IRC channel logs


back to list of logs

<EternalZenith>It's been quiet on here today
<atw>hey EternalZenith! Did your question about shepherd config files get an anser?
<EternalZenith>atw: More or less
<EternalZenith>It seems that there isn't one recommended way to start shepherd on user log-in
<atw>oh, I think I looked into that! at the moment I start it manually because I've slacked at automating it
<atw>When I was looking at it previously, I read and (guix services shepherd)
<EternalZenith>I can't seem to get Network Manager to reconnect to my saved wifi networks after my system wakes back up from sleep
<apteryx>EternalZenith: wouldn't that be 'shepherd' in your ~/.bash_profile?
<apteryx>(the solution to auto-starting a userspace instance of shepherd?)
<EternalZenith>apteryx: That's pretty much what I've settled on, but in .zprofile since I'm using zsh as my login shell
<apteryx>good! and that works as expected?
<atw>apteryx: and that works for graphical and non-graphical login?
<EternalZenith>Actually, I'm not entirely sure everything would work if I logged in multiple times
<EternalZenith>Since logging in on a different tty would attempt to restart shepherd for my user, restarting every service in a potentially troublesome way
<pkill9>EternalZenith: i have the same problem, i have an inconvenient fix that points in correct direction
<pkill9>EternalZenith: this fixes it for me:
<pkill9>well, it's not a perfect fix, but it's what i use
<pkill9>basically in /etc/NetworkManager/system-connections/<ssid>, change "permissions=user:the_user:;" to "permission="
<pkill9>then it will autoconnect when it finds the wireless network
<EternalZenith>I did it, but that feels really hacky
<pkill9>i agree, but i don't know what the root of the problem is
<apteryx>atw: I don't see why it shouldn't work :)
<apteryx>although you might want to implement some check to an existing shepherd process to avoid running it multiples times in cases of multiple logins, as EternalZenith pointed out.
<apteryx> it seems the --pid option of shepherd would help in achieving such a check
<atw>thanks both for the tips!
<emacsomancer>I've had an unexpected reboot - are there logs I can check?
<lfam>emacsomancer: I would check the kernel log (/var/log/messages)
<emacsomancer>lfam: thanks. I don't really see anything other than the restart itself, but there's nothing unusual nearby
<lfam>emacsomancer: You might also check the shepherd log, in the same place
<emacsomancer>lfam: where is the shepherd log stored?
<lfam>In the same directory, /var/log
<emacsomancer>I don't see any log file for shepherd
<lfam>That's strange, but I don't know why it would happen
<g_bor>hello guix
<g_bor>reconfiguring to core-updates on commit 418f1b241486ef7f98fdce1e6f8e53f2e7863fd9 was succesful.
<g_bor>This is the only problematic output, but I think we are already tracking this:could not change directory to "/root/guix": Engedély megtagadva
<g_bor>The files belonging to this database system will be owned by user "postgres".
<g_bor>This user must also own the server process.
<g_bor>The database cluster will be initialized with locale "en_US.utf8".
<g_bor>The default database encoding has accordingly been set to "UTF8".
<g_bor>The default text search configuration will be set to "english".
<g_bor>Data page checksums are disabled.
<g_bor>initdb: directory "/var/lib/postgresql/data" exists but is not empty
<g_bor>If you want to create a new database system, either remove or empty
<g_bor>the directory "/var/lib/postgresql/data" or run initdb
<g_bor>with an argument other than "/var/lib/postgresql/data".
<civodul>Hello Guix!
*vagrantc waves
<vagrantc>civodul: managed to get my local overdrive1000 updated to a recent guixsd ... took an absurd time to build
<vagrantc>hadn't updated since june or so
<vagrantc>civodul: sounds like you've got what you need for the one you were setting up though?
<civodul>vagrantc: yep it's up and running!
<civodul>'guix pull' & co. did take some time but not unreasonably so
<ngz>Hello. Is there a special process to rename a package (both the variable and the "name" field)?
<civodul>or maybe i'm super patient ;-)
<civodul>ngz: the only thing to pay attention to is to add a deprecated alias for the name
<civodul>so typically (define-public oldname (deprecated-package "old-name" new-name))
<ngz>I see. Thank you.
<vagrantc>also just tested pinebook a bit, guilt u-boot 2018.11-rc2 + a few patches (that should land upstream) ... so once u-boot 2018.11 comes out, and linux 4.19 comes out, should be able to run guixsd on the pinebook
<vagrantc>so an aarch64 laptop
<vagrantc>a bit slow processor-wise and not a lot of ram ... but ok-ish storage available
<vagrantc>eMMC up to 128G, if i remember
<vagrantc>also took it as an opportunity to wrap my head about (patch (search-patches ...
<vagrantc>hmmm... maybe only up to 64GB eMMC
<ngz>I created a new in a package's snippet. I need to update "" accordingly, I guess. So I added "automake" to native-inputs and call (invoke "automake"). I get an un-parsable error (invoke probably fails). I also tried to run "automake --add-missing" after removing old "", to no avail.
<ngz>Is there something obvious that I am missing?
<ngz>I also added (guix build utils) to modules before the snippet.
<civodul>vagrantc: the pinebook looks interesting
<civodul>do graphics work with free software?
<civodul>ngz: i'd look for ways to avoid adding a .am file :-)
<civodul>otherwise if you really need to add it, "autoreconf -vfi" should do the right thing
<ngz>No, it insists on using libtoolize, which fails.
<ngz>I'd rather not create a new ".am" file, but could find another solution so far.
<ngz>could not*
<civodul>you need libtool as an input, then
<civodul>native input
<ngz>I'm pretty sure I tried that already.
<ngz>I also removed "configure" to force "autoreconf -vfi" (it'a gnu-build-system package).
<ngz>Yes, still an error in "invoke"
<ngz>Can't exec "libtoolize": Aucun fichier ou dossier de ce type at /gnu/store/zydmcjb9y417b255kany3dzsjyy40q9w-autoconf-2.69/share/autoconf/Autom4te/ line 345, <GEN3> line 7.
<ngz>That's the error when calling "autoreconf -vfi" in a Guix environment on the source of the package.
<ngz>libtool is a native package.
<ngz>err native-input
<roptat>hi guix!
<roptat>ngz: what does `which libtoolize` tell you?
<vagrantc>civodul: pinebook has framebuffer video for now, nothing accelerated, but X basically works ... some acceleration may be on it's way in mainline linux
<ngz>roptat: nothing, there is no libtoolize
<ngz>I tried to replace autoreconf with autoconf + aclocal + automake -ac but invoke still fails.
***rekado_ is now known as rekado
<civodul>vagrantc: ok, sounds good (it'd be good enough for my own use cases :-))
<civodul>ngz: did you add libtool to the native inputs?
<ngz>Yes, I did.
<ngz>I'm trying another way to avoid writing a new I hope this one will give results.
<vagrantc>civodul: similar board would be the olimex teres-i
<tune>my time is wrong on icecat. it might be using utc time. my system time is fine, though
<tune>any ideas?
<civodul>vagrantc: ok, didn't know that one
<civodul>tune: what do you mean by "my time is wrong on icecat"?
<tune>multiple sites that use something time-based display the time 5 hours ahead of my system time. my system time being fine
<tune>it's specifically a web browser issue
<tune>riot (chat with timestamps) as well as a site that shows power usage and has the time at the bottom of the chart are both implying it's 5 hours later than it is
<tune>I noticed riot having the issue a day or two ago and thought it was something specific to that, but now I noticed it on another thing
<tune>another icecat issue, all my checkboxes (as in, ANY site that uses a checkbox) shows a red square, and it's not clear if it's checked or not
<civodul>tune: if you go the "about:home" in IceCat, there's a bunch of options you can enable/disable
<tune> here's the red box issue
<civodul>tune: some of the privacy-protection options might tweak the timezone sent to servers so that you're more anonymous
<civodul>link above is 403 for me
<tune>hm my second issue makes this hard to tweak
<tune>well feel free to suggest another host, but clicking the link again opens it fine for me
<tune> how's this?
<ngz>Bah. It failed.
<tune>I'm wondering if I disabled geolocation and that's breaking the time, but I'm not sure because of the red checkbox issue...
<roptat>tune: I had a similar issue where checkboxes didn't appear
<roptat>I fixed that by removing my ~/.themes
<tune>I was thinking it might be a theming issue. I only seem to have one theme, one I manually copied off another machine. I'm installing gnome-themes-standard to see if it gives me some more choices
<roptat>and installing the theme I used with guix
<civodul>ngz: maybe you can send the package and all to help-guix
<joshuaBPMan>hello, I think guix is being weird this morning. guix package --fallback -u . --do-not-upgrade mpeg says this:
<joshuaBPMan>building /gnu/store/ypdnq8bsaa0a7iiwwmjpn0xnwm079i8m-ffmpeg-3.4.4.drv...
<joshuaBPMan>I'm trying to avoid building ffmpeg, because it takes forever to build. I've got substitues available from
<joshuaBPMan>maybe I need to get substitutes from hydra too.
<civodul>efraim: just saw that you fixed Kodi, thumbs up!
<civodul>joshuaBPMan: perhaps ffmpeg is a dependency of some other package?
<joshuaBPMan>civodul: idk. I've got it updated now. I've got all three substitute servers defined, so that probably helped.
<ng0>NSA, Berlin, Hydra :)
<civodul>ng0: in this order :-)
***astronavt_ is now known as astronavt
<joshuaBPMan>ng0: the NSA has a substitute server? really?
<joshuaBPMan>Also how hard do you think it would be to convince my large public university to set up a substitute server?
<ng0>idk.. worth a try asking
<roptat>joshuaBPMan: what's this third substitute server you were talking about, then?
<joshuaBPMan>just a second...
<joshuaBPMan>you can configure guixSD to use those substitute servers when you launch guix-daemon
<joshuaBPMan>ExecStart=/var/guix/profiles/per-user/root/guix-profile/bin/guix-daemon --build-users-group=guixbuild --substitute-urls=''
<roptat>ah, I see
<mbakke>joshuaBPMan (IRC): You should not use directly, the mirror has the same contents and is faster.
<joshuaBPMan>mbakke: thanks for the tip!
<mbakke>There is a third substitute server though, that's not well known: :-)
<joshuaBPMan>mbakke: a third one!? awesome!
<pkill9>how fast is it compared with the default one and berlin?
<pkill9>in terms of download speed
<mbakke>pkill9: I don't know, haven't used it in a while. It is located in France on a good ISP IIRC.
<mbakke>I think it uses the same signing key as berlin.
<joshuaBPMan>oh. that's cool!
<joshuaBPMan>also what are some cool channels? I'm guessing that we don't have a stable one yet. Do we have a bleeding edge channel? or a security minded channel?
<pkill9>i don't think there are any, except for core-updates, etc
<joshuaBPMan>pkill9: ok. cool
<pkill9>but there is mbakke's chromium channel
<pkill9>but building it will take a long time so you probably don't want to add that one, but it's pretty cool
<joshuaBPMan>so I can install chromium. that's cool. the last time I tried chromium my laptop started crashing all the time. I think chromium was eating all of my memory.
<pkill9>LTS, stable and bleeding edge channels would be really neat though
<pkill9>i'm curious what happens when channels with conflicting packages are added
<civodul>pkill9: the 'guix' channel takes precedence
<efraim>I'm thinking of moving around some of my enlightenment packages and tossing them up as a channel
<roptat>efraim: why not adding them to guix?
<efraim>Some of them are more proof of concepts, like a calculator, and the pinentry-efl package is currently unclear about the upstreaming of the patch(es)
<nly>How to work with nodejs in guix?
<demotri>nly: node as such us not a problem, it is packed and usable. The problem is that (standard) npm packages have so many dependencies, cyclic dependencies, and package maintainers don't care about software freedom/building from source/bootstrapping. There is some importer in a branch, but it 's nearly impossible to use.
<roptat>nly: to be more precise, you can run any npm command with the guix package, but we don't have node packages yet
<roptat>nly: I've been working a bit on it, so if you want to have a look, here is the latest version of node-build-system and node importer:
<roptat>(as you can see, I only slightly improved over existing patches from jlicht)
<nly>Thanks roptat
<roptat>I still have an issue with the new icecat though, I can't use webgl because of a driver issue
<roptat>I had the same issue with firefox 60 on another distro, and it is fixed now in firefox 62
<roptat>so not icecat's fault
<apteryx>Is there a way to load an environment from a saved profile?
<apteryx>or is this enough: source ~/.config/guix/afo/etc/profile
<nly>I just did something horrible i think
<nly>ln -sf $(which env) /usr/bin/env
<nly>To make node work
<lfam>Yes, it breaks some of the "promises" of Guix's functional model, but it's so common that it's the example used to document the 'extra-special-file' service:
<amz3>nly: if that is the only hack to make nodejs work, I am buying it. Did you manage to get "hot-reloading" working?
<amz3>IIRC it's npm module that requires a python library that requires a shared library.
<nly>Pretty much that did it
<nly>Not encountered that yet amz3
<brettgilio>Hi, all.
<Laalf>hello! quick question: what service would i use to simply execute a command at startup?
<roptat>Laalf: depends on what you want to run?
<janneke>civodul: just a heads-up...i cannot get the new core-updates-next+%bootstrap-mes to build, so this will take some more time
<Laalf>roptat: a crontab with @reboot. which is not possible.
<roptat>why not?
<roptat>anyway, if you want to run a command at boot, you can create a service for that :)
<Laalf>roptat: also i never got mcron to work
<Laalf>roptat: yes but what service? i basically want to run screen -dmS name program && screen -dmS name2 program2 --flag
<roptat>have a look at my iproute2-service-type and iproute2-shepherd-service here:
<roptat>replace the name and commands run, and you get what you want :)
<Laalf>thanks. i will try
<roptat>I think you don't need screen
<Laalf>it might be helpful to run screen to have output when needed
<Laalf>this will be a fuckfest because id need to run sudo -H -u as well
<roptat>you can set the user as which the command is run in shepherd services
<roptat>oh maybe not actually
<civodul>janneke: sure, no worries
<civodul>janneke: is it build failures along the way?
<janneke>civodul: it's weird, i have no idea what's going on. gnu-make-boot0 did not build anymore; i had to add `LIBRARY_PATH=...' to the gcc-mesboot-wrapper to get it to build. that felt very fishy.
<janneke>now, gcc-cross-boot0 fails to build: undefined reference to `operator new(unsigned int)'
<civodul>janneke: oh, weird
<civodul>merging core-updates into core-updates-next shouldn't have caused this, no?
<janneke>so i'm a bit at loss now -- i may try to undo the core-updates merge and assert my %bootstrap-mes changes really works there; also, i may to try building on mbakke's gcc-wip7 branch
<civodul>i'd leave gcc-wip7 for later, to avoid having too many moving parts
<janneke>civodul: no, it shouldn't --so possibly my %bootstrap-mes patch series are problematic -- this is just very unexpected
<civodul>but yeah, maybe going back to the pre-merge state first could be helpful
<janneke>yeah...well mbakke made some nice fixes this gcc-wrapper area
<civodul>which reminds me i didn't comment on this series
<civodul>well, it doesn't have to be me :-)
<janneke>so, i have a few options but would have liked to push a new core-updates-next tonight, that may take some more days
<civodul>i understand the frustration
<civodul>sorry that this turned out to be trickier than i thought!
*rekado is falling seriously behind in reading email
<rekado>I’m currently preparing my PiGx/Guix talk for the conference in China.
<janneke>civodul: may be all my fault, but your kind words help :-)
<rekado>I’ve been reading most email that arrives at guix-devel, but have only rarely replied. If you find I missed something please feel free to ping me!
<civodul>rekado: sure, will do! :-)
<civodul>i usually explicitly mention you when i expect a reply
<civodul>i hope that works well with your filtering, scoring, and all ;-)
<rekado>generating my slides I notice that “gs” segfaults.
<rekado>this is the command that fails: gs -dBATCH -dNOPAUSE -q -sDEVICE=pdfwrite -sOutputFile=slides.pdf input*.pdf
<g_bor>hello guix!
<g_bor>rekado: I've seen you mentioned that cmake-build-system was removed.
<g_bor>When did that happen?
<rekado>g_bor: I didn’t express myself clearly
<rekado>what I meant is that *tensorflow* no longer comes with the necessary CMake files (since the latest release)
<g_bor>oh, I see.
<g_bor>Actually if I remember correctly it is using bazaar.
<rekado>but it comes with a contributed makefile and cmake files.
<rekado>the makefile only allows you to build the core library
<rekado>the cmake stuff was more comprehensive
<g_bor>bazel I mean...
<g_bor>yes, I got that.
<g_bor>how hard packaging bazel would be?
<g_bor>I guess it is a tough one...
<g_bor>we seem to have some parts of the dependencies...
<rekado>it’s very challenging.
<rekado>it depends on a large number of Java packages, and some of them are very difficult to package on their own.
<g_bor>Yes, impressive listing in third_party...
<g_bor>I have to go now, bye!
<rekado>me too, flight tomorrow
*rekado –> zzZ
<civodul>gs segfault is no good