IRC channel logs


back to list of logs

<niebie>boot from usb no good?
<nyberg>only have ubuntu livecd so trying to fix from here
<Copenhagen_Bram>How do I search for a package containing two words, but not only one word or the other?
<Copenhagen_Bram>video AND editor, for example
<Copenhagen_Bram>and not video OR editor
<nyberg>regex no?
<niebie>guix package -s '(video.*editor)|(editor.*video)'
<Copenhagen_Bram>well, yes, but i'm not an expert at regex. Why does `video&editor` seem to give me results that contain video OR editor?
<Copenhagen_Bram>*sees suggestion
<Copenhagen_Bram>that's slightly complex looking :/
<niebie>() = group
<niebie>| = or
<niebie>. = any character
<niebie>* = any number
<Copenhagen_Bram>what does & mean?
<Copenhagen_Bram>and thank you
<niebie>afaik not much, i don't use it
<nyberg>playing with sed with the manual up helps
<niebie>based on the results of my query btw, it seems to have worked out fine
<niebie>avidemux and mlt
<Copenhagen_Bram>mlt? i got avidemux and out
<nyberg>kernel panic, there we go. just need to play until I get it to boot fully
<Copenhagen_Bram>how to get what to boot fully? guixSD? what's wrong with it?
<niebie>'name: mlt
<niebie>version: 6.4.1
<niebie>outputs: out
<niebie>systems: x86_64-linux i686-linux armhf-linux aarch64-linux mips64el-linux
<niebie>dependencies: alsa-lib@1.1.6 ffmpeg@3.4.2 fftw@3.3.5 jack@0.125.0 ladspa@1.13
<niebie>+ libsamplerate@0.1.9 libxml2@2.9.7 pkg-config@0.29.2 pulseaudio@11.1 sdl@1.2.15
<niebie>+ sox@14.4.2
<niebie>location: gnu/packages/video.scm:1815:2
<niebie>license: GPL 3
<niebie>synopsis: Author, manage, and run multitrack audio/video compositions
<niebie>description: MLT is a multimedia framework, designed and developed for
<niebie>+ television broadcasting. It provides a toolkit for broadcasters, video
<niebie>+ editors, media players, transcoders, web streamers and many more types of
<niebie>+ applications. The functionality of the system is provided via an assortment
<niebie>+ of ready to use tools, XML authoring components, and an extensible plug-in
<niebie>+ based API.
<niebie>relevance: 6
<niebie>ah shit
<Copenhagen_Bram>maybe i did see mlt
<Copenhagen_Bram>and mistook the output type as out
<nyberg>grub isn't detected at all. no net capabilities as iso doesn't work with phone
<Copenhagen_Bram>yeh it's called mlt
<niebie>if you want to see packages like this, you could use 'guix edit mlt', to bring up the package definitions
<Copenhagen_Bram>nyberg: are you trying to put a distro on a phone?
<niebie>then you could page through them
<niebie>Copenhagen_Bram:gl, gotta go
<Copenhagen_Bram>also emacs not found
<niebie>echo $EDITOR
<niebie>that is likely emacs
<niebie>set EDITOR='Copenhagen_Bram editor pref'
<Copenhagen_Bram>EDITOR=zile worked
<Copenhagen_Bram>ACTION burns all the emacs users
<niebie>evil is a better vim
<niebie>but no flame wars seems like a good plan
<Copenhagen_Bram>actually i've been attracted to emacs lately so i'll try out evil
<nyberg>Copenhagen_Bram: no,boot guixsd on my x220
<pkill9>why is the version different when running `/run/current-system/profile/bin/guix --version ; ~/.config/guix/current/bin/guix --version`?
<snape>oops sorry for the spam on guix-patches
<snape>I should go to bed :p
<snape>too tired
***Copenhagen_Bram is now known as Stingy_Japanese
<OriansJ``>I find it interesting so few people in this community care about how to improve the situation of discrimination against those who are not neurotypical or is it fear that one might be labeled for having any perspective contrary.
<Stingy_Japanese>who's discriminating against non neurotypical people?
<Stingy_Japanese>btw i have aspergers
<OriansJ``>Stingy_Japanese: It was the topic of discussion on #guix-COC-discussion about how the Code of Conduct could be discriminate against those who have low EQ or are non-neurotypical and what could be done to improve the situation as was anounced on guix to ensure broader community engagement.
<lyr3>hey, I installed icecat under guixsd but it just not open
<lyr3>I would like to remove it and reinstall but as you know it just resymlink the previous installed icecat
<lyr3>there is any way to (purge) icecat then reinstall or even force install
<apteryx`>Hello! debbuging `guix offload', continued :)
<apteryx`>anyone is welcome to jump on the bandwagon ;)
<vagrantc>the only way i could get offloading to work was specifying --max-jobs=0
<vagrantc>seems a bit counter-intuitive to me, though
<apteryx`>I'm stuck before reaching this interesting point.
<apteryx`>I get: guix offload: error: build failed: program `guix-authenticate' failed with exit code 1
<apteryx`>But I'm getting closer to something useful with gdb debugging of guix-daemon at the remote; I just got: error: executing `guix-authenticate': No such file or directory
<apteryx`>printed by stopping on every fork and continuing
<apteryx`>(gdb) catch fork
<apteryx`>so it seems it just doesn't find that program in the PATH.
<vagrantc>you've set up pubkeys for both ends? e.g. machineA: guix archive --authorize < ; machineB: guix archive --authorize <
<apteryx`>It seems the guix-daemon adds "settings.nixLibexecDir" to the PATH to try to find that... but that doesn't help because on my system, nixLibexecDir := /usr/local/libexec, and guix-authenticate is not found there.
<apteryx`>vagrantc: yes, I've doubled checked the pubkeys setup is correct.
<apteryx`>the start of the 'guix offload test' output looks promising:
<apteryx`>vagrantc: Are you using a GuixSD machine as the offload machine?
<vagrantc>i've used both guixsd and debian machines
<apteryx`>vagrantc: any special magic PATH setting, or guix profile sourcing in your ~/.bashrc?
<apteryx`>(on the remote offload machine?)
<vagrantc>though there have been a lot of guix changes in the last few weeks; it's been a while since i tested
<apteryx`>my guix is 66 days outdated. That's the number of days I've been stuck on this puzzle ;)
<vagrantc>i don't recall doing anything unusual
<vagrantc>that's about the extent of what i've got :)
<apteryx`>OK, I'll try hacking my PATH and see if it progresses ^^
<apteryx`>thanks for sharing some hope
<lyr3>hey, I cant run icecat. there is any way to force reinstall it instead of re-symlinking the installed one?
<lyr3>I guess that guix reconfigure might help me as I installed icecat as user
<brendyn>lyr3: well that's what it means to "reinstall" in guix. It makes no difference. You could try deleting the config directory in your home
<lyr3>brendyn: I installed it under guixsd
<brendyn>what happens when you run icecat from a terminal? Maybe there are useful error messages there when it fails to run
<lyr3>it yields nothing
<lyr3>so weird
<brendyn>No terminal output at all?
<brendyn>Does it hang or return to the shell?
<lyr3>which icecat confirms it is installed
<brendyn>Maybe you can try running `strace icecat' to gain more information
<brendyn>You'll have to install strace
<lyr3>execve("/home/lyr3/.guix-profile/bin/icecat", ["icecat"], 0x7ffed071a510 /* 63 vars */) = -1 ENOEXEC (Exec format error)
<lyr3>fstat(2, {st_mode=S_IFREG|0644, st_size=130, ...}) = 0
<lyr3>write(2, "strace: exec: Exec format error\\n", 32strace: exec: Exec format error
<lyr3>) = 32
<lyr3>getpid() = 2908
<lyr3>exit_group(1) = ?
<lyr3>+++ exited with 1 +++
<brendyn>Wow ok I'm not sure what that means
<apteryx`>lyr3: Try 'pkill icecat; mv ~/.mozilla ~/.mozilla_bak; icecat'
<lyr3>no such directory ~/.mozilla
<apteryx`>OK :/
<apteryx`>so it's not personal configs which are messing it.
<brendyn>Does icecat use the mozilla directory?
<apteryx`>it does
<apteryx`>it's based on Firefox ESR, which does
<lyr3>cant say
<apteryx`>I have ~/.mozilla/{icecat,extensions}
<lyr3>there is any way to purge icecat?
<lyr3>or there are any other browser
<apteryx`>lyr3: is your guix up to date?
<lyr3>just installed and (guix pull)
<brendyn>I wonder if it is corrupt in the guix store?
<lyr3>it said corrupt something
<brendyn>purging it would require uninstalling and then running `guix gc'
<lyr3>there reinstall it
<brendyn>guix gc should remove everything from the store that isnt installed in a profile
<apteryx`>guix package -r icecat && guix gc $(guix build icecat) should do it I think
<lyr3>apteryx1: it is installing it
<lyr3>but I ve done a clean guix gc hahaha
<apteryx`>OK! good luck :)
<lyr3>so we can specify which package (guix gc) can free space...interesting
<apteryx`>hope nothing went wrong there... ;)
<lyr3>no luck
<lyr3>it keeps symlinking to that earlier install icecat
<brendyn>oh probably because you also have backup profiles
<brendyn>maybe delete all the old generations
<lyr3>I am just deleting icecat generation haha
<brendyn>Everytime you change something in your profile the current profile becomes a backup generation
<brendyn>guix package --list-generations
<apteryx`>lyr3, when your reinstall, if guix hasn't changed, icecat should be strictly the same
<apteryx`>so you'd see the same store location being symlinked to again.
<apteryx`>arg! I've added 'guix-authenticate' to my PATH, but it still won't find it!
<apteryx`>(guix offload test)
<brendyn>Why does guix need to download wayland, cups-filters and a bunch of stuff just so i can run `guix package -r monero' ?
<apteryx`>guix graph might give you answers; it's probably indirect dependencies
<apteryx`>ah, '-r'. this I'm not sure.
<apteryx`>Could be related to substitutes, but I'm not very knowledgeable about that part yet.
<apteryx`>substitutes or grafts
<apteryx`>I'd like to know as well!
<brendyn>Guix does so many weird things that don't make sense to me.
<mange>Is there something in particular that you're mystified by right now?
<brendyn>mange: About why guix is downloading stuff in order to uninstall a package
<mange>Oh, right. That makes sense, in context.
***Gamayun_ is now known as Gamayun
<pheonix83>Hey im about to download the iso and install again. last time i had trouble with the partitioning part of the install. can someone point me to a more detailed wiki with partitioning like partition sizes.
<wigust>pheonix83: What's the trouble did you have?
<pheonix83>wigust: not sure what i did it was couple weeks ago i tried it. but when i restart after installation i see grub menu loading then it reboot again and just looped liked that. grub didnt load.
***dc is now known as danio
<pheonix83>i remember i made a boot_grub partition, /boot partition /root partition and /home partition. dont know if i messed up the sizes for grub and boot or what. when i installed gentoo i just followed the wiki and all was good. dont really know fully what i am doing when it comes to partitioning tbh
<pheonix83>so without an actual guid saying make this partition this size and add this flag like so. im lost.
<wigust>pheonix83: The Guix documentation is talking about separate /boot partition
<wigust>pheonix83: I would recommend you don't separate /boot in a partition if you don't need it
<pheonix83>yeh thats the guide i followed, i dont use EFI so got a little confusing as it begins to talk about bios grub but all the exampls are for EFI. So do i add the bios_grub flag to the /boot partition?
<pheonix83>so i would only have /boot partition /root partition and /home partition
<wigust>pheonix83: i don't think you need any flags in non-efi system, at least from my experience with a non-efi machine
<wigust>pheonix83: so just skip this note and make sure you have mounted your boot partition in /mnt/boot as the documentation says
<pheonix83>oh that makes things much easier.
<pheonix83>ill see how i go this time :D thanks
<civodul>Hello Guix!
<jonsger>civodul: did version 1.0.3 of guile-bytestructures provides an important fix for guix or does it just make the package definition cleaner?
<civodul>jonsger: it provides an important fix, see
<jonsger>oke, thx
<roptat>hi guix!
<roptat>so I learnt something: transferring the guix package and creating a symlink from .config/guix/current doesn't create a gc root
<roptat>so after running guix gc I was left with the system guix, which is older
<civodul>roptat: ah sure, you need to register the GC root explicitly, otherwise guix-daemon cannot guess :-)
<civodul>registering the GC essentially boils down to creating another symlink in /var/guix/gcroots
<roptat>from /var/guix/gcroots to .config, or /gnu/store?
<roptat>civodul: since the new version creates a profile, it should be possible to download most of the packages of that profile from berlin or hydra by choosing the correct commit, right?
<roptat>or do they only build guix-manual?
<civodul>the gc root should point to ~/.config/guix/current
<civodul>i think
<civodul>the way 'guix pull' does it is by creating an "indirect GC root", which is slightly different: it creates a symlink in /var/guix/gcroots/auto that guix-daemon can automatically remove if its target no longer exists
<civodul>re substitutes, yes, berlin provides substitutes for all the derivations that 'guix pull' computes (that's unrelated to the fact that we're using a profile)
<roptat>civodul: can I use guix weather to see if the substitutes are available?
<civodul>roptat: not really because 'guix weather' expects a list of packages, but we should change it to be more flexible
<roptat>"Génération 3juin 14 2018 10:20:32(actuel)" this is terrible :/
<civodul>looks like translators removed the tabs, no?
<roptat>no, that my irc client I think
<roptat>should show "Génération 3 juin 14 2018 10:20:32 (actuel)"
<civodul>oh, so you were referring to the date translation?
<civodul>it's very bad indeed
<Rukako>hi guix
<civodul>hello Rukako!
<civodul>roptat: i plaid guilty: i think i did the bad SRFI-19 i18n a long time ago in Guile
<civodul>but SRFI-19 'date->string' doesn't lend itself well to i18n
<civodul>because in the format string you would specify things like "~Y-~M", meaning "year-month"
<civodul>and 'display-generation' in (guix ui) uses: "~b ~d ~Y ~T"
<civodul>perhaps we should put that in G_ as well, with a comment explaining that translators should use a format that works well in their language?
<civodul>roptat: WDYT?
<roptat>sounds good
<civodul>can you take care of it?
<roptat>civodul: sure
<civodul>awesome, thanks
<civodul>in the "TRANSLATORS" comment you can link to
<civodul>or maybe we should use the default, which is ~c
<civodul>ACTION tries
<civodul>hmm not any better:
<civodul>LC_TIME=fr_FR.utf8 guile -c '(use-modules (srfi srfi-19)) (pk (date->string (time-utc->date (current-time time-utc))))'
<civodul>;;; ("jeu. juin 14 10:37:25+0200 2018")
<civodul>yeah that's because of this hardcoded format in srfi-19: (define locale-date-time-format "~a ~b ~d ~H:~M:~S~z ~Y")
<civodul>so using G_ as we discussed above is the best solution at this point
<roptat>civodul: how does G_ work?
<civodul>just write (G_ "some literal string")
<civodul>and gettextize will automatically extract the string for translation in the .pot file
<civodul>and then at run-time G_ is essentially an alias for the 'gettext' procedure
<roptat>so how do you add a comment for translators ?
<civodul>Add a comment exactly on the line(s) before the G_ line that you want to comment, and start it with "TRANSLATORS:"
<civodul>there are examples in the code
<civodul>i thought translators knew this ;-)
<roptat>translators open .po files :p
<civodul>true :-)
<civodul>ACTION just pushed the store database patch series
<roptat>civodul: I don't know if you have seen that:!map
<roptat>guess where hydra is :)
<civodul>heh, funny
<civodul>that's ping latency though, which is the least of our problems :-)
<roptat>there are other kinds of measurements we can make
<roptat>but they are "low-level" (ping, traceroute, dns, ssl, http or ntp)
<roptat>how much space do I need on a server if I want to create a mirror for all architectures?
<civodul>as much as you want, in the sense that you can give nginx an upper limit
<civodul>also if you make it a lazy mirror, like, it'll be populated on demand
<civodul>so if nobody uses it it needs no storage :-)
<roptat>how much does it take on mirror.hydra?
<civodul>well 1.4T is the sum of the local store plus the nginx cache
<roptat>ok, and I guess there are also source tarballs
<civodul>yes, you can't discriminate between fixed-output derivations and other derivations anyway
<civodul>so the nginx cache is 99G
<roptat>why is the local store so big, then?
<civodul> is also a build machine of
<roptat>oh I see
<civodul>and it has a big hard disk, so...
<civodul>breakdown of the nginx cache:
<civodul>6.4G /gnu/nginx/cache/cas
<civodul>97M /gnu/nginx/cache/logs
<civodul>93G /gnu/nginx/cache/nar
<civodul>288M /gnu/nginx/cache/narinfo
<civodul>16K /gnu/nginx/cache/static
<civodul>"cas" is "content-addressed storage", i.e., the /file URLs that 'guix publish' serves
<jonsger>civodul: ssd or hdd
<civodul>jonsger: hdd
<Copenhagen_Bram>hello guix
<Copenhagen_Bram>i can't connect to the internet from my installation
<Copenhagen_Bram>wpa_supplicant fails to connect with my router even though it works on the live usb drive
<Copenhagen_Bram>i'm making an archive of connman to put on there lol
<Copenhagen_Bram>and how come internet worked on the installation when it had gnome?
<Copenhagen_Bram>hey pkill9
<Copenhagen_Bram>Is there anyone who can help me? Should I obtain a log from wpa_supplicant?
<rekado_>sorry, I don’t know how to help here.
<roptat>Copenhagen_Bram: do you use %desktop-services in your config?
<Copenhagen_Bram>roptat: yes i'm pretty sure i do
<Copenhagen_Bram> (services (cons* (xfce-desktop-service)
<Copenhagen_Bram> %desktop-services))
<roptat>Copenhagen_Bram: so I think you're supposed to use network-manager now
<roptat>nmtui should work
<roptat>use it to configure the connection, then nm will automatically connect when you start your computer
<Copenhagen_Bram>alright i'll try that
<jlicht>hey guix
<rekado_>jlicht: did you once ask for the mu4e patch to debbugs-gnu.el?
<jlicht>I did :-)
<jlicht>I've been using it for the last week, and it works quite well (with the caveat that I have to clear out my debbugs-<...> mailboxes from time to time)
<rekado_>sorry for not replying earlier. I find direct messages on IRC confusing as my office workstation is always connected and receives them when I’m out or when my laptop is offline.
<rekado_>so I don’t see them when I reconnect with my laptop.
***Guest12932 is now known as sturm
<jlicht>are you still using your modified debbugs, rekado_?
<jlicht>because I did not see it getting merged :/
<rekado_>I’m still using it.
<snape>by the way, mu4e users, when I search for a message, all messages in the same thread are displayed as well. Do you have that issue?
<snape>it makes the search pretty much useless
<lyr3>how do you guixsd users call .xsession? mine always issues a lot of "errors", I have to remove it...
<lyr3>I have no experience with login managers, but startx
<wigust>lyr3: Don't you use a Slim login manager?
<nckx>lyr3: Mine gets called automagically by slim...
<wigust>lyr3: It could read your .xsession which should have a last line calling a window manager
<rekado_>I use .xsession. It needs to be executable and it needs to end on an “exec my-window-manager” line.
<wigust>BTW .xsession should have executable permissions
<rekado_>snape: you can hide related messages with W
<lyr3>wigust: yep,slim.
<lyr3>rekado_: Oh interesting,maybe that is why
<snape>rekado_: oh. Thank you so much!! Very useful
<nckx>Huh. +1.
<snape>I don't understand why mu4e-headers-include-related defaults to t...
<Copenhagen_Bram>Hi! The nmtui worked
<roptat>Copenhagen_Bram: cool :)
<civodul>'guix substitute' seems to occasionally hang for me since core-updates
<civodul>anybody else experiencing it?
<efraim>My guix cant find the system command since the store patches, but no one else has said anything
<rekado_>civodul: haven’t noticed this.
<lyr3>what is the difference of (%desktop-service) and (%base-services)? or where I could find it?
<lyr3>found haha
<civodul>efraim: guix obtained by 'guix pull'? or from your checkout?
<rekado_>oh, I’m having new problems with “guix: offload: command not found” in the cluster environment.
<civodul>rekado_: : guix obtained by 'guix pull'? or from your checkout? :-)
<rekado_>I configured it in “./pre-inst-env guix environment --pure guix --ad-hoc coreutils”
<civodul>could you check what config.log has to say about guile-ssh?
<civodul>perhaps "rm config.cache" before running ./configure
<rekado_>I make sure that “guixr” loads an environment where “guile-ssh” is available.
<nckx>I've added some fonts to my operating-system and reconfigured, but I can't get fc-list (TIL) or other programmes to find them — even after running ‘fc-cache -f’. Am I doing it wrong?
<nckx>(Is there a separate cache for system-wide fonts?)
<nckx>Hm. The files are right there in /run/current-system/profile/share/fonts/truetype.
<rekado_>civodul: it’s probably just something with the guixr wrapper script, which does a little too much
<civodul>rekado_: so that wrapper is still around? :-)
<rekado_>it does a few things: set the GUIX_DAEMON_SOCKET, GUIX_PACKAGE_PATH, prints a warning on the login node, and implements a simple “load-profile” command
<rekado_>other than that it merely calls “/gnu/remote/guix/pre-inst-env guix "$@"”
<rekado_>oh, and it sources a profile containing Guix runtime dependencies, such as gnutls or guile-json.
<rekado_>I don’t think this is still needed.
<rekado_>hmm, I don’t know why this is failing.
<rekado_>I see: @ substituter-started /gnu/store/fwghwzg6b3z6fahg97mf17rqr3fidhvk-r-shiny-1.1.0 /gnu/remote/guix/nix/scripts/substitute
<rekado_>followed by:
<rekado_>guix: offload: command not found
<rekado_>guix build: error: build failed: unexpected EOF reading a line
<civodul>so the issue is in the environment of guix-daemon itself
<civodul>like incomplete/incorrect GUILE_LOAD_PATH, etc.
<civodul>(you can use --no-build-hook as a workaround)
<rekado_>the daemon is started as “/gnu/remote/guix/pre-inst-env guix-daemon …”
<lyr3>hey, after install aspel (guix package -i) now my manifest is empty and guix package wont install anything...
<lyr3>(guix package: error: unsupported manifest format)
<lyr3>I can rollback, ofc, but it happens a lot so maybe its wiser to file a bug or there any tip to avoid that?
<rekado_>now I remember why I used that separate profile containing gnutls and guile-json: with only pre-inst-env “guixr build anything” complains about a missing gnutls.
<rekado_>I find this weird.
<rekado_>I start the daemon with “/gnu/remote/guix/pre-inst-env guix-daemon” — should this not be enough?
<rekado_>I see in /proc/pid/environ for the guix-daemon pid that GUILE_LOAD_COMPILED_PATH is set, but not GUILE_LOAD_PATH
<rekado_>oh, I take that back. Both are set to /gnu/remote/guix
<rekado_>and that’s not right.
<rekado_>sourcing the profile containing gnutls before running the daemon fixes this, but results in the “guix: offload” error.
<rekado_>oh well, I’ll just disable the build hook in the daemon command.
<rekado_>hmm, that does not work.
<rekado_>passing it now as a default option with “guixr”, but that’s ugly :-/
<lyr3>wow, icecat is way better and enough than I previously thought. haha great work!
<civodul>rekado_: another possible issue is that the guile-ssh that's in GUILE_LOAD_PATH of the daemon is the wrong one (linked against a different libc)
<rekado_>hmm, I find that unlikely, because I used the updated Guix to re-build the profile containing guile-ssh.
<rekado_>but I’ll look into this at some later point.
<rekado_>thanks for the hint!
<rekado_>the guile-fibers tests take a very long time.
<rekado_>I wonder if we need to run them all (including the speedup demonstrations).
<janneke>i'm feeling a bit stuck with bootstrapping, on building the full gcc-boot after binutils/gcc-core-boot/glibc
<janneke>hoping to tap into some expert knowledge from you, maybe g_bor[m] or phant0mas?
<janneke>there seems to be a problem with LD and possibly crt1.o/crtn.o; possibly an installation or configuration problem
<janneke>glibc builds fine with this LD, however, the first gcc-core-boot also builds fine
<nyberg>I was wondering if the issue with guix pull failing on "Wrong type argument in position 2: #<unspecified>" would be the result of my local install or the sources pulled.
<nyberg>not so keen on selecting generation 1 again
<rekado_>nyberg: no, this is a problem in Guile. You need to try again. (Sometimes “--cores=1” helps.)
<g_bor[m]>janneke: Hello!
<g_bor[m]>I1m glad that you are making progress.
<g_bor[m]>Can we go over to bootstrappable and see what is not working?
<nyberg>rekado_: consistent faliure even on different commits, might just roll-back
<nyberg>specifically /gnu/store/6wjgrn737frsf2a02mgx25q1xhl576dl-compute-guix-derivation.drv
<janneke>g_bor[m]: great!
<pkill9_>anyone else unable to decrypt with gpg since updating?
<pkill9_>actually i'll try restarting brb
<pkill9>still not working
<pkill9>gpg: public key decryption failed: No pinentry
<pkill9>gpg: decryption failed: No secret key
<pkill9>it does correctly say what key it's encrypted with however
<sahithi-ihtihas>Hello Guix!
<nckx>Hullo, sahiti.
<sahithi-ihtihas> when i tried this part of code in ui.scm I got colorised string for string with "starting phase" but not for string with "phase"
<sahithi-ihtihas>Changes I made is I added spaces to substrings , This worked for "starting phase" but not for "phase"
<atw>pkill9: I think I had to install pinentry as well as gnupg
<pkill9>i already have it installed though
<sahithi-ihtihas>Can someone help me, where did i go wrong
<nckx>sahithi-ihtihas: This is just a wild guess, but aren't you expecting two spaces after ‘phase’?
<nckx>i.e. your first regex has ‘)(’, the second ‘) (’.
<nyberg>hmmmm, guix pull --bootstrap fails too
<nckx>Does having the same package in both inputs and native-inputs ever make sense?
<lfam>Hmm... a large number of crypto library updates
<g_bor[m]>pkill9: You also have to configure gpg to use the pinentry you provided. Most probably you want pinentry-tty or pinentry-gtk2.
<pkill9>how do i configure it to use that pinentry?
<rain1>lfam could it be related to the floating point bug
<rain1>even though the fix is in the kernel
<lfam>rain1: I don't think it's related but I'm not sure yet. It's a new side-channel for extracting keys
<g_bor[m]>pkill9: you create a file ~/.gnupg/gpg-agent.conf with content 'pinentry-program /home/your_user_name/.guix_profile/bin/pinentry-gtk-2' in case of pinentry-gtk2.
<rain1>gosh. I wasn't aware of this. so many bugs found recently
<g_bor[m]>You can adjust that path when using pinentry-tty.
<g_bor[m]>(that should be on one line, I'm not sure about IRC line break behavior)
<g_bor[m]>sahithi-ihtihas: nckx is probably right. The line most probably does not match this regexp.
<pkill9>ah thanks g_bor[m], I had already specified it but recently changed it
<pkill9>changed where the pinentry program is i mean
<pkill9>forogt you need to specify it
<roptat>civodul: I've made the change we discussed this morning
<roptat>should I regenerate the pot files?
<roptat>(I can't remember how)
<lyr3>yeah, emacs 26.1 arrived
<civodul>roptat: you don't need to regenerate them, unless you want to send them to the TP
<lyr3>flymake is back again
<roptat>civodul: I'd like to make sure my changes are correct
<roptat>also, is "guix: ui: Allow translation of dates." a good commit comment?
<civodul>then you can run "make update-pot" (?) in po/guix
<civodul>it's a pretty good commit log IMO :-)
<civodul>you can include the bad translation example in the log as well, to clarify thing
<roptat>update-pot doesn't exist
<roptat>yes, I found it :)
<roptat>civodul: I wonder how that could work in Japanese
<roptat>for instance, it should be "21日4月2018年" for april the 21st
<roptat>(I hope it shows correctly)
<roptat>anyway, I'll push my changes
<civodul>roptat: this example should work
<roptat>oh, a reboot fixed my locale issues \\o/
<civodul>i'm more concerned with non-Gregorian calendars
<civodul>i think we should develop extensions to srfi-19 for other calendars
<g_bor[m]>hello guix!
<g_bor[m]>I'm now into fixing libressl. What is the better way? Add a patch, or modify origin to use a commit from the git repo?
<g_bor[m]>I would favor using the git repo.
<lfam>g_bor[m]: It's at the implementer's discretion, in my opinion :)
<lfam>Some considerations: How many patches would there be? If there are many, that's annoying. How big is the Git clone? If it's very large, it's better to use a patch to save bandwidth and time.
<lfam>My copy of the libressl-portable Git repo is only 3.7 MiB, whereas the OpenSSL repo is >300 MiB.
<civodul>that big a difference, woow
<lfam>Yes, I wonder what's up there
<lfam>It's a fork but they must have truncated the history
<lfam>Yeah, the earliest libressl commit is from when they forked in 2014
<lfam>I'm wishing we had some way to maintain a list of packages and check them off as we process them...
<g_bor[m]>lfam: never mind, it seems that there is a release already.
<lfam>Great :)
<nyberg>Since the error persists no matter which options, commits, and rolling back hasn't worked so far I'm looking at reinstalling or diving in to find the error
<civodul>lfam: what do you mean by "as we process them"?
<civodul>you mean when reviewing patches?
<civodul>nyberg: what error?
<lfam>civodul: This crypto side-channel bug affects a lot of our crypto packages. It's just a little tedious to keep track of which packages have been reviewed and fixed, especially since I am trying to delegate this work to a group :)
<lfam>ACTION rm guix-register
<civodul>oh i guess i need to catch up on this one
<civodul>is this the recent libgcrypt CVE?
<civodul>oh and you update the old gpg as well
<lyr3>oh that explain clearly: In addition, ~/.xsession files are honored. When available, ~/.xsession must be an executable that starts a window manager and/or other X clients.
<lfam>It's this thing:
<lfam>I don't think the gpg-1.4 update is related
<nyberg>civodul: guix pull fails with either or
<nyberg>depending on the commit used
<lyr3>hey, where I find what services and package %desktop-service install? I want to change to %base-service but I will remove just some packages from %desktop
<civodul>lfam: interestingly this page is all JS...
<lfam>Maybe they are testing their exploit!
<civodul>there's a trick to (ab)use valgrind to detect non-constant-time implementations
<rekado_>ACTION prepares a short presentation on Cuirass and the GWL for the team at work
<civodul>using the uninitialized memory thing in Valgrind
<civodul>rekado_: Cuirass! woow
<rekado_>I’m beginning to use Cuirass to offer CI for everyone’s manifests.
<civodul>nyberg: the second error looks familiar, let me see
<rekado_>that’s easy, because people don’t use manifests.
<civodul>ah ha :-)
<roptat>civodul: that hack with valgrind sounds interesting, do you have resources?
<civodul>nyberg: what if you just run "guix pull" alone? current master should not have the problem with the 'invoke-error?' thing
<OriansJ``>incase anyone here missed this bit of news:
<civodul>roptat: it's from Schwabe, djb et al., see the talk by Schwabe at
<civodul>"Verifying ECC software"
<nyberg>civodul: Current master
<civodul>nyberg: thanks, let me investigate a bit
<civodul>nyberg: what do you have in /etc/guix/machines.scm?
<nyberg>currently commented out all machines
<reepca-laptop>hm, which package could I find the dc command in?
<civodul>nyberg: ok, so you must either delete it, or add '() at the end so that it returns the empty list
<civodul>here 'guix offload' crashes because machines.scm returns something that's not a list of machines
<civodul>this should be handled gracefully though
<nckx>reepca-laptop: bc contains a ‘dc’ executable.
<nyberg>civodul: Thank you, it's building
<nckx>Seems to be the only one.
<reepca-laptop>nckx: ahh, thanks.
<nyberg>civodul: would a patch for returning a slightly better error than the current be merged? I wouldn't mind writing it.
<rekado_>nyberg: a better patch would be to make it not fail in this case.
<rekado_>i.e. an empty file or a file with only comments should behave like one that contains the empty list.
<lfam>I just learned that crypto++ includes an implementation of Shamir's Secret Sharing Scheme
<efraim>is that the one that shards secrets?
<civodul>nyberg: i'm working on it :-)
<lfam>efraim: Yes. I looked recently and couldn't find an implementation among the Guix packages
<bavier>I get an "wrong type to apply: #f" error for packages that use svn-fetch :(
***Guest93996 is now known as Pigeon
<bavier>e.g. for `guix build -S netpbm`
<Copenhagen_Bram>Hi guys
<Copenhagen_Bram>ACTION is trying out ratpoison
<lyr3>should not guixsd iso image come with wget installed? it would so useful if we just download our config.scm and voila
<reepca-laptop>... I ran "guix package -i bc" and now it's downloading ghc. Interesting. "guix graph --type=derivation | grep -- '-ghc'" doesn't show anything. Wonder why it's doing that.
<lfam>bavier: Does it happen with other svn-fetch packages, or just netpbm? Or perhaps just packages hosted on sourceforge?
<lfam>lyr3: The tool `guix download` can be used as a basic downloader
<reepca-laptop>(err, with "bc" added before the pipe...)
<lfam>lyr3: But you're right, maybe it's time to reevaluate what people need in the image
<efraim>it can be installed with 'guix package -i wget' but even I forgot about that the last time i installed :)
<nyberg>rekado_: I'll look into it.
<lyr3>efraim: last time I (tried) to install anything under the iso installer it run out of memory hahaha
<Copenhagen_Bram>lyr3: it will if you git clone guix and edit install.scm
<Copenhagen_Bram>and then run git system disk-image on it
<Copenhagen_Bram>That's what I did. My guixSD usb has links, wget, and weechat on it
<lfam>reepca-laptop: That's strange! Can you make a bug report?
<lyr3>Copenhagen_Bram: the one under /gnu/system?
<Copenhagen_Bram>lyr3: yup :)
<lyr3>interesting. ok,I will generate my own iso then! haha
<reepca-laptop>lfam: aye, will do
<bavier>lfam: yes, other svn-fetch packages too, like e.g. many of the texlive packages
<bavier>lfam: and the command errors out before it even gets to the download part
<bavier>reepca-laptop: perhaps the cabal profile hook is mistakenly firing
<lfam>bavier: It's working for me with the java-cisd-args4j package
<civodul>or perhaps it's rightfully firing
<bavier>indeed, perhaps rightfully
<bavier>lfam: thanks for checking on your side
<lfam>It also works for me with netpbm. So I don't know what's up :)
<bavier>I'm not sure how to debug further, because the backtrace has a bunch of catch/gexp/run-with-store in it
<Copenhagen_Bram>lyr3: You'd better put curl and netcat on it too, I wish I had. Then I could use or from guixsd live
<Copenhagen_Bram>Oh and I don't think it's an iso, it's actually a raw image
<Copenhagen_Bram>Also don't bother changing your directory to where you'd like to put the image, because it just creates the image in /gnu/store
<reepca-laptop>huh, /gnu/store/sdqm4va5rfgq1m7gwznay1i84lrq9vjw-ghc-8.0.2.drv is listed in the "the following derivations will be built" part
<Copenhagen_Bram>Hmm. I wonder if you could disk-image config.scm and then dd it to the hard drive you wish, for instant easy guixsd installation
<Copenhagen_Bram>Oh and another tip learned from my first days of guixsd: You might consider adding cryptsetup to your config.scm, for some reason it's actually not included in an installation
<reepca-laptop> this is what "guix package -i bc" has produced so far
<Pigeon>Hello! Could someone help me with installing guixsd on vm please? I would like to have enctrypted root partition (sda1), and unencrypted /boot partition. Both are ext4 formatted, but after running "guix system init /mnt/etc/config.scm /mnt" there are two warnings "File system ext2 doesn't support embedding", "Embedding is not possible. GRUB can only be installed in this setup by using blocklist[...]", and an error "will not proceed with
<vagrantc>Pigeon: a split /boot partition isn't really supported
<vagrantc>ACTION forgets if a bug was filed about this or not
<Copenhagen_Bram>vagrantc: I found that out the hard way (I think). But I didn't get any ext2 error. It just installed and then failed to boot
<Copenhagen_Bram>Although, I recall making the mistake of typing `try_user_config ahci0` instead of `try_user_config ahci0,msdos1`. Who knows what would've happened if I did it like that?
<vagrantc>Pigeon: grub can support decrypting the / partition where /gnu/store is
<nckx>Sounds more like the error you get when you're missing the BIOS boot partition on a non-UEFI system, not /boot.
<lyr3>Copenhagen_Bram: noted! >
<Copenhagen_Bram>Pigeon: Does guix system init fail to run and then give you that error, or does it go through the whole installation and then when you boot you get that error?
<Copenhagen_Bram>lyr3: huh?
<nckx>*not /boot-related.
<lyr3>curl/netcat tip into my file
<Copenhagen_Bram>nckx: either there's an invisible boot partition because cfdisk says the start sector is at 2048, or I have uefi but you don't need a /boot/efi partition for efi
<Copenhagen_Bram>because I have just one partition on my drive
<nckx>Copenhagen_Bram: I was referring to the GRUB error message above.
<Copenhagen_Bram>oh it's a grub error message, interesting
<Copenhagen_Bram>hmmm, maybe it's possible to manually load the encrypted partition from the grub command line
<Copenhagen_Bram>i peeked into the grub config, it looks like it has to load /gnu/store/hoobitybabbityhashes-linux-libre.version
<Pigeon>I don't exactly know what am I doing. Just following installation manual, but It doesn't show how to install botloader, while using encrypted system partition. It only shows how to encrypt it.
<nckx>Pigeon: A separate unencrypted /boot isn't supported yet.
<nckx>Any separate /boot, for that matter.