied desktop.scm as said in manual <jmd>ok. Now you have two options: 1) you can set the title field to 'label and put the label of the filesystem in the device field; OR 2) set title to 'device and put the device name in the device field. <jmd>Note that depending on how you created the filesystem, it may or may not have a label. <jmd>I think most people prefer option 1, because that way the system remains bootable if the device enumeration changes. <jmd>pksadik: So to answer your earlier question, you can replace "my-root" with anything you like. <jmd>(well almost anything) <jmd>The System Installation Instructions don't mention how much disk space is needed. I wonder what the minimum is? <ng0>less than 12 GB I suspect <jmd>We ought to gather some ballpark figures. <civodul>"guix size $(guix system build bare-bones.tmpl)" <civodul>but yeah, we should put figures in the manual <janneke>jmd: my ballpark would be: twice of what you use for your debian system <janneke>my current advice would be: use your whole disk :-) <jmd>I've started work on a curses-based installer. So it would be good if it could suggest test for filesystem size. <jmd>janneke: I have a machine with a 30MB disk. <janneke>jmd: wow, I once had a machine with a 5MB disk, long ago that was *very* large <jmd>My first machine didn't have a hard disk. just 2 floppies. <ng0>static linking and musl / uclibc could be of help. I'm workingon uclibc-ng but this is paused for other items preference now, will pick it uo though <ng0>but static will not be possible afaik <ng0>barebones: total: 676.2 MiB <iyzsong>yeah, it requires more disk space than Debian, ArchLinux, etc. and it grow fast without gc :-) <janneke>jmd: but indeed, you'll have a challenge there... <janneke>/gnu/store/dpnl9m7d7ylrrp3si6x19lwqcbgms9gz-linux-libre-4.7.3 201.6 201.6 100.0% <jmd>iyzsong: Sadly even with a gc. <jmd>Currently there is no way to remove old configurations. <ng0>it will be a challenge to reduce the size for the live-system i work on <ng0>i think 4GB is probably okay these days <iyzsong>jmd: system generations? it can be remove manually by root, then a run of gc will free that. <ng0>the smallest usb disks you can buy are 16 and 32GB anyway <jmd>iyzsong: Possibly. how does one remove them manually? <iyzsong>jmd: delete old 'system-xx-link' in /var/guix/profiles, be careful not delete the current one. <jmd>"disks" are soon going to be found only in museams. <jmd>iyzsong: which is the "current one" ? <iyzsong>there is a 'system' which point to the current 'system-xxx-link'. <janneke>iyzsong: yes, also i think current users of GuixSD will want to contribute and build quite some packages that do not have substitutes yet <ng0>jmd: disks for a reason <ng0>jmd: there'll be other mediums it will be used on though i think <janneke>does anyone have a working setup of guile-wm yet? <ng0>can someone review the new net-lib-psyc series next week, but don't do the psyced packaged. I'm testing on package and service currently and made a mistake in license etc <ng0>everything is very low in priority for me, but everything psyced and gnunet related is important to move on with what I'm working towards <iyzsong>janneke: agree, and I feel see so many (and growing) patches in guix-devel waiting for treating :-) <jmd>We should use aegis for developing Guix. <jmd>Great for managing reviews. <ng0>if we get to the go-build-system and rust one, we can package gitlab. i ncase that drowned somehow. <ng0>and with cargo, I can finally get to work on our projects new prototype version, and also package panopticon, a reverse engineering software which got me on this rusted mission earlier this year :) <ng0>i think the criteria for switching was whatever we switch to it has to be packaged for guix? <rekado>speaking of aegis: I think it fails to build on Hydra (at least for some architectures) <ng0>ah, it is already there. good :) <iyzsong>um, aegis look too professional, a bit scare for me :x <rekado>rebuilt all Haskell packages without errors. Yay <ng0>did you take over updating ghc? <rekado>just removed all propagation from all packages. <rekado>one thing to note: the importer should refuse to add packages to inputs that come with GHC. <rekado>seems that bytestrings are provided by GHC already <rekado>adding ghc-bytestrings to package inputs explicitly will lead to problems <rekado>now that all haskell packages are clear about their inputs updating should be a little more straight-forward. <ng0>yes, I've seen that while working on darcs. confusing. <rekado>the hardest part is writing the git commit message… ***[0xAA] is now known as Zer0Pings
***[0xAA] is now known as Zer0Pings
***[0xAA] is now known as Zer0Pings
<civodul>i think we'll be ready to freeze core-updates <pksadik2>I can't umount /mnt. is there any bg process running? may I just restart? <iyzsong>yeah, iirc, the cow-store service uses /mnt, it will be unmount by shepherd when reboot. It's safe. <pksadik2>oops... I did herd stop cow-store, and the partition usage has now reduced by 2 GiB. <iyzsong>run the 'reboot' command, if it doesn't boot.. then install again :- <ng0>gnunet test suite could be fixed in latest commit.. I'll test either today or on sunday. I was just slow with getting a core dump, but someone else had the exact smae problem :) ***pksadiq is now known as Guest32424
***Guest32424 is now known as pksadiq`
<pksadiq`>Though I ran into a few issues, mostly my mistakes. <pksadiq`>So now, How can I get gnome-shell up and running? <ng0>i think it was as easy as including gnome-desktop-service in your config.scm of the system and running guix system reconfigure /path/to/config.scm .. i think, because I'm not focused on this channel at the moment :) <catonano>ng0: if someone else had the exact same problem, then the Guix machinery wasn't at fault. Was it ? <ng0>In all the time i do gnunet for gentoo, i never enabled the test suite there... i assume that the problem was not exclusive to guix ;) <ng0>i mean.. even without any build system, building as user, this problem appeared. <ng0>currently building the latest revision <ng0>so far everythingf passes <ng0>i'm working on a guix.scm for gnunet svn.. so that you can just use the latest svn head, included in the source. <catonano>ng0: I hope they will include it upstream. <ng0>I include it upstream. <catonano>Wonderful. I can't wait for exploring Gnunet <ng0>sometimes it's diffcult to understand when I say 'for' when it can be understood in many ways.. I work primarily as part of Secushare and GNUnet, Guix just happened as a part of a subproject I run on my own for Secushare <ng0>damn, 1 test is still failing <ng0>nope.. didn't fix it. back to mantis. <ng0>catonao, 0.10.2 is now scheduled for 2016-11-01 if everything goes well.. at least the roadmap got updated to this date :) <snape>There seems to be a bug with emacs 25 and geiser. When I do geiser-repl-import-module in a Geiser REPL, and then <TAB> to auto-complete, I get "No Geiser REPL for this buffer (try M-x run-geiser)". <snape>well, actually, it is not guix related, sorry for the noise :) <snape>guix system reconfigure <your-config.scm> does it <pksadiq>snape: it warns me that I may be downgrading my system? my guix is working fine. I need to make grub find other installed OSs <snape>maybe you did not run "guix pull"? <snape>that would be why it warns you about downgrading your system <snape>and... I don't know how to make grub find about other OSs <pksadiq>snape: Hm.. I copied grub config from my old grub.cfg. Let me try if it's working. <adfeno>Do any of you use Haunt to provide a website? <adfeno>Strange, I'll make a quick test. <adfeno>ERROR: no code for module (system reader) ... when running `haunt build`. <adfeno>Or even when simply openning `guile` and doing `(use-modules (system reader))`. <adfeno>I'll try `guix environment --pure "haunt" "which"` to see what is happening. <ng0>dat commit message for ghc :O <davexunit>adfeno: there's a patch of the mailing list about this issue <davexunit>GUILE_LOAD_PATH and GUILE_LOAD_COMPILED_PATH needs to be set appropriately <davexunit>this is a flaw in haunt's build system that I'm not sure how to fix. the binary should ensure that the dependencies are on the load path <adfeno>Actully I'll try replacing "haunt" for "guile" instead. <adfeno>Let's see if I can help a little <adfeno>davexunit: I don't know how to do it, but: Can't we use wrap-program to do that? <adfeno>As I can see, wrap-program comes from "guix build utils" module. <adfeno>Of course, we could also set the variable to the user profile itself, but this would also probably require a change in documentation. <OrangeShark>`guix environment --ad-hoc haunt` is what you want then <ng0>guix.scm could be done any minute.. i'll push an not very polished version to svn later today. <adfeno>Hm... interesting, I can talk to you all now, but my browsers can't seem to access the Internet. <ng0>catonano: what do you mean? <catonano>I mean I didn't know that the roadmap had been updated and I acknowledge that <ng0>catonano: Revision 38060 adds a very first rough undocument, with tests enabled, guix.scm to gnunet.org/svn/gnunet <ng0>ideally there'll be more documentation (see mediagoblin) and the ability to launch a guix-env. <ng0>thanks to christopher webber ;) <catonano>what do you mean qith "see mediagoblin" ? <ng0>source of mediagoblin. <ng0>i had something like this in work when I ran into the hash problem.. this is now solved, just need to make my file cleaner etc <adfeno>davexunit: I managed to set GUILE_LOAD_PATH and the _COMPILED_ counterpart in my .bashrc (because I don't feel confortable doing to in .profile yet). <adfeno>I just did `haunt serve` and visited the site, and it's working, but... How do I stop the server now? <adfeno>The terninal is busy, but I can't do Ctrl +D, nor Ctrl + C. <adfeno>I guess only sending SIGTERM through a task manager is possible for now. <adfeno>Sounds good to me anyways, avoids accidents. :) <rekado>does anyone know the status of our Python upgrade to 3.5? <rekado>The latest version of Blender can only be built with Python 3.5 now. <rekado>adfeno: have you tried hitting Ctrl-C twice? <rekado>adfeno: it’s a common problem with the web server in Guile. <rekado>the first Ctrl-C appears to be swallowed <rekado>I only know this because I have the same problem in my application using the Guile web server :) <efraim>rekado: python-3.5 is in core-updates <rekado>efraim: oh, thanks. I didn’t find it… <rekado>oh, I looked in my outdated local branch <adfeno>rekado: I did some research, and mostly, the interpretation of double-SIGINT (Ctrl + C twice) is application-dependent. <adfeno>But most cases, the first one is traped by the usual trap set in the application, and if the post-trap code is really messed up, a second SIGINT exits the trap. <adfeno>In the recipe actually, not artanis itself. <adfeno>What's the difference between "%outputs" and "outputs"? <rekado>adfeno: “outputs” is a regular variable that is bound in (lambda* (#:key outputs #:allow-other-keys) …) <rekado>“%output” is special, which is indicated by the leading % <drakonis>so it appears that all the guix non free repos might have went the way of the dodo and i can't seem to get the install image maker to work <drakonis>something something unbound variable login-service <ng0>ah, i think i remember this, some software <drakonis>the scheme that writes the usb install binaries <drakonis>and aren't you the gentoo dev person that was talking about transitioning some gentoo things to scheme or guix? <ng0>i'm just stuck with gentoo for package maintenance reasons, that's all.. I would like to see something like pfl in guix, probably easy doable as we already have something similar. <ng0>actually i have pfl working here, but without portage it is much less functional <ng0>you can query some results, but not all the files. <drakonis>i'm trying to replace the libre kernel in the install image, as it doesn't load here <ng0>i use gentoo to build gnurl in addition to guixsd, and i maintain an overlay, though the offer in gentoo direction still stands to get conversation going to include the software in portage.. but i like the collaborative maintainance i do in the overlay <ng0>if you succeed with that, do you mind sharing your results off-list or some place? I might need to consider this for the system i am working towards. <ng0>youbroketheinternet-overlay, located at gnunet.org <drakonis>i'd actually like a alternative to guix's libre kernel, as it limits the available hardware way too much right now <ng0>I need to build a system which is targeting the sad reality which does not take hardware freedom in respect always.. i have to think globally, on a very large scale of audience i need to reach <ng0>ZombieChicken: yes. although guix is functional the service needs to be fixed. <ng0>otherwise expect 30 minutes shutdown :D <drakonis>a opposite to what fsf's endorsed distros usually represent <ng0>it's the worst openrc i ever wrote <ng0>lynX has improved my failing package, so it's also his effort. collaborative. <drakonis>i found a scheme that produces a linux kernel <ng0>haven't tried nix, but we also have that now. <drakonis>nix is the "predecessor" to guix, in a manner of speaking <ng0>I mean, i haven't tried the ebuild. <ng0>I use nixos, slowly getting into it. <drakonis>i'm going to install nixos if i can't replace the kernel <drakonis>but i figure that shouldn't be a hard task, as i found someone else's scheme for this <ng0>where did you find it? github? <ng0>okay.. it's further down the road for me if i need to consider this at all. i have multiple options :) <ng0>I'm working on a live-system for a project.. there's a roadmap etc but the tl;dr is that most of it will end up in their upstreams, currently that is guix <drakonis>gentoo has been lacking in that department for some time now