IRC channel logs


back to list of logs

<ng0>Question: does the Guix homepage encourage users to get back in touch with us if any problem occurs or to tell us of anything they want to see improved, changed, fixed, when they are not completely familiar with anything being used? some "feedback" should also be in consideration when improving the tools (where toolset also includes reviewing etc where I could do more)
<catonano>ng0: would you mind publishing your work anywhere ? I could try to help you. Also
<catonano>Also other guixers could try to help ypu
<ng0>work.. the work in progress packages or the longterm gnunet pull etc thing? for the longterm one i need to work on things.. I can move stuff into one repository (I have 20 - 30 guix repos) with many branches named accordingly and let users pull from my server. one thing which needs badly help by someone dedicated enough to want it is the dbacl packages ... you need to either disable all tests (bad idea imo) or
<ng0>fix 200+ tests
<ng0>i could let my server repository mirror to also notabug and if necessary even to github, though i like the workflow we established with the gentoo overlay and really dislike pull requests and all that weirdness. a pull request in our workflow is "when you have a fix/addition and wnat to participate, do it, message us with either a patch or a direct url to your git server in onion space or clearnet"
<ng0>catonano: i'll try to publish it next week or next month, whenever i have time. some appointments i have soon are in the way.
<catonano>ng0: take your time, it was just an idea
<ng0> this has a quoted part of the "readme" file, which should show what i mean :) A todo on my side for a long time now is to make a more generic explanation on the setup.
<catonano>ng0: the setup of what ? Of a git service running on an "onion" ?
<ng0>no, the workflow.
<ng0>i made no progress with vlc. i know that whatever we had to create plugins.dat needs to be run before the tests, not after install.
<ng0>no visible progress
<ng0>lst thing for today: i believe pybitmessage can be faster packaged if I ignore the discussions with upstream. I can try to contribute my …again… and hope that it does get included, but this should really not be my work, it should be them doing it.
<catonano>ng0: gotta go now. Good night
<ng0>oh, 20-30 is just the packages i managed to keep track of.. there are more in the backups and ones which just sit unchanged because of some dependencies, like rust etc which i need to monitor for the day the dependencies are funcitonal
<ng0>i hope rust had its release now where it integrated the switch which would enable us to build it
<Sleep_Walker>can I somehow easily express, that my initrd should contain some additional packages? (busybox in my case as scheme is hard and bournish not sufficient)
<Sleep_Walker>it would really help me with debugging
<civodul>Hello Guix!
***kelsoo1 is now known as kelsoo
<ng0>hi. need to do some paperwork today, but I just noticed that after a guix pull this morning, building notmuch fails with gpg not found
<ng0>notmuch? or gmime..not sure
<ng0>i'll paste the log, maybe someone feels like fixingit
<ng0> |
<civodul>ng0: sounds like it's due to the gpg2 -> gpg rename
<ng0>gpg2 binary is now gpg symlinked or the other way around?
<ng0>iirc the switch is gpg-is-gpg2
<civodul>i think gnupg@2.1 provides 'gpg' only, not 'gpg2'
<civodul>gnupg@2.0 provides both
<ng0>163708a664276dc4dae8180992afae26159434f5 i see.
<ng0>i only have gpg2.1 in this profile. the switch should not break other applications, or we have to fix those packages
<civodul>yeah we have to fix gmime
<ng0>are we on 4.6.4 linux-libre already?
<ng0> this fixes suspend apparently.
<ng0>but my system config is broken so i can not reconfigure
<ng0>something with swap-devides breaks
<ng0>same config i used for setting this system up.. weird.
<nee`>Can I burn the guixSD image to a DVD? Is it possible? I have an old computer that can't boot from usb and wanted to quickly check if it runs guixsd. It doesn't have grub installed, so I can't use that to boot from usb.
<ng0>can it not boot from usb?
<ng0>sorry i skipped that part of the sentence
<ng0>could you dd the image of guixsd to an harddrive and go from there?
<ng0>there's nothing at the moment which can produce an bootable optical disk for the way guixsd releases are produced.
<efraim>I have a machine like that, I had to use plop to boot from usb
<nee`>ng0: Okay thanks. The HD way is possible, but too much work right now.
<efraim>i have to look at it more actually, I don't know what its license is
<efraim>plopkexec is gpl2 according to the license file
<efraim>nee`: you could burn plopkexec to a cd/dvd and use that to run the guixsd installer from usb
<ng0>the problem with gmime is the part where gpg gets replaced by gpg2 in the added patch phase
<ng0>i did not write it, have to look at it, but maybe i can fix it. the paperwork is boring and annoying.. can't force myself off from distraction
<ng0>notmuch depends on gmime? then i can testbuild that afterwards.. maybe it was just this one line to remove
<ng0>does line 250 in gnu/packages/mail.scm serve any other purpose than replacing gpg occurences with gpg2?
<ng0>apparently yes.
<iyzsong>it should also replace '/bin/rm' and '/bin/mkdir'.
<ng0>i don't see how i can change 250 and 253 to leave gpg be gpg but let it do the rest it did
<ng0>or not yet
<iyzsong>replace '(if ... base)' with 'base'?
<ng0>got it
<ng0>i can check in a minute what i did set
<ng0>changing 250 to: (prog (which (if (string=? base) base)))) worked, but it is obviously too muhc in the line. just (which base) did not work
<ng0>what would you suggest to change before i send the patch?
<iyzsong>so you get error with '(which base)'? did that mean 'gpg' is not provided by gnupg?
<ng0>gpg is now provided by gnupg, gpg2 no longer
<ng0>i can try (which base) again to be sure
<iyzsong>ah, ok :-
<ng0>my bad. which base works.
<ng0>or it looks like it
<ng0>building nochmuch to be sure
<ng0>I'll be sending the patch in the next minutes
<koosha>I wanted to install Scheme , which package should I get ? (In Debian and Guix)
<civodul>koosha: there are many Scheme implementations, but i recommend "guix package -i guile" :-)
<civodul>or "sudo apt-get install guile-2.0"
<ng0>has there been a bug report on gmime or am i the person who did report and fix at the same time?
<ng0>idid not check emails
<civodul>koosha: to search for Scheme implementations, you can run: guix package -s scheme | recsel -e 'location ~ "scheme"'
<civodul>ng0: i think there hasn't been any report other than yours
<koosha>civodul: Oh , okay . So I install Guile . Thank you .
<ng0>i think the lispf4 bug is still open.. does someone with knowledge in the languages it is written in care to investigate why it fails to execute correctly on guix? or does someone know appropriate mailinglists to ask for people to look into this?
<ng0>i should get the gnunet vpn setup on the systems and stop using tor.. i'm hitting so many man-in-the-middle recently exits it's no longer funny.
<ng0>oh.. yeah. do i need to sign my commits now, even if i at this moment did not connect my savannah account with guix and did not ask for access (I'd rather gather much more experience in some parts of git workflow and then ask)
<ng0>or is the new requirement not affecting this as long as commiters sign it as usual
<civodul>ng0: it only affects committers
<wingo>do you just have to sign the HEAD commit that you push?
<wingo>or do you have to sign more commits
<civodul>wingo: all the commits that you push
<civodul>there's info in HACKING on how to set it up
<ng0>maybe link to even
<ng0>but for me it's not "git commit -a -S -m 'message' , it is -s
<wingo>kids, my lawn, etc etc
<wingo>i'll figur it out at some point :)
<ng0>would there be any trouble in sending in signed commit patches for the people who have to commit it later then?
<ng0>is libreoffice binary substitute broken? I get no "libreoffice" binary when it finishes grafting.
<davexunit>ng0: that's because libreoffice doesn't provide one
<ng0>oh.. what's it called?
<davexunit>ng0: check the /bin directory of the output
<davexunit>it's soffice or something weird like that
<efraim>I have something in my .git/config that makes me sign everything
<Sleep_Walker>I put that into ~/.gitconfig
<ng0>i put it many things into aliases. too many guix checkouts, too many other repositories which differ.
<Sleep_Walker>I have something which can kernel config experiments but I'd really appreciate help with adding more package outputs into initrd
<Sleep_Walker>hm, I could inject it into helper-packages
<efraim>CVE against perl
<efraim>there's a real url at the end, that's what the mail header showed
<lfam>efraim: Are you looking for the patches?
<paroneayea>hi everyone! so here's a thing about geiser(-next)
<ecraven>is there any way to have "dynamic" dependencies? say, if X is installed, then compile this way, if Y is installed, this other way, and if X or Y are installed or removed, rebuild (but neither X nor Y are required dependencies)?
<adfeno>ecraven: I'm not an expert on this (I'm not even a programmer) but, would this break reproduceability of the package?
<efraim>I can't today, movers come tomorrow so I have to finish packing up the apartment
<paroneayea>geiser-next supports mit-scheme
<paroneayea>but, it requires that a file be compiled *by* mit-scheme and shipped with geiser-next
<paroneayea>which would make mit-scheme a geiser dependency
<paroneayea>I'm not sure how many in #guix would be thrilled by that, because it would mean pulling in texlive-texmf
<lfam>efraim: Okay, I'll look into it
<lfam>Good luck with your move!
<paroneayea>I could make a derived package like
<paroneayea>geiser-mit-scheme or geiser-all or something
<paroneayea>I don't think it's possible to do the dynamic loading thing, afaik, ecraven
<paroneayea>(btw this is the same thing ecraven is asking about :))
<paroneayea> more info
<adfeno>Hi shoaloak :D
<shoaloak>I am trying to install GuixSD on Oracle's Virtualbox
<shoaloak>Everything went well thus far
<shoaloak>but when i ran guix system init /mnt/etc/config.scm /mnt
<shoaloak>it started doing a bunch of stuff
<adfeno>People, can someone help answer ecraven's question up there? ↑
<shoaloak>and now i have the feeling it is hanging on jibberish-base-initrd.drv
<lfam>adfeno, ecraven, paroneayea: Sounds like a case for the mailing list :)
<ecraven>hm.. I understand the problem with reproducability :-/
<lfam>shoaloak: It appears to be doing nothing?
<adfeno>ecraven: Keep in mind that it was my own opinion.
<adfeno>I don't have any reference to that.
<shoaloak>i see no activity
<paroneayea>adfeno: yes, I think it would break reproducibility also
<paroneayea>but maybe I should email the mailing list
<paroneayea>should I open a bug or should I just email devel for now?
<paroneayea>I guess it depends on the decision ecraven makes also on whether this ends up being a thing guix has to do anything about
<shoaloak>oops wrong link
<paroneayea>esp if we have the option where it just (load)'s things
<adfeno>shoaloak: Wrong link again?
<lfam>That link works for me
<shoaloak>no that link is correct
<adfeno>Oh OK
<shoaloak>as you can observe, no internet or hdd activity
<shoaloak>can i interrupt the proces and try again?
<shoaloak>or should i first check some stuff?
<lfam>shoaloak: I haven't had that issue before, although I've never used VirtualBox. Guix has great integration with QEMU so I use that. In this case, I would try looking at the process list (`ps`) and then tracing what I think is the correct process, to see what it's doing
<paroneayea>unrelated: magit users, how are you handling the new "sign all your commits" workflow?
<shoaloak>how am i supposed to do that?
<shoaloak>is screen enabled
<shoaloak>or does it have ttys?
<lfam>shoaloak: You can go to another tty
<lfam>shoaloak: You will probably need to install strace: `guix package -i strace`
<efraim>sorry, commit.gpgsign=true
<lfam>Or whatever tracer you want to use
<lfam>shoaloak: I haven't heard about this issue before. It might be worth a bug report.
<shoaloak>hmm maybe
<shoaloak>but maybe i just fudged up my config.scm
<lfam>Perhaps, but I don't think it should cause this issue
<shoaloak>@ bootload i put device "/dev/sda"
<shoaloak>is this correct?
<shoaloak>or should've put /dev/sda1?
<lfam>shoaloak: /dev/sda is correct, assuming that's the name of the device
<shoaloak>it is
<shoaloak>crap than it probably is a bug
<lfam>We don't test on VirtualBox, so it's possible
<lfam>shoaloak: Bug reports go to :)
<shoaloak>Okay than i will file up a bug repport.
<shoaloak>shame though, was curious to see the GuixSD system.
<shoaloak>And my laptop unfortunatly does not support qemu/kvm
<lfam>shoaloak: QEMU doesn't require KVM
<lfam>Although it does improve the experience...
<lfam>Yes, really :)
<shoaloak>than maybe there is still hope
<lfam>When you invoke the QEMU emulator, you have to pass '-enable-kvm' if you want to use KVM
<adfeno>shoaloak: Check that out, use QEMU, :D
<shoaloak>okay i installed qemu on my host
<shoaloak>but i have no experience /w it (as you may have guessed)
<shoaloak>how should i run the image?
<shoaloak>qemu-x86_64 guixsd-image ??
<shoaloak>nope that fails
<lfam>shoaloak: There is a pending addition to our manual that documents how to do this. Try it out and give some feedback so we can merge it!
<lfam>It's not so easy to read in that format...
<lfam>I think the important parts are "create a qcow2 formatted image file" and "Boot a virtual machine with the USB installation image"
<lfam>shoaloak ^
<shoaloak>thanks will try it
<lfam>I'm on Debian, and the networking didn't require me to inspect the kernel settings. I don't know if you distro supports the features "out of the box"
<ng0>can someone take a second approving look at the gmime patch?
***Digitteknohippie is now known as Digit
<efraim>python2 on core-updates failed on mips64el
<koosha>Hello friends !
<koosha>What interpreters are available for Guile ? I used the guile itself but it didn't satisfy me .
<davexunit>koosha: well, you can use elisp on guile 2.1.x
<davexunit>but scheme is better than elisp, so if scheme didn't satisfy you then I doubt scheme will.
<davexunit>er, I doubt elisp will.
<koosha>Sorry I didn't ask my question truely :D I mean the envierment .
<koosha>davexunit: I'm not comfortable with the software envierment .
<alezost>koosha: is it like python is not satisfied, and ipython is better? do you mean something this?
<koosha>alezost: Exactly !
<alezost>koosha: well, there is only 'guile' for CLI, but if you use Emacs, you may look at Geiser
<alezost>I personally never use 'guile' in console, I always use Geiser
***kelsoo1 is now known as kelsoo
<koosha>alezost: Also It's possible to write one with probebly python which can use guile itself for interpreting . Right ?
<alezost>koosha: I'm not sure I understand but there is "wisp" which is guile but looks like python. But all these are questions for #guile
<koosha>alezost: Oh , okay . I ask it there . Thank you :)
<nidhogg>I am trying to configure a menu-entry in grub configuration. My linux kernel is at "(hd0,3)/boot/vmlinuz-linux-libre". But when I specify that in "linux" under "menu-entry", the generated grub menu has "(hd0,3)/boot/vmlinuz-linux-libre/bzImage" as the kernel. How can I avoid this extra bzImage from being appended?
<lfam>I can figure out exactly which patches I need to cherry-pick from the Perl Git repo to patch against CVE-2016-1238 and CVE-2016-6185. Apparently the fixes are on the 'maint-5.22', which corresponds to our package, but it's not clear to me which commits are relevant.
<lfam>Bad typo. I meant to say that "I can't figure out ..."
<lfam>And Perl just made the bug report private again... :(
<Sleep_Walker>does anyone have idea, why this static package has glibc and bash-static as dependency?
<davexunit>Sleep_Walker: libc is always dynamically loaded
<davexunit>bash-static, dunno.
<Sleep_Walker>davexunit: but statically linked package doesn't require it...
<davexunit>Sleep_Walker: glibc is an exception.
<Sleep_Walker>one of us is very wrong now :(
<davexunit>I don't know a lot about static linking, but everything I've ever read says that glibc cannot be statically linked.
<Sleep_Walker>AFAIK static linking means take all the code and put into one file (including library functions where glibc can be one of them)
<Sleep_Walker>my guess is that glibc is implicit dependency so I have to set inputs to empty list
<Sleep_Walker>surprisingly bash-static is dependency of glibc
<Sleep_Walker>now just to figure why the glibc is there
<adfeno>Wow! :D
<Sleep_Walker>really - see gnu/packages/base.scm:622
<davexunit>Sleep_Walker: you are correct about static linking, but what I'm saying is that you can static link everything *BUT* glibc
<davexunit>that's just the way it is
<Sleep_Walker>well, yes, that wouldn't be useful
<Sleep_Walker>but again, glibc is not required but pulled as dependency
<davexunit>it is required
<davexunit>your binary won't run without it
<Sleep_Walker>do you want my binary to try? >:)
<davexunit>you need a C runtime
<Sleep_Walker>you can use busybox-static built against uclibc on system with glibc (and without uclibc)
<Sleep_Walker>OK, different approach
<Sleep_Walker>busybox.scm - no inputs, just some native-inputs
<Sleep_Walker>from documentation:
<Sleep_Walker>inputs (default: '())
<Sleep_Walker>so where glibc comes from?
<davexunit>keep in mind that there are implicit inputs
<davexunit>provided by the build system
<davexunit>like gcc, glibc, make, etc.
<davexunit>the exact inputs depend on the build system in question
<Sleep_Walker>but all these should be native-inputs, right?
<davexunit>I'm using "inputs" generally
<davexunit>gcc, make, etc. would be native inputs because they need to run on the build machine
<davexunit>but glibc would be an input because it needs to be compiled for the target machine architecture
<Sleep_Walker>bash-static depends on bison O_o
<Sleep_Walker>hm, I went to deep, time to hunt some food
<Sleep_Walker>davexunit: thanks
<lfam>Hmph, the patch series for perl CVE-2016-1238 doesn't apply cleanly to our source code
<davexunit>Sleep_Walker: when you say "depends", do you mean that it references it or just that it is an input to the build?
<davexunit>it's unlikely that bash needs bison at runtime.
<Sleep_Walker>guix package -s bash-static | grep bison
<lfam>The patch application failures result in a bunch of '.rej' files that look like this:
<davexunit>Sleep_Walker: it's a native input.
<davexunit>which makes sense.
<davexunit>what do you find odd about it?
<baconicsynergy>im having trouble finding the list of service modules in the manual. can anyone point me in the right direction?
<baconicsynergy>im attempting to modify ntp-service-type from %desktop-services, but it says its an unbound variable...
<baconicsynergy>this is what im trying to do
<davexunit>baconicsynergy: ntp-service-type is not defined in (gnu services desktop)
<davexunit>you need to import the correct module
<davexunit>which is (gnu services networking)
<davexunit>ACTION goes afk
<lfam>baconicsynergy: Also, I believe the variable is named openntpd, not openntp
<baconicsynergy>is there a scheme variable for that module, like %networking-services?
<lfam>I'm not sure. Maybe bavier (who made that paste) has some advice :)
<lfam>baconicsynergy: If %desktop-services is unbound, you probably to import that service module. (use-service-modules desktop)
<lfam>There are other examples of that in the 'doc/os-config-*.texi' templates
<baconicsynergy>oh i didnt know those templates existed!
<lfam>Heh, really?! I'd have been totally lost :)
<lfam>Even more than usual ;)
<baconicsynergy>lolol. where are these docs, specifically?
<lfam>In our Git repo. Do you have it cloned locally?
<baconicsynergy>im about to!
<lfam>Cool :) Those templates get used to generate examples in the manual
<lfam>baconicsynergy: It looks like you need to build Guix from the Git repo in order to generate those example files.
<Sleep_Walker>CVE-2016-1238 looks trivial
<lfam>One important thing to note: when you do ./configure, make sure to `./configure --localstatedir=/var`, assuming your localstatedir is in /var/guix
<lfam>Sleep_Walker: Can you reply to the ML thread with your advice? :) I'm going AFK for a bit
<Sleep_Walker>lfam: I'm just on lunch, I'll try to send updated patch