<vagrantc>i've sometimes wondered about re-implementing libreboot in guix ... so rather than a boatload of patches and shell scripts, you'd have a boatload of patches and guile code... :) <apteryx>oh, is libreboot like linux-libre, taking some upstream and stripping the non-free bits from it through patches/scripts? <vagrantc>in my experience, it also showed more information to the user, rather than just a white screen that flashes by in 3 seconds, it actually displays the text for the various key-combinations that do things <nckx>vagrantc: That's just a setting though. <vagrantc>maybe so, but there's a lot to be said for default behaviors :) *nckx prefers the silent default so meh. <nckx>Apparently flashing *boot without a BIOS (no, *boot is _not_ a free BIOS replacement) makes it impossible to run Windows on the device. \o/ Can't wait. *nckx quips ‘now that's what I call Secure Boot’. <vagrantc>well, you can load seabios as a payload, not sure if that would break it for you :) <nckx>vagrantc: Linux doesn't need it & rescue images can be booted from GRUB, so I don't see the point (not sure about the Hurd though). <nckx>Plus it adds _seconds_ to boot speed X-( <vagrantc>so you can upgrade grub without reflashing? <nckx>I've been less up on GRUB development lately but it seems pretty ‘mature’, no? <nckx>Only thing I'd reflash it for is faster LUKS decryption, it's not like new hardware support is an issue. <nckx>TBH I wouldn't mind flashing once a year or so (this is all big talk from someone who hasn't flashed it at all, yet). <nckx>For a big feature such as that. <nckx>I assume it would involve re-encrypting my drives anyway so it's a Project in any case. <vagrantc>apparently, guix hasn't yet updated to a cryptsetup version that defaults to luks2, though <vagrantc>apparently luks1 -> luks2 can be done in-place <nckx>Flashing too, if nothing goes wrong 😛 <vagrantc>surprise to me too ... just was reading about all this earlier today <nckx>Shame there's no response. <vagrantc>time for everyone who wants grub and luks to switch to guix <nckx>(^ hence my general impression of GRUB development.) <nckx>Which is a shame, since none of the hip new fooloader fads provide a tenth of the flexibility and peace of mind that GRUB does. Bah. <emacsomancer>`guix package -u` seems to have resulted in local building of 6 different versions of rustc. I haven't explicitly asked for rust/rustc to be built and it's be churning for several hours. <nckx>How does one ’explicitly [ask] for rust/rustc to be built’? <nckx>Ah, never mind, I imagined a not where there was none. <emacsomancer>I never know when things will be built locally and when there'll be binary substitutes. I'm not sure why it's trying to build so many different rustc versions though <nckx>Because (roughly) the only way to build rust@n is with rust@n-1, which can only be built with rust@n-2, etc. You can't just compile the latest rust with gcc, for example. <nckx>The current path is already the minimal one ☹ <nckx>So I guess someone changed something that changed rust@n-1-1-1... and unfortunately the build farm is still sweating away at it. <tortoise>Hey, if I want to write a package that has a custom license, how do I create the license object? ***jonsger1 is now known as jonsger
<nckx>tortoise: Extremely custom as in unique to that package? (non-copyleft "file://path/to/licence") ; or http[s]:// <nckx>Yup. Just make sure it's really free ☺ <nckx>I'm not sure whether remote URLs or paths into the tarball are preferred. Maybe peek around the current packages using it. <nckx>Anybody else with creepy low-pitched audio in some videos in Icecat? *nckx is watching Satanic Announcer Baseball. <nckx>Agh… The guix-daemon on my build farm had crashed and it wasn't building anything. That'll be a fun catch-up. ☹ <nckx>No new Icecat for me either. <emacsomancer>I didn't have that video audio issue that last time I noticed. <nckx>It's only certain files. Youtube's fine, my not-super-legal MLB stream is not. But it's not only live streams or dodgy sites. <emacsomancer>I don't do a lot of video-watching inside of Icecat. I'm using pulling down things via youtube-dl. Doesn't work as well for streams, I suppose. Could you use VLC for the stream? <nckx>emacsomancer: Youtube-dl chokes on what I presume is encryption. So MPV is out as well. VLC hadn't occurred to me to be honest. <nckx>Weird how my first choice on borrowed $proprietary_os boxes is my last on GNU. <emacsomancer>I find VLC doesn't perform nearly as well as mpv (and I presume you have as well), but I've occasionally used it for streaming. <nckx>Yup. I'm used to extremely low-end laptops that can't afford VLC. I just got one that probably can. Strange new world. <emacsomancer>even on high end machines, I find that VLC isn't as good as mpv. Sometimes it stutters; and skipping in videos works much better in mpv. Plus, I think mpv's (admittedly minimal) interface is slick. <nckx>When I paste the m3u8 URL ☹ <nckx>Thanks for the idea. VLC hadn't occured to me and it was a good bet. <emacsomancer>I just like having a second browser (just in case). Ethics/privacy/etc. aside, Firefox/Icecat is technically superior to Chrom{e|ium} anyway and more pleasant to use. <nckx>Using Epiphany (my backup GUI browser ;-) gives: Your browser does not support the playback of this video. Please try using a different browser. <nckx>What are the odds there's a whitelist Chrome|Firefox|Edge somewhere. <emacsomancer>I mean, the problem is that for anything that might not work in Firefox/Icecat, it's unlikely to work in Epiphany. <nckx>If anyone feels like helping me with their Icecat results: tobias dot gr / mlb is where the illegal content at. <nckx>emacsomancer: Not exposed in the UI AFAICT. <nckx>But I'm very unfamiliar with it. Luckily I hardly ever have to fall back to a different browser. <emacsomancer>nckx: I get stuck at "just a moment" with a cloudflare ddos screen <nckx>It magically broke, it might magically fix itself. ***vagrantc_ is now known as vagrantc
<tortoise>how does guix package -i work on files that don't evaluate to the package in question? whenever I write my own and do guix package -f file.scm or guix build I need to end a package file with it's name so that the program evaluates to the package instead of just having a definition <tortoise>That might be unclear I'm not sure. I want to define my own packages that only have the package definitions, but guix package -f only seems to work if I end the file by evaluating the package I need. <dadinn>I found an issue with the binary install script... it keep failing with error "Could not obtain list of Guix releases." <dadinn>this seems to be due to wget on an https:// url <dadinn>Also, the init system is incorrect... it thinks it is sysv-init... this is because systemctl doesn't work well under chroot <civodul>dadinn: could you paste the output that you got from the install script? <lprndn->civodul: Thanks for yesterday. With your advice, I succeeded what I was trying. It's still hacky but it works! <lprndn->I'm asking way too late but did someone worked on a lightdm service for guix? <lprndn->cool thanks! so it should be on the way... <dadinn>civodul: the https problem is due to not having the cacerts installed on debian... it is a freshly debootstrapped instance <dadinn>civodul: I think it is fair to ask that I install those beforehand, I already install info, and dirmngr just for guix, so ca-cacert is an additional requirement <dadinn>civodul: but regarding the systemd init config... unfortunately I am not able to run the necessary systemctl commands in a chroot... any workaround for that? <dadinn>civodul: does the `guix archive` command talk to the guix-daemon? <dadinn>civodul: because that seems to be the only reason the script needs it to be enabled <dadinn>civodul: sorry not enabled, but started <dadinn>civodul: `systemctl enable` is just a symlink AFAIX, so it should be doable even in a chroot :/ <dadinn>civodul: nevermind I think I should just stick to executing the the install script in a running system <roptat>I think I won't be able to bootstrap scala after all, it's just too hard <roptat>I wonder if it's acceptable to send a patch series to add a binary scala and sbt built from that scala? <roptat>it's a bit hard to know what to do especially when the first thing that sbt does is to download itself and another version of scala... <civodul>sending what you have could at least bootstrap the discussion :-) <civodul>but yeah, it sounds like a really hard problem <roptat>kind of, but I found out that the scala documentation was not complete (and even contradictory), so my parser is not complete and probably wrong too <civodul>it would be nice not to waste it though <civodul>perhaps there are people in Scala circles that would be interested <roptat>I'll work on a plan to merge some of my own packages too, I have a lot of them that I've been just too lazy to send patches for <roptat>working on a separate collection of packages is nice because it's easy to develop, but it's more work to merge in guix afterwards too... <civodul>ideally it shouldn't be harder to get them into Guix for someone already familiar with the conventions and all <roptat>it's only harder because I don't pay too much attention to conventions in my own package set <roptat>we don't support auto-detection of other distros, instead you have to either not use grub from Guix System, or declare your other distros in your config file <roptat>if you're using windows, we don't support that (yet) unfortunately :/ *civodul is unsure whether this is "unfortunate" <abcdw>`guix system reconfigur config.scm` reports that system is 8 days old and I need to do `guix pull`, but I already did it right before reconfigure. What am I doing wrong?( <roptat>civodul, well it prevents me from booting into my haiku system too <roptat>abcdw, you probably forgot to update PATH or run `hash guix` <civodul>roptat: which means you can contribute more to Guix! :-) <civodul>seriously though, aren't you able to specify a menu-entry for Haiku? <roptat>no because I need to write "chainloader +1" somewhere and that's not possible with our grub abstraction <roptat>so I can boot from the grub console, but not from an entry <linarcx>roptat: Hi, i have windows7, arch linux, debian, nixos on the same hdd. A question: How to declare other distros in my config files? which config files? <civodul>we should discuss it with Danny and Mathieu <civodul>the problem roptat describes is for OSes other than GNU/Linux <abcdw>roptat, message appeared because I did sudo -s, hash showed path to user's profile. Thanks for help. <linarcx>civodul: Yeah, i see it before, but i don't have any idea about it. what is : `linux "/boot/old/vmlinux-2.6.32"`? How can i find read address to another distro kernel addresses? <linarcx>or this: ` (initrd "/boot/old/initrd")` <civodul>linarcx: you first need to find the "boot parameters" of that other distro: kernel file name, initrd, kernel arguments, etc. <linarcx>civodul: How can i find these informations? <civodul>it depends on the distro so i can't really help you much here <abcdw>I have %base-desktop-service and on each boot on first tty I see `New session cNNN of user gdm. Removed session cNNN.` for 5 hundred times. and nothing happens after that. I can login, start sway and so on. But those lines are very annoying. Why they appears? <civodul>abcdw: it's printed by elogind when the gdm user "logs in" (i.e., when GDM is started) <civodul>but you should see it only once, when gdm starts <civodul>and then probably also when you log out <civodul>roptat: i was thinking that it'd be nice to have fr_FR@tutoiement, as well as a variant that would explicitly use gender-neutral wording <civodul>are these things discussed by the French translation team? <civodul>are there guidelines or is it really up to each translator? <civodul>abcdw: but when you wrote 500 times, maybe that was actually less than that? :-) <abcdw>civodul, every time the number in the message increased by one and last number that I saw was c505. <civodul>abcdw: could you see if you can reproduce it in a VM (with 'guix system vm'), and then send the config and steps to reproduce to bug-guix@gnu.org? <abcdw>civodul, sure, will try to reproduce. <roptat>civodul, about gender-neutral/tutoiement, were you talking about guix, the manual or both? <roptat>how do you select a variant as a user? <civodul>roptat: i was thinking about translations to French, generally speaking <civodul>locale names can have variants after the "@" sign <roptat>that's the "modifier", right? like euro, etc? <roptat>and gettext can load different files depending on the modifier? <civodul>i'm not quite familiar with the details though <roptat>civodul, I just sent a mail to the traduc mailing list to ask about that <abcdw>civodul, on system vm message appeared only once, probably I will need to reinstall guix from the scratch. <roptat>abcdw, maybe it's because sources are read-only in your case <divansantana>perl-io-socket-ssl / git fails to build. Not sure if this has been asked yet, if so apologies. <roptat>I actually got a similar problem with an OCaml package <roptat>divansantana, can you show us the build log at /var/log/guix/drvs/99/9jzy3bx8aahj2h9fkc0xag6729pzq9-perl-io-socket-ssl-2.038.drv.bz2 ? <divansantana>I'm not sure why it's trying to build from source too. I thought, at what of the times I would try, there would exist a prebuilt package. <roptat>what does "guix describe" tell you? <roptat>also, have you authorized substitutes? <roptat>I just guix pull'd and got a substitute for perl-io-socket-ssl <roptat>also make sure .config/guix/current comes first in your path and run "hash guix" <abcdw>roptat, what does it mean? And what can I do with it? <roptat>abcdw, in my case, it meant that the install phase installed a file with read-only permission (it didn't change it) and then tried to overwrite it <roptat>"add write for user, to prevent a failure in the install phase" <roptat>the reason for the failure is that in the case of a git checkout, the files are directly in the store, so they are read-only. in the case of a tarball, the tarball is read-only but uncompressed files have write permission <roptat>in my case, that lead to the install phase failing with the same message as yours, and as you're also using a git checkout, it might be the same reason ***tilpner_ is now known as tilpner
<abcdw>roptat, I obtained an archive from github and now it's not failing in that place, thank you. <roptat>you shouldn't use a github archive, because they are not stable <abcdw>roptat, ok, I will keep it in mind, but now I just want to build a package somehow) <roptat>sorry, no idea, but one thing is sure: it doesn't have network access when building a package <abcdw>roptat, hm, that's mean that I should specify those rust packages as inputs somehow <pkill9>does guix have a /var/log/syslog equivalent? <pkill9>i've never actually used /var/log/syslog before though *abcdw Imported around 20 crates and still importing :/ <dustyweb>civodul: those were all people who *were* on freenode through the matrix bridge I think? <dustyweb>because matrix's bridge server (and a bunch of other infrastructure) got attacked <galaxie>Huh. I was wondering why there was so many messages. <civodul>dustyweb: yeah, that's my understanding, i was kidding :-) <civodul>but it certainly has an impact on Freenode <dustyweb>I'm still not sure why there's a *central* bridge server <dustyweb>why not have each separate federated matrix node do its own bridging <dustyweb>I find many things about matrix to be very strange <abcdw>how to do such thing: (define (string->symbol "symbol-name") "symbol-value") ? <bavier>nckx: yasnippet troubles? commit 6e36c6ed8 <bavier>civodul: there's a typo in guix.texi. Section "Limitations", 'dekstop' should be "desktop". <civodul>bavier: i don't see that typo in current master, is it still there? <bavier>civodul: ah, sorry, must have been fixed. I guess I was browsing the pinned version <civodul>i recently removed a few things from that section <nckx>bavier: I remember that from guix-commits@, it's just been fixed. <jonsger>dustyweb: I think almost all matrix users are on the matrix.org instance, so this is centralized either ways... <nckx>bavier: Eh. Whoops. Sorry. <nckx>I launched epiphany last night and it seems to have hijacked URLs in my terminal. <katco>hi, can irc users see this message? <katco>yay! sorry for any spam i've been causing while i figure out matrix's irc bridge <katco>dustyweb: that's how it's supposed to work. but many of us are using a central matrix server, not our own federated one. <katco>i was actually watching the fosdem talk about matrix, and the goal of the matrix team is to actually get rid of matrix.org as federation improves. i don't think they have a desire to admin their own instance -- at least not one this large <dustyweb>I've come to believe it's an antipattern for people who work on decentralized systems to host a "flagship" node <dustyweb>and it's healthier for the network for the core people to not be tied up in that every day <jfred>I'm a bit divided on that. On one hand it promotes centralization, but on the other hand when the community is small enough not having a flagship node makes it harder for new people to join <kmicu>nckx: happy mimeapps configuration. <nckx>kmicu: xdg-settings set default-web-browser icecat.desktop did the trick, but finding that out took time I would've rather wasted on something else. 😠 <nckx>PSA: You're not helping anyone by making your programme do things like that. <nckx>I expect that from Chromium & other commercial crapware. <kmicu>Better do not look inside xdg-open script then xD <jonsger>diaspora had a similar issue, where the biggest instance wasn't maintained well anymore (old software, slow, etc...) <katco>i agree with this. i definitely wouldn't look at matrix as favorably as i do had i not been able to try it out on a well maintained instance first <nckx>Today, on Overdrive: ERROR: In procedure dynamic-link: In procedure dynamic-link: file: "/gnu/store/dd8ld51ilm098kj2apx3wr53y124w8yi-libgcrypt-1.8.3/lib/libgcrypt", message: "file not found" <nckx>During ‘guix pull’. Which is broken. Which is bad. <nckx>I'd expect a failed ‘guix pull’ not to break ’guix pull’ (since it hasn't switched guixes yet) so what'm I missing. <lprndn->Hey. Does anyone know what are 'c1', 'c2' etc. sessions on guix? It <nckx>Problem seems to b the missing extension on libgcrypt. /gnu/store/dd8ld51ilm098kj2apx3wr53y124w8yi-libgcrypt-1.8.3/lib/libgcrypt.{la,so,so.20,so.20.2.3} do exist. <nckx>You can list them through loginctl, that's all I know. <lprndn->nckx: ok thanks! I need to go but i'll start with that. ***rekado_ is now known as rekado
<rekado>I’m still alive, but I’m not currently following the IRC channel or mailing list <rekado>still recovering from (planned) surgery; I’m only online some of the time when the pain killers kick in :) <rekado>had surgery on Saturday; I’m on sick leave for two weeks, but recovery can take three weeks or more. <nckx>rekado: Get well soon & take it easy :) <nckx>sneek: later tell civodul: Thanks for fixing http-parser. <apteryx>rekado: Give it the time it needs! But I hope you get well soon :-) ***maddo_ is now known as maddo
<nckx>abcdw: First (and total) guess is a circular reference in your code. Does the *build* hang, or the *computation* of the derivation (♪)? <nckx>Although I don't see an obvious problematic reference. <nckx>Does RAM usage actually have anything to do with it? <efraim>Could it be a problem with the build system? <abcdw>nckx, I created a swap file and set GC_INITIAL_HEAP_SIZE=40G, without setting this var to big enough value build fails. And htop says that all 16G of ram and 30G of swap used. <nckx>I'm not great at debugging circular refs myself; my usual method is ‘guess and poke until you hit something’. <nckx>since there's no info to go on. <abcdw>Can I plot a graph with some guix flags/commands? *kmicu wishes rekado quick and painless recovery. <dongcarl>(And yes I know it's stupid and against everything Guix stands for but for my particular use-case I need 0 rpaths) <nckx>dongcarl: You'd replace the pre-configure phase with a copied phase that omits that block. <nckx>Yes, it sucks, but that's the granularity you have. *nckx prefers small 'do-one-thing phases over 'pre-foo phases for that reason. <nckx>Or maybe you could do some lispy code-as-sexp manipulation. ‘Probably.’ Haven't tried. *nckx is being super helpful here yw. <nckx>dongcarl: If you have the time, I'm curious about what you're building. Bitcoin something Guix something? <dongcarl>nckx: yeah I'd love to explain if it's not too offtopic for this channel <dongcarl>Basically we have a deterministic build system already <nckx>If it involves Guix in any way I don't see how it can be. <dongcarl>But it basically just builds inside an Ubuntu VM with libfaketime and our own set of patches <dongcarl>I think that the Ubuntu distribution is too large of a trusted binary <dongcarl>I hesitated between NixOS and Guix, but Guix seems more focused on bootstrappability <dongcarl>So when I first started, I thought the easiest way would be to ship `guix pack` tarballs <dongcarl>So I'm on a journey to produce the same tarballs as our normal Gitian builds <dongcarl>Not deterministically the same, just functions the same. <harding>dongcarl: you should link your WIP PR. :-) <nckx>Hence the desire to de-guix the rpath. <dongcarl>Right now I've solved that problem, and the only problem is the interpreter path <dongcarl>Once I solve the interpreter path, we're golden (at least for x86_64-unknown-linux-gnu) <nckx>While I'd prefer to see it done ‘properly’ (not that I know how), maybe running patchelf on the result would be an acceptable hack for now? <nckx>It's been ages since I've used it but it could modify interp paths last I did. <dongcarl>nckx: Yup! I've already got it working with `patchelf`, so the journey is to do it 'properly' <dongcarl>nckx: would you know how to append to inherited configure-flags? <nckx>Nice to hear peeps doing that in this age of DockeraaBuildSystem. <nckx>dongcarl: Yes, but off the top of my head. <dongcarl>`DockeraaBuildSystem` nice one I'm gunna use that <nckx>dongcarl: make-glibc-locales in (gnu packages base) seems to be a pretty minimal example of that. If it fits into your current way of inheriting things. <dongcarl>nckx: Ah! `substitute-keyword-arguments`! <tune>does anyone here use xmpp on guix? can't figure out how to enable encryption in dino, can't get logged in at all in gajim <tune>oh dear I think it was just lagging massively. I'm now logged in several times over <nckx>eolie on Guix defaults to opening google.com. <nckx>Is this something that we want? <dongcarl>Does anyone know how to configure spacemacs to reformat my guile code the way the Guix repo does? <dongcarl>I think spacemacs' Guile/Scheme layer uses geiser in the back, but not sure if that's used to reformat things <civodul>dongcarl: Geiser is used to communicate with the Scheme implementation <sneek>Welcome back civodul, you have 1 message. <sneek>civodul, nckx says: Thanks for fixing http-parser. <civodul>dongcarl: but i think Emacs-Guix provides scheme-mode customizations <civodul>also, the Git repo has a .dir-locals.el file <dongcarl>civodul: Thank you very much, I need to check that out <dongcarl>does .dir-locals.el have special meaning? <apteryx>it's a Emacs builtin way to define variables 'per-directory' <dongcarl>It seems that I need to build guix first to use ./etc/indent-code.el.in though? ***paroneayea is now known as dustyweb
<apteryx>correct, that .in is a template used by the build system <civodul>dongcarl: but since you're using Spacemacs, you probably don't need indent-code.el at all? <civodul>i suppose Spacemacs honors .dir-locals.el, doesn't it?