<cbaines>interestingly it seems like the downtime effects a lot of FSF hosted stuff, e.g. gnu.org and fsf.org too ***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | guix pull servers are currently down | 1.0.1 is out! get it at https://guix.gnu.org | videos: https://guix.gnu.org/blog/tags/talks/ | bugs and patches: https://issues.guix.gnu.org | paste: https://paste.debian.net | Guix in high-performance computing: https://hpc.guix.info | This channel is logged: http://logs.guix.gnu.org/'
***ChanServ sets mode: -o nckx
<nckx>"FSF services are temporarily down for maintenance. Thanks for your patience!" *vagrantc wonders about high availability with guix <reepca>I wonder how much space all sources for every package in guix would take, cumulatively... <reepca>just in case I one day want to go full hermit mode <reepca>vagrantc: guix build: error: .: unknown package <nckx>reepca: They should fit in 100 GiB, although that info stale af. <idnull>it seems that this problem persists In procedure module-lookup: Unbound variable: openscenegraph (I copied suckless.scm to my repo) channels.scm http://dpaste.com/01K2A03 <jlicht>idnull: you are missing %default-channels in your declaration, AFAICS <jlicht>I have something like (cons* (channel ....) %default-channels) in my channels.scm, FYI <mbakke>idnull: I believe the openscenegraph problem should be fixed with commit 84a0c6ff63afc296c1bc5873f783693dccc5ded2 ***ChanServ sets mode: +o nckx
***nckx changes topic to 'GNU Guix | 1.0.1 is out! get it at https://guix.gnu.org | videos: https://guix.gnu.org/blog/tags/talks/ | bugs and patches: https://issues.guix.gnu.org | paste: https://paste.debian.net | Guix in high-performance computing: https://hpc.guix.info | This channel is logged: http://logs.guix.gnu.org/'
<reepca>Welp, reconfigured with gdm and sure enough I can't start a graphical session now. 'New session c506 of user gdm', yikes. Meanwhile whenever I booted with sddm recently, I had to ctrl+alt+f2 my way to a tty to restart it, since it wouldn't start on its own (though herd reported it started). So I guess I've come full circle now, back to slim! ***ChanServ sets mode: -o nckx
***Digitteknohippie is now known as Digit
<retropikzel>I have distorded audio in icecat, any ideas what could help? <shrdlu68>Hello world, I believe I've found my next distro. <rk4>(suggestion (topic #guix-converations) (body should be conducted in s-exps)) <shrdlu68>How do I install the system without a USB, DVD, or PXE? Is there something like Gentoo's Stage3 tarball? <rk4>if you can access the boot disk and installation disk, while running in a linux distro, i believe you can do an install <shrdlu68>Scenario: I want to install Guix System in a cloud VM, where I can't boot from the Guix iso. However I can boot the VM in "rescue mode", which runs some distro like Ubuntu, where I can partition the file system and install Guix System ***ng0_ is now known as ng0
<rk4>when i did that i used a guix live image, but i don't recall anything remarkable about it, the main advantage was that the guix package manager was already installed <shrdlu68>The manual says that the the command above "This makes /gnu/store copy-on-write, such that packages added to it during the installation phase are written to the target disk on /mnt rather than kept in memory. This is necessary because the first phase of the guix system init command (see below) entails downloads or builds to /gnu/store which, initially, is an in-memory file system." <shrdlu68>I do have the guix package manager, after running the installer script referenced in the Manual. But I don't have herd. <rk4>see i thought you could just mkfs, mount, guix system init blah.scm /mnt, but i don't know enough to tell if that step is essential. :( wait a while, i'm europe is waking up :) <ramajd>Hi, does anyone knows that `nslookup` is located in which package in guix ? thanks <pinoaffe>retropikzel: I also have distorted (seemingly lowpassed) audio in some media players in icecat, particularly the twitter one, but I have no clue what could fix it <pinoaffe>I've just been playing twitter videos using mpv as a workaround <retropikzel>pinoaffe: I usually play all my videos with mplayer or such, but countrys public broadcasting companys live radio is distorted. Not all audio, so I would guess it has something to do with codecs maybe? <pinoaffe>it might, or it might be one of pulseaudios "special" features <pinoaffe>shrdlu68: you can install guix (the distro) using guix (the package manager) from nearly any linux system, in a similar manner to how one would bootstrap, say, a debian system <shrdlu68>pinoaffe: So `herd start cow-store /mnt` is not essential? <roptat>nckx, no idea what went wrong, I didn't touch it <roptat>maybe a filesystem issue, or a gc that didn't work as expected? <shrdlu68>pinoaffe: So how do I bootstrap Guix System from an Ubuntu Live disk? <pinoaffe>shrdlu68: it is not documented nor supported, but it should be possible if you get both herd and guix set up on the ubuntu system <shrdlu68>pinoaffe: A tarball that I can untar onto / and chroot from there, Gentoo-style, is the only other option I see. I've tried `guix system init /mnt/etc/config.scm /mnt`, but I end up running out of space, since the filesystem is on the livedisk. <pinoaffe>huh, so why aren't you mounting a proper blockdevice onto /mnt? <pinoaffe>I guess it might be possible to prepare a tarball on another machine, but that might give ya some issues with BLKIDs and the like <roptat>pinoaffe, the issue is that guix system init fills /gnu/store, and then copies it to the target <efraim>are ilmbase and openexr even supposed to be installed separately? <roptat>pinoaffe, I think that's the goal of cow-store: instead of installing on /gnu/store, it installs on /mnt/gnu/store directly, but can still access it as /gnu/store <roptat>but you can probably replicate the behavior of cow-store manually on the command line <pinoaffe>or maybe do some mounting tricks or even a pivot_root to ensure /gnu/store is backed by a blockdevice from the get-go <idnull>hello once again, I've manage to build my packages, but I have one more question. I need to put config.h in the build dir (for dwm/st)? How can I do this? <idnull>lambda*: invalid argument list in subform (#:allow-other-keys) of (#:allow-other-keys) <pinoaffe>idnull: I don't see what you'd be doing wrong in the lamba* definition, but you're not supposed to add a string describing the location of your desired config.h, you're supposed to add guile that modifies the tarball, adding your desired configuration <lprndn>idnull: I might be wrong but if you don't need any keys you could go with (lambda* _ ) <pinoaffe>what you did now is add a phase that does nothing but return a string <pinoaffe>and you could indeed probably just do (lambda _ ) <idnull>pinoaffe: how can I make a substitution of a config.h then? <oldosfan>does anyone here know when Guix is going to bundle a newer version of GNOME? <civodul>oldosfan: i think there was work towards an upgrade, but it takes time <oldosfan>perhaps, if I can get building GNOME from source to actually work on Guix lmao <civodul>well the package definitions are already there <civodul>so it's a matter of updating them and testing things <shrdlu68>I seem to have successfully done `guix system init` by chrooting. Haven't rebooted yet, though. <roptat>there's a new version in core-updates, no? <roptat>are we going to merge core-updates soon? :) <roptat>also, I think there's wip-gnome-updates or gnome-updates branch <idnull>ambiguous package specification `dwm choosing dwm@6.2 from gnu/packages/suckless.scm:79:2 <idnull> how to choose from my repo not guix one? <lprndn>idnull: I would suggest you to change the name of your custom package ;) <idnull>And while I'm here, I want to build st from master, can one skip sha256 somehow? <roptat>if st has a source from git, --with-branch=st=master I think <roptat>that should skip hash validation <g_bor>:) jsut bumped into the next home-manager thing :) <roptat>not an issue for me: I use a USB key, so I just set up a link from ~/.gnupg to a directory where the key is always mounted <idnull>ty:all: for help I'll try later rebuilding from master without sha256 once again later <g_bor>right now it fails with can't create directory ~/.gnupg <roptat>as usual, you can create a symlink to your data directory <g_bor>:) yes, but I would like to this without it if possible <roptat>.gnupg is supposed to be at least partially mutable <jonsger>oldosfan: gnome3.30 is in the core-updates branch and should arrive in the next days in "normal" guix :) <g_bor>yes, I am aware of that... I would like to go as far as possible with this. <roptat>if you can come up with a home definition, it'd be great :) <g_bor>roptat: I'm on it. Finally I would like to do something like %guix-development-environment, and just having to put the gpg and shh key in place,and have everything else working out of the box. <idnull>it only took me two days (i'm a little slow yeah) <g_bor>roptat: what would be an easy way to create an empty directory in the resulting profile, ala mkdir-p? <g_bor>It looks like to me that is not exported directly. I guess it would help while developing. <g_bor>Maybe I am missing something... <roptat>g_bor, the root-home-type doesn't support that directly <roptat>you can symlink to an empty directory though <g_bor>roptat: an make that read only... neat trick <roptat>you can use a file-like object too <idnull>jonsger: this was the easiest part: now I need to copy new moon, slstatus, wireguard and find a way to make a theme package (and set this theme as a transaction) <roptat>like (computed-file "empty" #~(mkdir-p #$output)) <g_bor>roptat: I was aware of that, thanks. <civodul>i haven't watched it yet but it might be of interest <civodul>you should blog about Guix Home, BTW! <roptat>didn't watch it yet, but that's how I found the other video :) <roptat>yeah... now is not a good time for me <roptat>a few other videos from that conference look interesting <shrdlu68>Can't seem to find my hard disk after rebooting. Am I missing kernel modules? <shrdlu68>I'm trying the "bare bones" configuration. <roptat>shrdlu68, probably, but guix should have warned you during guix system init <civodul>roptat: i saw that and i wasn't convinced that "recursive attribute sets in Nix" were a good fit for an intro to FPM :-) <civodul>it's as if we were discussing syntax-case in an introductory talk on Guix ;-) <g_bor>roptat: I guess I made some mistake somewhere... I now get guix home command not found. <g_bor>I remember it was working earlier though. <roptat>g_bor, check your GUILE_LOAD_PATH <roptat>did you change anything in guix/scripts/home.scm? <g_bor>The laod path only has the /run/current-system... <shrdlu68>roptat: I didn't get a warning, althoug I'm doing all this in a chroot. <roptat>g_bor, there's a bug in guix where it doesn't load the channel before checking for subcommands, so you have to set GUILE_LOAD_PATH <roptat>shrdlu68, maybe that's the reason <roptat>shrdlu68, you should boot on another distro and check what is needed for your disk <g_bor>roptat: how do you manage that? Do you st it up from a shell startup fiel? <g_bor>Ok, let's try it again: do you set it up from a shell startup file? <roptat>currently I do it manually, but I could add it to my .bash_profile <roptat>(which is a file-like object, by the way :)) <g_bor>:) I was also thinking about that. <g_bor>I remember talking about this before, but I just can't find it right now. What is the correct value? <shrdlu68>Does someone have the .config of the default Linux-Libre kernel? <roptat>just add (simple-file-home bashrc ".bashrc") to the list <g_bor>roptat: what is the correct value of GUILE_LOAD_PATH? <roptat>it should contain an item that points to your guix home git checkout, or to ~/.config/guix/current/share/guile/site/2.2 <g_bor>ok, thanks. just what I was looking for. <g_bor>roptat: one more question, why did you use computed-file for bashrc? It looks like a plain-file? <pkill9>wow the silver searcher is a lot faster than grep <pkill9>is any work being done on making the create importer able to import recursively? <jonsger>shrdlu68: gnu/packages/aux-files/linux-libre/ there you should find them <jlicht>pkill9: I think there is a patch floating somewhere that does exactly that <jlicht>I recall it being functional with some rough edges <jonsger>pkill9: I think efraim has something for that <pkill9>it just installs the license files <efraim>the 'import crate foo@version' and import crate foo --recursive' got approved together and foo@version got merged first, the recursive one wasn't rebased <efraim>pkill9: are you sure there's a binary? <pkill9>there should be, I'm trying to check with the -K option, but I can't cancel the build in time <efraim>looking at the source it looks like it's a library <efraim>also i'm not sure the recursive importer makes sure it imports the correct version <atw>Hello, I am trying to move everything that's not an implementation of lua into lua-xyz.scm (this is because I plan to add some lua libraries and this seems like best practice). But, when attempting a clean build, I get a confusing failure: https://paste.debian.net/1102524/. I'm stumped! <roptat>atw, did you forget to import (gnu packages lua) in (gnu packages lua-xyz)? <roptat>I think lua-5.1 is used somewhere where it's not imported <atw>hmm, good thought, I'll just grep everywhere for it <roptat>the error with zip is just a side-effect <roptat>I get it every time I make a syntax error or there is an unbound variable <atw>every occurrence of "lua-5.1" I could find is in a file with the appropriate "#:use-module (gnu packages lua)" <roptat>did you transfer its definition to lua-xyz by mistake? :p <atw>just checked again and I did not <atw>roptat: I've got to run to work, would you leave a message with sneek if anything occurs to you? TIA! <mbakke>efraim: do you remember why freeimage was reverted to use the bundled libjxr? (commit f347c24acc14e080dc2801561edca0d525a90257) <roptat>looks like what I used for my own website :) <jonsger>civodul: and what happens if we have 11 parens :P <jonsger>but serious, this is nice and even without JS :) <efraim>mbakke: wasn't that a while ago? Likely it either didn't compile correctly or it wouldn't actually link to it, probably the later <efraim>I'd be tempted to merge ilmbase and openexr into one package since the double inclusion of headers drives freeimage crazy and openexr already propagates ilmbase <pkill9>did the Guix manual have a margin before? <civodul>jonsger: ten parens should be enough for everyone :-) <civodul>pkill9: the manual had a margin but i've made it wider recently <civodul>rather, i've reduced the max-width of the body <pkill9>ah, it feels nicer I think, it's easier to read <mbakke>efraim: wdym by driving freeimage crazy? <civodul>davexunit: perhaps we should provide this syntax-highlighting mode directly as part of your package? <mbakke>efraim: it's from april last year... AFAICT none of the freeimage dependents work since the unbundling <efraim>Hmm, the unbundling was so it would build on arm and aarch64. If none of the dependents work that's a problem <efraim>I'm away from my computer ATM, I thought the freeimage issue now was the double declarations between the headers ***nckx is now known as bug_bot
***bug_bot is now known as nckx
<idnull>is it possible to bundle service with a package? I want to package iNET Wireless Daemon (it can substitute wpa_supplicant and works with connman and NM) <roptat>you typically define a service that uses a package, not the other way around <idnull>ok, so firstly I create a package, then service, then I put package as a service dependency? <roptat>as a simple example, you can have a look at the opensmtpd service for instance <jetomit>ell should be OK, I haven't figured out the dbus bits for iwd yet <idnull>New Moon is working, but with an issue tar (child): palemoon-: Cannot open: No such file or directory <idnull>'t do something with archive (maybe create one) <nckx>idnull: You can try strace'ing New Moon (or its children). <rain>heyyy, so, is there a nice way to make the regex in substitute* into a literal? like, all the special characters that have special meanings in regexes, i wanna opt out of them. <nckx>rain: So like a REGEXP-ESCAPE procedure? I dunno. <rain>nckx, hm. then i guess i'll just have to figure out scheme regexes. :/ <nckx>Welp, it's literally called regexp-quote 🙂 <nckx>rain: ‘Scheme regexes’? What's different about them? <rain>nckx, oh nice, it exits. well, i'm not sure if they are compatible with any regex i know. (eg. Lua has a very different (and simpler) regex implementation compared to Perl) <nckx>I've always considered non-Perl regexes to be ‘normal’, and Perl regexes to be, well, not. <nckx>Doesn't Lua call them patterns, not regexes, to avoid exactly this confusion? *nckx agrees there are too many ‘regexes’ though. <nckx>Guile's regexes are bog-standard to me but it's the double-escaping that makes them fun™! <rain>nckx, that looks like a group of people raising their hands (like the \o/ smiley) <nckx>I thought of a marathon runner crossing the finish 🙂 <nckx>‘Yay, finally wrote a regexp that works.’ <nckx>Or a (match-regexp (perl-regexp "…")), (scheme-regex "…"), … syntax. All the regexes in one codebase! Imagine the fun. <nckx>civodul, rain: Thanks though, I had no idea there was really such a thing as a ‘Scheme regexp’ (sigh). <rain>welp, i'm trying to escape but it seems to affect an (assoc-ref) call somewhere for... some unknown reason... <rain>nckx, i think regexp-quote isn't quoting something properly: (make-regexp "LIBCLANG_LIBS += $${CLANG_LIB}" 1) Invalid preceding regular expression <nckx>(regexp-quote "LIBCLANG_LIBS += $${CLANG_LIB}") → "LIBCLANG_LIBS [+]= \\$\\$[{]CLANG_LIB[}]" <nckx>Which make-regexp accepts without complaint. <nckx>rain: Not saying you're wrong, but I don't see what led you to that diagnosis. OTOH, I'm working, not in the backslash-counting zone. <nckx>Does it work outside of the (substitute* …) syntax? In a (let …)? <nckx>`(,( sometimes fucks with macros in ways that aren't entirely clear to me. <rain>nckx, it doesn't seem to :/ <rain>but i gotta go, class is over <rain>thanks for the help! ^u^ <nckx>You're welcome. Good luck. <nckx>sneek: later tell rain: It's always helpful to include the actual error message in addition to your code: after downloading & building your package, Guix said: Wrong number of arguments to #<procedure assoc-ref (_ _)> <nckx>sneek: later tell rain: which 1) is suspiciously relevant for a Scheme error message, what's going on and 2) immediately makes the obviously wrong ‘(assoc-ref inputs "clang" "/lib"))’ obvious. *nckx is unsure whether rain2 is rain. <sneek>Welcome back rain2, you have 2 messages. <sneek>rain2, nckx says: It's always helpful to include the actual error message in addition to your code: after downloading & building your package, Guix said: Wrong number of arguments to #<procedure assoc-ref (_ _)> <sneek>rain2, nckx says: which 1) is suspiciously relevant for a Scheme error message, what's going on and 2) immediately makes the obviously wrong ‘(assoc-ref inputs "clang" "/lib"))’ obvious. <nckx>Well, you fooled sneek 😃 <civodul>nckx: srfi-115 is not really about "scheme regexps", it's about providing a non-stringy DSL for regexps <civodul>thus without ambiguities, special characters, special characters that aren't that special, and all that <nckx>civodul: OK, now that sounds sane. <nckx>As in not reinventing/forking regexpen yet again. Whether '(+ (or space punct)), '(* digit), or '(: "*" ($ word) "*") are more or less readable, well… <nckx>I obviously still need a bit more time in the Scheme re-education mines. <nckx>I got to the ‘SRE Syntax’ syntax header right after typing that. <nckx>civodul: Right, but I assume your ‘without ambiguities, special characters, special characters that aren't that special, and all that’ advantages go back out of the window then, no? <nckx>It's interesting but I don't grok the ‘rock’ yet. I'm gonna read this straight through when I have the time to do so 🙂 *civodul regained SIEVE access, mail filtering should improve \o/ <refpga>Hello, I need to edit the source code of a package before installing it (using the same compilation process). Any ideas on how to proceed? ***Digitteknohippie is now known as Digit