IRC channel logs

<lfam>Howdy <joshuaBPMan_>

<rekado_>I just ran guix system init on an Ubuntu machine, but forgot to update the disk label, so now the thing fails to boot.

<rekado_>is there an easy way to override the disk info at boot?

<roptat>rekado_: maybe press 'e' on grub to temporarily edit the menu entry?

<roptat>or you could boot a liveCD to run e2label

<rekado_>I’m connected via serial console and I’m not fast enough to see GRUB.

<rekado_>(because I need to disconnect the serial interface, then issue a reset in the management interface, then restart the serial console.)

<rekado_>booting a liveCD remotely won’t work

<rekado_>I’m in the rescue REPL, though, so maybe there’s something I can do from there.

<roptat>I'd don't know the REPL enough to help you

<rekado_>that’s okay. Thanks for the hint with e2label.

<rekado_>ah, was fast enough this time to get into the GRUB console.

<temp_trisquel>Hi.

<temp_trisquel>I have forgot LUKS password but my HDD is mounted.

<temp_trisquel>I'm using installation USB.

<temp_trisquel>What can I do?

<temp_trisquel>Ah, I remembered it.

<garlox[m]>Hi guix. Developing a guix package for the first time using the "./pre-inst-env" approach, I notice it first bootstraps the whole compiler toolchain through compiling. That means it takes many hours until I have a working development environment. Is there are way around this? E.g. using the packages I already have on my system guix?

<garlox[m]>Ah, solved it myself. The hydra mirror must be manually authorized for the guix development checkout:

<garlox[m]>> sudo -E ./pre-inst-env guix archive --authorize < hydra.gnu.org.pub

<garlox[m]>It is documented here: https://www.gnu.org/software/guix/manual/html_node/Substitutes.html

<temp_trisquel>Hi.

<temp_trisquel>GuixSD says my mouse is removed and then, almost immediately it's added again.

<temp_trisquel>Every [some short time].

<temp_trisquel>Sometimes it makes executing commands harder (I see only part of command).

<temp_trisquel>rekado: I have failed to install EFI GuixSD, so I'm installed BIOS.

<temp_trisquel>*rekado_

<temp_trisquel>rekado_: iomem=relaxed causes reboot with unbootable OS.

<temp_trisquel>Will guix system reconfigure reinstall OS? I don't need OS to be reinstalled.

<temp_trisquel>I need to update packages.

<Gamayun>temp_trisquel: It will install updated packages, and change your OS configuration to point to those packages...

<temp_trisquel>Oh no. Because the "mouse spam", I can't see was guix system reconfigure successful.

<temp_trisquel>It spammed the whole screen. What can I do?

<temp_trisquel>How can I start Tor at boot on GuixSD?

<temp_trisquel>I also need Tor DNS.

<Apteryx>Hello Guix!

<ng0>sneek: later tell temp_triquel: It helps to read the documentation, which covers questions on system services, reconfigure, usage, etc

<sneek>Okay.

<temp_trisquel>sneek: later tell sneek: test

<sneek>Weirdo.

<temp_trisquel>triquel?

<temp_trisquel>I'm using Trisquel on this computer.

<temp_trisquel>sneek: later tell test

<temp_trisquel>sneek: later tell temp_trisquel: test

<sneek>Okay.

<temp_trisquel>1

<sneek>Welcome back temp_trisquel, you have 1 message.

<sneek>temp_trisquel, temp_trisquel says: test

<temp_trisquel>1

***temp_trisquel is now known as temp_trisquel1

<temp_trisquel1>later tell user_trisquel: test

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Okay.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Got it.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Okay.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Okay.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Will do.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Got it.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Okay.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Will do.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Got it.

<Xe>why are you doing this here

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Okay.

<temp_trisquel1>sneek: later tell user_trisquel: test

<sneek>Will do.

***temp_trisquel1 is now known as user_trisquel

<user_trisquel>test

<sneek>Welcome back user_trisquel, you have 13 messages.

<sneek>user_trisquel, temp_trisquel says: test

<sneek>user_trisquel, temp_trisquel1 says: test

<user_trisquel>I want to ban sneek.

<user_trisquel>Just for fun.

<user_trisquel>l

<sneek>user_trisquel, you have 1 message.

<sneek>user_trisquel, user_trisquel says: test

<user_trisquel>1

<sneek>user_trisquel, you have 25 messages.

<sneek>user_trisquel, user_trisquel says: test

***user_trisquel is now known as temp_trisquel

***temp_trisquel is now known as user_trisquel

<user_trisquel>1

<user_trisquel>sneek: help

***user_trisquel is now known as temp_trisquel

<temp_trisquel>sneek: who is sneek

***Gravious is now known as scarfeater

<adfeno>Hi #guix! :) I just did `guix pull && guix package -u` and noticed that the build for Git 2.14.0 is failling here.

<adfeno>...phase 'check' fails after test "ok 12 - test disallow multiple asterisks in one word". The fail is as follows: [Makefile:36: test] Error 2. ... [Makefile:2415: test] Error 2.

<quigonjinn>is there a way, instead of passing the flag --no-bootloader to 'guix system ...', to integrate not installing grub in the configuration file?

<adfeno>About my git error from before: Fixed it by installing guile-git in root's profile

<adfeno>:)

<temp_trisquel>Hi.

<temp_trisquel>Is DNS on port 53 allowed in tor-service?

<rekado_>I have a problem running “guix pull” on an installation from 0.13.0.

<rekado_>it tries to build guile-git, but the repository has moved:

<rekado_>“remote: Project 'amirouche/guile-git' was moved to 'guile-git/guile-git'”

<efraim>I thought there was supposed to be a 301 redirect

<efraim>mark reverted my guix-daemon patch that allowed the daemon to work on aarch64, time to figure out a fix

<efraim>and write up a talk proposal for august penguin

<efraim>this feels absurd to say as it now feels obvious, it seems indentation matters in our daemon code

<adfeno>efraim: Why it feels absurd and now feels obvious?

<efraim>i saw the other code intented but never bothered indenting my own if statements

<rekado_>hmm, so with this guile-git problem I cannot upgrade Guix on this machine :-/

<rekado_>efraim: indentation matters??

<efraim>either that or the stack on aarch64 doesn't need to be multiples of 16 on the board I'm testing

<efraim>guix build: error: build failed: cloning builder process: Invalid argument

<efraim>looks like indentation matters in #ifdef statements

<adfeno>efraim: #ifdef are you talking about Tab prefixed lines?

<adfeno>in Makefile files?

<efraim>i was working in nix/libstore/build.cc

<adfeno>At least in the case of make files, I know that Tab-prefixed lines denote *shell* commands.

<efraim>that I didn't know

<efraim>my original patch from february said "if statements are hard, fix this"

<lfam>efraim, rekado: I noticed that issue with the guile-git source previously, but the guile-git developer told me that he could not set up the redirection for some reason. I didn't push for a reason but maybe we should

<adfeno>So, one cannot use "ifdef [somestuff] [begin of line][some other stuff][begin of line][tab]ifdef [more stuff] ... and so on... because the second ifdef is an unknown command.

<efraim> https://paste.pound-python.org/raw/rH251r2nuMAGEYmQ405d/

<efraim>i should really set up a .wgetpaste to paste to somewhere else

<rekado_>lfam: I think it’s weird that we build from source when running “guix pull”. It would be nicer not to have to build guile-git from source as a user of “guix pull”.

<lfam>Is it not eligible for substitution?

<efraim>i was thinking we could compile to .cache/guix/... and then copy to the store, and offer a `guix pull --full' or something to deal with ABI changes, rather than compile fully each time

<lfam>I'd rather not make the user think about ABI changes. We already get so many repeat questions about other things...

<adfeno>lfam: +1

<lfam>I'm not saying we should rule out any improvements, but I think people will get too confused by having to choose

<efraim>i haven't run `guix pull` on any box with 2 GB of ram in about 2 months since it takes hours

<rekado_>lfam: it never even looks for substitutes. It fetches the Guix repository archive and then goes straight to fetch guile-git.

<lfam>Room for improvement :)

<efraim>substitutes enabled?

<efraim>i worked on make-u-boot-package, soon it shouldn't cross-compile if its the same architecture

<lfam>If I prepare patches to upgrade the whole texlive system to 20170520, can you test that biber-2.7 works with it, rekado_?

<efraim>my libgo patch doesn't work on aarch64, but I think i traced the issue back to CGO, so I'm going to need to make it into a real patch and send it off to the list for testing on an x86_64 machine

<lfam>efraim: I'll test it if you explain how

<lfam>What branch would I test it on?

<efraim>it can be on main

<lfam>Okay, I'll build it now (my main tree is on core-updates)

<efraim>its needed for gccgo@7, which should then be able to be used in lieu of go@1.4 to compile go

<efraim> https://paste.pound-python.org/raw/lKCdMVrpHHBNeBkRmxJx/

<efraim>i think i just added it to the gcc@7 patches

<lfam>So, add it to the patches for gcc-7 and then what steps should I take?

<efraim>go produces a binary at /gnu/store/...-go/bin/go?

<efraim>copy gccgo-4.9 and rename to gccgo-7

<efraim>and then in golang change ("go" ,go-1.4) to ("go" ,gccgo-7)

<efraim>and I think that should be it

<efraim>I get the fork/exec error on aarch64

<efraim>you know what, I bet I forgot something with switching gcclib

<efraim>if so, around line 222 "gcc:lib" should switch to "go"

<efraim>oooh, I really hope the bootloader talk at GHC is recorded

<lfam>GHC or GHM?

<efraim>GHM

<rekado_>lfam: yes, if you send the patches I can test biber.

<rekado_>ACTION –> zzZZ

<lfam>sneek: later tell rekado, efraim: Looks like redirecting between gitlab projects is indeed not possible: https://gitlab.com/gitlab-org/gitlab-ce/issues/35385

<sneek>Okay.

<efraim>well thats unfortunate

<lfam>sneek: later tell rekado, efraim: Oops, wrong link. https://gitlab.com/gitlab-org/gitlab-ce/issues/17361

<sneek>Okay.

<lfam>Yeah, so we will have to solve this ourselves :/

<paroneayea>hm

<paroneayea>is webrtc not working in icecat right now?

<lfam>Somebody else asked there in here recently, so it's probably not just you

<paroneayea>I think what happened is the video player was moved from gstreamer to ffmpeg

<paroneayea>presumably for security reasons

<paroneayea>looks like that happened in commit 002c3e6dd6709ba47823214111aa56aa6e0670c6

<paroneayea>so that's my tguess.

<temp_trisquel>paroneayea: WebRTC leaks your IP.

<paroneayea>temp_trisquel: yeah, well, I want to use it in this case anyway :P

<temp_trisquel>paroneayea: paroneyaea = paranoia?

<paroneayea>temp_trisquel: that's what they'd like you to think ;)

<lfam>WebRTC leaking one's IP is not really an issue for most use cases, right? Presumable you are comfortable sharing your IP if you are using a web browser, unless you are using Tor.

<temp_trisquel>Internet must be destroyed.

<paroneayea>lfam: yeah that's how I feel. Granted, you're also connecting to peers via webrtc

<paroneayea>but

<paroneayea>is trusting peers worse than trusting a server?

<paroneayea>anyway, I'm guessing that isn't the reason we don't have webrtc at the moment, I'm guessing a change to ffmpeg is not related?

<lfam>For me, in general, if I am connecting to a remote computer, I don't care if it knows my IP

<adfeno>lfam: +1

<lfam>Anyways, I'm not sure what's up with webrtc in icecat. Perhaps a bug report is warranted :)

<lfam>efraim: I think that building gccgo takes longer than building vanilla gcc!

<temp_trisquel>Can somebody help to write standard for malware that will destroy the Internet (attack all computers possible).

<efraim>I image it should be about the same as building gfortran

<temp_trisquel>Also, never run software on your computer, even through Tor.

<lfam>temp_trisquel: I recommend you read the book 'The C Programming Language' (aka K&R C) and start writing network-facing software

<lfam>I think it will be indistinguishable from an attack ;)

<temp_trisquel>*this software

<temp_trisquel>I think, both standard and software can be published "for educational purposes".

<temp_trisquel>This malware must be run on others, compromised computers.

<temp_trisquel>The main issue is bandwidth.

<lfam>temp_trisquel: All jokes aside, this channel is for discussion of GNU Guix, GuixSD, and related programs. It's not for collaborating on malware or planning attacks. Please stay on-topic

<temp_trisquel>If computer will send a message for a million PCs, many users will found their speed is low.

<paroneayea>lfam: +1

<temp_trisquel>lfam: At least, we can include this malware into Guix.

<paroneayea>temp_trisquel: I think you heard lfam's message, and so now's a good time to stop.

<temp_trisquel>Will this malware be included into Guix if somebody will ask?

<temp_trisquel>It must be available for many architectures and operating systems.

<lfam>temp_trisquel: The Guix projects tends to make decisions by consensus but my personal opinion is no, we won't include things described as malware

<adfeno>Also, including malware would be against the GNU FSDG.

<lfam>It's not in the spirit of the project, and the legal liability could be severe

<adfeno>that would make GuixSD be non-free distro.

<paroneayea>okay, there's not even any need to carry on this conversation any more

<adfeno>paroneayea: +1

<paroneayea>there are plenty of other forums on the internet that are relevant to malware-making conversations, and #guix is decidedly not one of them.

<temp_trisquel>Non-free?

<paroneayea>we're not even going to debate the freeness of it.

<paroneayea>find another venue.

<temp_trisquel>This malware must be public domain with sources available.

<temp_trisquel>To be compiled on unsupported OS.

<ym>Hello. Will I get last guile and emacs (-wip branch) development environment after installing GuixSD?

<adfeno>temp_trisquel: I call your attention to discuss this in one of: #fsf #gnu #libreplanet

<paroneayea>ym: hi! are you talking about getting the latest branches of each, or using the guile-emacs package?

<efraim>we don't currently package the dev branches/releases of guile or emacs

<paroneayea>we have before of guile, but not right now :)

<ym>Ok, got it, thanks.

<lfam>efraim: Building go-1.8.3 fails with:

<lfam>go build _/tmp/guix-build-go-1.8.3.drv-0/go/src/cmd/dist: : fork/exec : no such file or directory

<lfam>efraim: This is the diff: http://paste.lisp.org/+7K43

<lfam>Plus the patch you shared

<lfam>efraim: This is a little old but they seem to have worked around the same issue: https://github.com/golang/go/issues/10092

IRC channel logs

2017-08-05.log