IRC channel logs

2019-06-18.log

back to list of logs

<sebboh>xavierm02_: I warn you, I'm just making guesses in the dark, without even bothering to lookup the documentation for operating-system-host-name, but.. did you try (set-field my-os (operating-system-host-name "QWE")) ?
<xavierm02_>sebboh: Same error
<sebboh>SAME error? I didn't expect that. Maybe the error wasn't in the part I changed.
<xavierm02_>I think it is. If I replace this with my-os, it works.
<Marlin[m]>hi guix
<Marlin[m]>Gonna be travelling
<nullsix>thanks nckx is this syntax wrong, didnt pass
<bandali> https://www.openwall.com/lists/oss-security/2019/06/17/5
<sebboh>xavierm02_: I see. Well, I'm out of ideas. :(
<nckx>bandali: Remotely panic a box by sending it TCP packets? Am I reading that correctly? \o/
<sebboh>nckx: you may be interested in this. https://www.rodsbooks.com/efi-bootloaders/controlling-sb.html Apparently is *is* possible to replace the public keys that motherboards use for verification re: restricted boot (so called secure boot). For motherboards with onboard setup utilities that lack the functionality, there are external tools that can do it (all thanks to UEFI being standardized).
*nckx applies workarounds to their servers.
<bandali>hey nckx o/ yup looks like it
<nckx>sebboh: Thanks!
<alexanderbarbosa>weird, Info dont like people getting funny with fonts size... everything fall apart, if one insist :D
***webstran- is now known as webstrand
<fishinthecalcula>hey guix
<fishinthecalcula>i packaged the numix theme
<fishinthecalcula>it's my first time contributing to guix so i probably missed something
<fishinthecalcula>i run guix lint and indent-code.el on my new package definition and guix size says the package is 2 MB
<fishinthecalcula>am i supposed to check something else before submitting a patch?
<nckx>fishinthecalcula: ...and it works? Sounds good to go!
<fishinthecalcula>i'm not really sure how to apply it to my local machine but ls -R /gnu/store/...-numix... has the same contents as my numix installation from pacman
<fishinthecalcula>do you know how apply my package to my xfce installation?
<fishinthecalcula>nckx my main problem is that I run guix on Arch and my package is inside my local guix checkout so i'm not rellay sure how to load it inside xfce which comes from pacman
<nckx>Thanks for tagging me. I'm afraid I don't.
<nckx>fishinthecalcula: Try xfce4-appearance-settings in the xfce4-setting package. 🤷
<nckx>Else I haven't a clue.
<Marlin[m]>fatal error: linux/errno.h: No such file or directory
<Marlin[m]>hi guix
<Marlin[m]>i get the above error whenever i try compiling anything with chicken scheme
<fishinthecalcula>> fishinthecalculator: Try xfce4-appearance-settings in the xfce4-setting package. 🤷
<fishinthecalcula>will try, thank you very much
<Marlin[m]>just got two python packages working
<Marlin[m]> https://pypi.org/project/bpsrender/
<Marlin[m]> https://pypi.org/project/bpsproxy/
<Marlin[m]>part of the blender power sequencer addon
<xavierm02_>Does someone know a way of pretty-printing an operating-system object in a somewhat readable way?
<xavierm02_>Like replace the list of services by a list of their names etc.
<Marlin[m]> https://framagit.org/marlin1113/marlin-guix-packages
<Marlin[m]>this is my personal channel
<Marlin[m]>if you guys wanna help me getting these packages to work
<Marlin[m]>nckx looks like this is needed for lutris https://github.com/mvz/gir_ffi-gtk
<Marlin[m]>it says it requires go instrospection bindings for gnome
<Marlin[m]>also complains about the lack of typelib or something
<Marlin1113>waaat
<Marlin1113>dude
<Marlin1113>i just packaged typelib
<Marlin1113>and it still complains about the lack of typelib
<Marlin1113>;-;
<alexanderbarbosa>hey, ive installed libfixposix but a local installion cant find its share object
<alexanderbarbosa>how to fix it?
<alexanderbarbosa> Error opening shared object "libfixposix.so":
<alexanderbarbosa> libfixposix.so: cannot open shared object file: No such file or directory.
<alexanderbarbosa>
<alexanderbarbosa>same goes for sqlite...so
<tune>thanks to whoever packaged mako
<bandali>seconded
<alexanderbarbosa>hey, this line completely destroyed my guix system...
<alexanderbarbosa>it could find even "ls"
<alexanderbarbosa>export PATH="/home/alexanderbarbosa/.guix-profile/bin:/home/alexanderbarbosa/.guix-profile/sbin"
<alexanderbarbosa>guix package --search-paths -p ~/.guix-profile/
<bandali>that’s because you’re completely replacing PATH instead of prepending to it
<alexanderbarbosa>cant say why
<alexanderbarbosa>bandali: id
<bandali>id?
<alexanderbarbosa>bandali: indeed... that is what --search-paths gave me too export
<bandali>that’s *weird*
<bandali>i normally use something like this:
<bandali>export PATH="$HOME/.config/guix/current/bin${PATH:+:}$PATH"
<alexanderbarbosa>bandali: just did that... thanks anyway....
<bandali>i don’t recall ever completely replacing PATH
<bandali>alexanderbarbosa, cheers
<bandali>alexanderbarbosa, btw, if that’s indeed output that a guix subcommand gave you, i wonder if it makes sense to report a bug
<bandali>granted, i haven’t look at the docs for --search-paths so it may be the correct behaviour
<bandali>hmm, looking through the docs, it seems to be the expected behaviour
<bandali>did any tutorial or something suggest you to directly set PATH from that command’s output?
<alexanderbarbosa>OOohhh...
<alexanderbarbosa>--search-paths "forgot" to list export LD_LIBRARY_PATH="$GUIX_PROFILE/lib:$GUIX_PROFILE/lib64"
<alexanderbarbosa>it says only, gcc shared libraries ... export LIBRARY_PATH="$GUIX_PROFILE/lib:$GUIX_PROFILE/lib64"
<alexanderbarbosa>LD_
<alexanderbarbosa>or it is just export LD_LIBRARY_PATH="$GUIX_PROFILE/lib:$GUIX_PROFILE/lib64"
<bandali>not sure
<alexanderbarbosa>:/
<bandali>try asking tomorrow when others are around :)
<nckx>Oh, they left.
<nckx>sneek_: later tell alexanderbarbosa: That PATH is correct: ‘ls’ is provided by your system profile so it shouldn't be in there. Did you want ‘--search-paths=suffix’? LD_LIBRARY_PATH shouldn't be set unless you know what you're doing; Guix is right not to suggest that.
<sneek_>Got it.
<nckx>sneek_: botsnack if you do.
<sneek_>:)
<tune>04:50:51 < rekado> tune: I see in your paste that you haven’t added the recommended use-modules line.
<tune>it recommends two different things and they don't work so I undid them. sorry for the confusion
<tune>I'm also not clear which of the top 3 lines the things go in
<tune>I'll probably take another stab at it now after I do a guix pull
<tune>oh, can I have multiple use-modules lines?
<tune>I've just attempted that after adding to the existing one didn't work and it seems to be building stuff now. hopefully nothing breaks as a result
<tune>anything guix system users have to do to mitigate that recent SACK issue?
<tune>I heard someone mention a sysctl file but I don't have one
<nckx>tune: just echo 0 | sudo tee /proc/sys/net/ipv4/tcp_sack etc.
<nckx>tune: and (use-modules (srfi srfi-1)) definitely works.
<tune>Yeah, I was just adding the srfi bits to an existing use-modules line I had. I don't really know what I'm doing
<nckx>tune: Hm. That should also work provided it wasn't a Guix-specific use-package-modules or use-service-modules form. (use-modules (foo bar) (but foo too) (yoo hoo))
<tune>guix system: error: failed to load '/etc/config.scm':
<tune>ice-9/boot-9.scm:2803:6: In procedure resolve-interface:
<tune>no code for module (gnu system nss rnrs lists)
<tune>maybe this counts as guix-specific
<nckx>tune: You're asking for one module called ‘gnu system nss rnrs lists’. You have to tell Guile where the name ends and the next one (if any) begins, see my example above.
<nckx>(use-module (gnu system nss) (srfi srfi-1)).
<tune>ah, I see. I ended up not needing srfi. difficult to understand what was and was not working but sometimes it suggested srfi and sometimes it suggest rnrs and lists
<tune>in the end I added a new line like this: (use-modules (rnrs lists))
<tune>and it seemed to work
*nckx smells a rat.
<nckx>tune: Is this to get ‘remove’ to use with modify services?
<tune>yeah
<tune>someone suggested some lines for me to add to remove gdm from desktop-services so that it stops overriding sddm
<tune>gdm doesn't seem to work with wayland so it's just a nuisance at the moment
<nckx>I'm almost certain that you'll need the (srfi srfi-1) for that, not the (rnrs lists) one. Yes, there are two removes. Yes.
<nckx>s/for/one for/
<tune>hm well it built without error this time... I guess I could add and rebuild anyway
<tune>I'd hate to reboot and have something not working
<nckx>tune: Yes, it won't remove anything from the list. So you need to switch.
<tune>switch as in not use the rnrs and lists thing?
<nckx>Errors got nothin' to do wit' it.
<nckx>tune: Yep. And (srfi srfi-1) instead.
<tune>alright. thanks.
<nckx>tune: curl https://bayfront.guixsd.org/.well-known/logs/2019-05-30.log | grep -w rnrs -C2
<nckx>for slightly more on the difference ☺
<tune>I've been brought to an unresponaive tty with blinking cursor now
<tune>s/aive/sive
<nckx>‘Progress.’ :-/
<tune>can't ssh in to see my tmux session and check if the reconfigure is still doing stuff
<nckx>After reconfiguring? Sounds like X was killed.
<nckx>Hum.
<tune>using wayland, and I can't change TTYs
<nckx>(I don't know what you were trying to remove or why, I just happened to see the message about modules.)
<nckx>I can't help you with Wayland.
<tune>got something to happen hitting eject on my dock
<tune>the old thinf you had me change was working after a reconfigure so I've gotta say I am not feeling good about this
<tune>guess I will reboot it
<nckx>Well, yeah, because it did absolutely nothing.
<tune>you're likely right, pardon my ignorance
<nckx>(remove 'fizbotheclown system-services) can never break your box if there's nothing to remove...
<tune>good news, on fresh boot it has auto started sway which means sddm works and wasn't blocked by gnome
<tune>so thank you
<tune>s/gnome/gdm/
<nckx>Yay!
*nckx checks #linux-libre again...
<tune>oh yeah, on that note I notice I am on 5.1.10 kernel and I hear .11 fixes that SACK thing
<tune>should I do that echo proc sudo tee thing again since I've just rebooted?
<nckx>tune: Yep.
<nckx>tune: That's because a deblobbed 5.1.11 hasn't been released yet by the linux-libre project.
<tune>alright
<nckx>I don't know how they work (something something scripts) but it always takes quite a while.
<nckx>In Internet time, anyway.
<nckx>HOURS
<nckx>tune: We have a sysctl service, apparently, if you want to excercise your Scheme arm some more.
<tune>I'm not really sure what it is. Should I have it running?
<nckx>tune: It's not something that runs. It's just an automated way to echo those values on start-up.
<tune>Oh, got it.
<nckx>Similar to that sysctl.conf you mentioned above (that Guix doesn't have; this is our version). They both work the same way.
<nckx>echoaaS.
<tune>getting an issue with ssl certs now it seems
<tune> https://paste.debian.net/1088280/ here I'm trying to open a youtube video with mpv+youtube-dl
<tune>I think youtube was working before my changes and reboot so I doubt it's a ytdl issue
<tune>I also notice that pulsemixer is not showing up correctly, the bars showing volume just show up as black
<tune> https://github.com/ytdl-org/youtube-dl/issues/20597
<tune>> Nothing to do with youtube-dl. Most likely you are missing root certificate from yt-dl.org's security chain.
<tune>I wonder what I'm supposed to do here
<tune>I have nss-certs installs
<tune>looks like curl is broken too
<slyfox> https://www.gnu.org/software/guix/manual/en/html_node/X_002e509-Certificates.html maybe?
<tune>I already have nss-certs and I added those lines to my .profile... not sure what else I can do but it's still broken
<tune>also it worked very recently and I already had nss-certs before. I wonder if I broke it somehow
<tune>"nss" is in my use-modules line so maybe it's not working for some reason after my additions
<rekado_>sneek_: botsnack
<sneek_>:)
<rekado_>sneek_: later tell xavierm02_ You can use (operating-system (inherit my-other-config) (hostname "different") …) to override individual fields of an operating-system configuration.
<sneek_>Okay.
<rekado_>sneek_: later tell xavierm02_ Pretty printing can be done with the “pretty-print” procedure that’s part of the “(ice-9 pretty-print)” module.
<sneek_>Okay.
<rekado_>tune: you can have more than one use-modules line. The error indicates that you added just “srfi-1” to the list of service modules, which is wrong and not what the hint recommended.
<tune>that's not what happened, I was apparently combining it with the previous module
<tune>I didn't know I had to start a new set of parens, I thought it was just a big list
<tune>so I had something like (gnu system nss srfi srfi-1))
<tune>I got past that but now I have an ssl issue that is likely related to other changes I made
<rekado_>yes, that’s wrong because that would mean that there’s a module in a file gnu/system/nss/srfi/srfi-1.scm
<tune> http://sprunge.us/KICwiV is there anything here that is likely breaking my ssl certs?
<rekado_>no.
<rekado_>what error do you get?
<rekado_>tune: unrelated, but you probably shouldn’t add both glibc-locales and glibc-utf8-locales to the global list of packages. In fact, you don’t need them at all.
<civodul>Hello Guix!
<civodul>rekado_: python-ipython broke, something with documentation
<civodul>i looked at updating it but it looks like quite a bit of work
<civodul>any advice
<civodul>?
<tune>rekado_: well, youtube-dl and curl fail with https stuff for example
<tune>rekado_: are you saying I should remove both of them? I recall needing to add one or the other at least
<tune>rekado_: https://paste.debian.net/1088285/
<civodul>tune: did you check https://www.gnu.org/software/guix/manual/en/html_node/Application-Setup.html#X_002e509-Certificates-1 ?
<tune>I had nss-certs and things worked before, then after a reboot it all stopped
<tune>to be clear I still have nss-certs, but I'm not sure there's anything related to that for me to do
<civodul>could you check the env vars that the manual mentions?
<tune>I did add those to my user's .profile, I didn't have them yet
<tune>that didn't seem to fix it either
<tune>could it be breaking from removing gdm from %desktop-services?
<civodul>no no
<civodul>what does /etc/ssl/certs contain?
<tune>many many things
<tune>for some reason when I paste it to a site it gets all these garbage characters added
<tune> http://sprunge.us/0cyf9B
<civodul>ok, sounds good
<civodul>then maybe you need to unset SSL_CERT_FILE and SSL_CERT_DIR
<civodul>because programs will default to /etc/ssl/certs anyway
<civodul>well actually i have:
<civodul>SSL_CERT_FILE=/etc/ssl/certs/ca-certificates.crt
<civodul>SSL_CERT_DIR=/etc/ssl/certs
<civodul>and everything works fine
<tune>I'll try but I only added them after the problem started. although I did tweak some config.scm things back and forth a bit
<tune>oh interesting
<tune>mine has me looking in home
<tune>export SSL_CERT_DIR="$HOME/.guix-profile/etc/ssl/certs"
<tune>export SSL_CERT_FILE="$HOME/.guix-profile/etc/ssl/certs/ca-certificates.crt"
<civodul>on Guix System it's enough to add 'nss-certs' to the 'packages' field
<civodul>with these settings, you'd have to install nss-certs in your own profile
<civodul>which is unnecessary here
<tune>ah okay
*civodul goes afk for a bit
<tune>I just exported them as you had them and got a youtube video to open now from that shell, but if I launch it with rofi it doesn't work. maybe I have to login again
<tune>okay I think it's fixed now
<tune>thanks
<rekado_>ipython fails to build; I suspect it’s related to the sphinx upgrade on staging
*rekado_ sees that civodul mentioned it a little earlier
<rekado_>civodul: I’m working on the ipython upgrade now
<rekado_>I also split python2-ipython (version 5.8.0) from python-ipython (version 7.5.0)
<rekado_>I’m about to write the parasail authors to clarify their license requirements, but I wonder if the citation requirement really makes it non-free. It seems to me that it’s not any different from the old BSD license.
<rekado_>i.e. it’s annoying and incompatible with the GPL, but it’s still free.
<rekado_>see https://www.gnu.org/licenses/bsd.html
<rekado_>bleh, latest ipython documentation requires ipykernel, which requires ipython.
<rekado_>I suggest we break out the documentation to a separate package.
<Digit>i see guix 0.16.0 in gentoo. is 0.16.0 far behind the 1.0.1 download linked on gnu.org/software/guix/ ? would it be trivial to upgrade the guix-from-portage via guix once installed?
<linuxserverguixk>I install guix on archlinux, using the binary, but guix install hello is display Warnning: setlocale LC_ALL: cannnot change locale (en_US.utf8)
<Digit>ah, i gather from gnu.org/software/guix/blog/ 0.16.0 is just 2 versions behind. that answers my first question at least.
<linuxserverguixk>how to fix it
***root is now known as Guest8056
<civodul>rekado_: neat!
<civodul>yeah i had just noticed the loop with ipykernel
<civodul>BTW i have backcall here, but perhaps you've already packaged it too?
<rekado_>yes, I got it already
<civodul>cool
<rekado_>almost done with the Python 3 side of things (including docs).
<civodul>Digit: yeah; normally 'guix pull' will work to upgrade to current master
<rekado_>the Python 2 versions still don’t build properly
<civodul>ok
<civodul>well thanks a lot!
<rekado_>we should try harder to avoid breakage like this. We can’t just have the scientific python stack and ipython break because documentation fails to build.
<Digit>thanks ludovic. :)
<rekado_>ipython is now fixed. There’s now a new python-ipython-documentation package providing the docs.
<civodul>yeah!
<civodul>thank you, rekado_!
<civodul>rekado_: i think i'll email the list to announce that {mirror.,}hydra.gnu.org service is discontinued, WDYT?
<efraim>I noticed the CI system on hydra wasn't getting updates for weeks now
<civodul>efraim: yes, it's officially discontinued now
<civodul>Mark was the only person still taking care of it
<efraim>bayfront and berlin share a signing key?
<civodul>yes, for historical reasons
<civodul>we should probably change that
<rekado_>civodul: sounds good.
<efraim>Well I removed hydra from my OS configs and from my vm-image config
<civodul>efraim: i've changed / and /index.html to return 301
<civodul>but the rest is unchanged
<efraim>curl hydra.gnu.org
<efraim>bash: curl: command not found
<efraim>I guess that shows how often I use curl :)
<civodul>wget FTW!
<efraim>we all know the primary use of curl is 'curl | sudo bash'
<civodul>definitely
<rekado_>oh, there is no tensorflow substitute because gemmlowp uses sse4.1, which isn’t supported by all build nodes.
<rekado_>should it be SSE2 at most?
<bavier`>rekado_: yes
<minall>Hello guix!
<minall>I want to make a question...
<bandali>hey, go for it
<minall>Thanks!
<minall>I've just installed Icecat, and i had no fonts, i'm on a clean guixSD installation
<minall>All I had to do was install throught 'guix install' the fonts font-dejavu and font-liberation, then restarted icecat and solved
<minall>I just think that i shouldn't had this problem with a clean installation, i don't know... maybe add these fonts to the icecat package? or maybe the problem was my machine?
<bandali>i’m not on a Guix System installation (what you refer to as GuixSD, the old name) right now, but i believe that’s expected
<bandali>the choice of font is left to the user
<bandali>it may be a good idea at some point to make installing *some font* mandatory for things like icecat though
<minall>Ohhh i see
<bandali>maybe open a bug report about it?
<minall>Yep, I think is perfect to left the choise to the user... but, maybe a messages installating icecat, to inform the user that He has to install the fonts, just to help more unexperienced users?
<minall>bandali: I don't think it will be necessary, since is not a bug, I think is just a suggestion
<bandali>minall, yes, but suggestions are more than welcome :) and necessary for improving software
<minall>I had to search in the internet to have it solved, but a message that says: you should install fonts, or some default options would be nice
<bandali>if not as a bug, feel free to post to the guix-devel list
<minall>bandali: I love to help, since I appreciate what you're doing, guix is awesome! (not the awesomewm)
<minall>bandali: How do i post it in the guix-devel list? I'm very new in this jaja
<bandali>minall, ;)
<bandali>to post to the list, you just send an email to guix-devel@gnu.org
<bandali>though i recommend subscribing to the list if you’re interested
<bandali>emails from subscribers generally go through faster/easier
<bandali>you can see info about the list and how to subscribe to it here: https://lists.gnu.org/mailman/listinfo/guix-devel
<bandali>also https://www.gnu.org/software/guix/contact/ is useful too
<minall>bandali: Thank you!, I'll do it, and continue using GuixSD for any problems that i can help
<minall>Thanks! i'll go to the page directly
<bandali>minall, cheers!
<minall>bye bye !
<bandali>bye :)
<kmicu>It’s a bug ;)
<civodul>i guess we could add a couple of fonts to %desktop-services
<cbaines>I've encountered ungoogled-chromium refusing to start without fonts
<bandali>may i suggest Liberation Sans?
<cbaines>I had to install fontconfig and font-dejavu to get it to work
<kmicu>It can be any font which makes an installed by Guix program usable. Later a user can change preferences in usual places (Gnome’s dconf, fontconfig, gtkrc, etc).
<kmicu>(Treating global mutable things like fonts or opengl in systems like Nix/Guix is an open issue. Above my pay grade. 😺)
<joshuaBPMan>hello, i'm having trouble starting the latest version of sway.
<joshuaBPMan>I'm getting this error:
<joshuaBPMan>sway: ../mesa-19.0.4/src/compiler/glsl/serialize.cpp:555: void read_uniforms(blob_reader*, gl_shader_program*): Assertion `vec_size + prog->data->UniformStorage[i].storage <= data + prog->data->NumUniformDataSlots' failed.
<joshuaBPMan>would deleted thi mesa shader cache help?
<joshuaBPMan>rm -r .cache/mesa_shader_cache/ ?
<joshuaBPMan>I think I did something like that last time and it seemed to help.
<joshuaBPMan>actually it fixed the problem.
<joshuaBPMan>yup. that worked. thanks. I'm not certain why this is a problem in guix land.
<vagrantc>does the cache store paths and upgrading invalidates them somehow?
<vagrantc>joshuaBPMan: does the cache store paths and upgrading invalidates them somehow?
<vagrantc>although one would think it would just clean out invalid cache items, or at least ignore them...
<kmicu>iirc there’s an issue on the mailing list about GDM and ~/.cache/mesa_shader_cache and how to handle that.
<kmicu>(roptat making ~ read only has an another solution.)
<rekado_>cache invalidation really is an upstream problem, but it pops up more commonly on Guix because of the unique package prefixes.
<demotri>I read that hydra was taken offline. I find that said: Even though it was slow, the user interface was much more powerful than that of Cuirass. What was the reason to take it down?
<rekado_>demotri: let’s improve the user interface of Cuirass then.
<kmicu>All those lovely CI charts are gone? Well, time to improve Cuirass!
<rekado_>demotri: the hydra software is no longer maintained. Mark also stopped maintaining the service, which was a huge manual task.
<rekado_>(i.e. maintenance of hydra.gnu.org, which would not be usable without continued manual intervention, database manipulations, targeted GC, etc)
<demotri>rekado_: You are putting pressure on this by creating facts :-)
<rekado_>it’s more like redirecting existing pressure
<kmicu>(Guix’s Hydra*)
<vagrantc>i haven't tried recently, but it seemed like berlin wasn't building (many) armhf substitutes
<rekado_>kmicu: (yes)
<rekado_>vagrantc: there were problems due to the fact that we had only a limited number of actual armhf build nodes.
<rekado_>vagrantc: now that they aren’t used by hydra.gnu.org they could be added as build nodes to berlin.
<vagrantc>rekado_: ah, cool
<kmicu>demotri: UI is usable but the source code is in C and Perl so we shall not miss it. 😺
<rekado_>demotri: FWIW I had no hand in discontinuing hydra.gnu.org, but I fully support this long overdue action.
<demotri>rekado_: OK, I understand. It already looked like it was much work to maintain the old Hydra, and it was a matter of time when it finally went down.
<vagrantc>rekado_: by "could be" i'm guessing that requires someone to do some work to make it happen?
<rekado_>Cuirass recently gained a package search feature; if you have a feature that you’d really like to see in Cuirass I’d be happy if you could outline it on the mailing list and help create it.
<rekado_>vagrantc: I don’t want to unilaterally declare what to do with the nodes; we could also add them to bayfront. This is a decision that the Guix sysadmins should make together.
<demotri>I wanted to look more into Cuirass but I don't have the time, I also have some half-baked packages around that I would like to get out at some time.
<rekado_>vagrantc: it’s not a lot of work. It only really requires keys to be exchanged and configuration files to be updated (+ reconfigure, + firewall changes by the MDC’s IT department)
*vagrantc nods
<sirgazil>Today I re/quit
<vagrantc>i guess i could ask these questions on the mailing list thread, too
<rekado_>demotri: oh, don’t get me wrong: I’m not pushing you to hack on Cuirass.
<sirgazil>Sorry, I'm having problems here...
<demotri>rekado_: Now that pressure is there, I will at least have a closer look at the UI and write down what I'm missing.
<rekado_>demotri: I’m just saying that we don’t have a clear view on what Cuirass is missing.
<kmicu>Charts!
<rekado_>demotri: I added what I missed most (package search), but I’m sure other people have other needs and it would help to have them articulated so that people can work on them.
<rekado_>kmicu: I don’t think I’ve ever seen a chart on hydra.gnu.org
<kmicu>Friendly reminder that https://gitlab.com/alezost-emacs/build-farm is a thing.
<demotri>rekado_: Don't get ME wrong :-) I feel like a long time that I want to play around more with Cuirass, and a bit of pressure doesn't hurt :-)
<rekado_>demotri: heh :)
<kmicu>There were lovely charts to monitor closure sizes and build times https://hydra.nixos.org/job/nixpkgs/trunk/icecat-unwrapped.x86_64-linux#tabs-charts
<efraim>Time to remove hydra from guix? When was the last time it built?
<rekado_>kmicu: was this even available in the version deployed on hydra.gnu.org?
<rekado_>efraim: I’d be happy to see it go.
<bavier>efraim: fine by me
<kmicu>rekado_: yep, linked those charts from hydra.gnu.org here too! ;)
<kmicu>Nope.
<kmicu>Checked logs. I was linking to NixOS’ Hydra charts all the time 🤦
<kmicu>(Or maybe they were not called tabs-charts but tabs-history…)
*efraim will get around to it in a few hours
<civodul>efraim: fine with me too
<civodul>we have Cuirass + build-farm.el + guix weather -c
<civodul>life is good :-)
<civodul>it can be better, but let's work from there
*kmicu experiences a strong Mandela Effect about graphs on hydra.gnu.org right now.
<civodul>"Mandela effect"?
<sirgazil>Hi, I restarted the computer today a
<kmicu>civodul: I’m pretty sure we had graphs (but not closure size) on e.g. https://ci.guix.gnu.org/#tabs-history
<kmicu>Great, stale clipboard cache 🤔
<civodul>heh :-)
<kmicu> https://hydra.gnu.org/build/3473960#tabs-history
<civodul>yes, Hydra does graphs like you wrote
<civodul>i'm sure the Guid Data Service will have that and more Real Soon
*civodul looks at cbaines
<civodul>*Guix Data Service
<kmicu>Thank you. I thought I’m the case of https://en.wikipedia.org/wiki/Mandela_Effect aka false memory about existance of graphs on hydra.gnu.org.
<civodul>heh
<sirgazil>Sorry, I'm having serious problems with my system...
<sirgazil>Can't log in to GNOME but can log in to TTY
<sirgazil>Some errors are preventing me from using TTY correctly thouth
<sirgazil>When I try to log in to GNOME desktop, I'm taken back to log in screen.
<sirgazil>In TTY I can log in using the same password, but there is no guix command to try to roll back to previous generation.
<rekado_>sirgazil: I feel your pain. This is likely because of stale cache files.
<sirgazil>I don't know what to do...
<rekado_>sirgazil: remove (or move away) ~/.cache/gnome-shell
<rekado_>(do this after logging in on a different TTY)
<rekado_>when even GDM doesn’t appear you may need to remove /var/lib/gdm
<rekado_>this is really bad and I hope we can find a way around this
<rekado_>civodul: I updated the python-on-guile package, so you can try to use its bin/python on glibc 2.29.
<sirgazil>rekado_: I can't find any gnome-shell in ~/.cache
<rekado_>civodul: I noticed a few minor problems with the bin/python (not all packages are added to the load path and the compiled load path setting is wrong)
<rekado_>sirgazil: oh, try ~/.local/gnome-cache
<rekado_>there are probably one or two files there (something about notifications and applications).
<sirgazil>rekado_: I found and deleted "~/.local/share/gnome-shell"
<sirgazil>I'll try to log in again.
<notnotdan>jackhill:
<notnotdan>oops sorry
<notnotdan>jackhill: sorry, wrong hilight :<
<jackhill>no worries ☺
<notnotdan>what is the etiquette for bugging someone about reviewing patches?
<rekado_>notnotdan: can you tell us the debbugs number?
<civodul>notnotdan: i'd just send a friendly ping by email, no worries
<civodul>as long as you don't do it every single day ;-)
<jsgrant_>Am I crazy, or did Guix (Distro) used to support LVM? Seem to remember installing via it like 2-3 years ago when I was first messing around
<notnotdan> this is the debbug https://debbugs.gnu.org/cgi/bugreport.cgi?bug=36138
<notnotdan>I think the last person to submit a related patch (for chocolate-doom) was Jakob, so I wanted to bother him, but he is not a reviewer IIRC
<bandali>afaik guix never supported lvm
<jsgrant_>bandali: Hm, odd; Guess may just be flawed memory after 3 or-so really eventful years ... is/was this on the roadmap maybe?
<rekado_>jsgrant_: it is still a desirable feature to have, but I don’t know of anyone actively working on it.
<bandali>jsgrant_, :) there were maybe one or two attempted patches, but i don’t think they went anywhere
<bandali>and yup, what rekado_ said
<jsgrant_>Saw lvm2 was packaged generally, guess that is for 3rd-party (non-system) drives then?
<vagrantc>i've definitely used lvm on guix, but not for the rootfs or anything that mounts at boot, just to access other drives
<sirgazil>No luck, still can't log in...
*vagrantc replaced gdm with sddm...
<sirgazil>Another problem is that the "guix" command is not found any more.
<vagrantc>since i wasn't using gnome anyways, that dramatically reduced the install size...
<vagrantc>guix disappearing sounds pretty challenging
<sirgazil>So, if the problems I'm having are because of the latest pull and upgrade I did on June , I don't see how I can roll back.
<sirgazil>vagrantc: Yeah, I don't know what to do. I'm seriously considering going back to Debian.
<sirgazil>*June 15
*sirgazil goes for lunch
<vagrantc>sirgazil: are there /var/guix/profiles/per-user/*/current-guix* links?
<vagrantc>you could call guix with the full path ...
<vagrantc>e.g. /var/guix/profiles/per-user/USERNAME/current-guix-N-link/bin/guix and see if one of those works
*vagrantc has also been working on not having to choose between Debian and Guix, but that's another issue entirely...
<sirgazil>vagrantc: Yes, there are many links and "current-guix" works
<bavier>cbaines: I'm curious to hear what you have to say about perl programs in Guix at the guile-days
*bavier will not be there
<sirgazil>Oh, no, spoke too soon, let me see...
<kmicu>Could you choose an older/working entry from boot menu?
<sirgazil>kmicu: I'll try that right now...
<sirgazil>No luck.
<sirgazil>Using a menu entry from May, I got the same log in problem.
<sirgazil>And no "guix" command found...
<vagrantc>what's PATH set to?
<sirgazil>/home/sirgazil/.guix-profile/bin:/home/sirgazil/.guix-profile/sbin
<sirgazil>I set PATH in my ".bash_profile" using the output of "guix package --search-paths"
<vagrantc>that seems to be missing quite a bit
<vagrantc>mine is /run/setuid-programs:/home/vagrant/.config/guix/current/bin:/home/vagrant/.guix-profile/bin:/home/vagrant/.guix-profile/sbin:/run/current-system/profile/bin:/run/current-system/profile/sbin
<nckx>sirgazil: That's not what --search-paths is for.
<nckx>What happens when you remove the lines you added like that?
<vagrantc>you shouldn't need to manually set PATH if you're running Guix System
<sirgazil>nckx: I set the paths that guix told me I should set when I install some packages.
<nckx>sirgazil: Guix didn't tell you to add them to any file. Anyway, remove them now.
<sirgazil>guix displays a message at the end that says you should set some variables.
<nckx>If you set PATH manually you're on your own.
<nckx>sirgazil: Yes.
<nckx>Not that you should edit your .profile to add them.
<nckx>Anyway, that message is gone now ☺
<sirgazil>nckx: /quit
<kmicu>sirgazil: unfortunately multiple Guix System issues hit you at once but recovery should be as easy as reverting any PATH related changes from files in (not tracked by Guix hence no rollback) home.
<kmicu>(^^ and that’s why what’s not tracked by Guix sucks. Like whole ~ config which can render the Guix System rollbacks useless.)
<str1ngs>it's hardly practical to have guix track ~/
<kmicu>It’s very practical and that already exists in Nix ;)
<str1ngs>for which use case? I have configs that run on multiple OS's some that are not supported by Nix or Guix even.
<kmicu>(Guix/Nix not following FSH with ready-only store was also hardly parctical.)
<kmicu>The use-case is rollbacks and reproducibility. Assimilating ~ is just pushing Guix/Nix further. Nothing surprising.
*kmicu is excited for roptat’s read-only ~ for that very reason.
<sirgazil>nckx, kmicu, vagrantc: Thanks. I could log in again.
<sirgazil>So, setting those variables in my ".bash_profile" was a problem.
<sirgazil>After installing packages, everytime I saw the guix message recommending setting some environment variable, I would copy and paste the value to my ".bash_profile".
<sirgazil>The message has changed, though. It used to display a list of environment variables, but now it recommends some commands.
<sirgazil>So, in the GNU system installed with Guix I should never set those variables? I'm confused about those messages then.
<sirgazil>I remember seeing those messages when I installed Lyx, ibus, python, and many other software.
<sirgazil>Now I have to check if things that didn't work as I expected were not working because of that.
<sirgazil>It seems I'm not ready for this system. I've been misinterpreting instructions...
<vagrantc>sirgazil: usually simply logging out and logging back in again would set them appropriately
<vagrantc>sirgazil: your experience is probably exactly why the message now says something different
<sirgazil>Or the system is not ready for my level :)
<xavierm02_>If I have two packages with the same name in two channels, how do I know which one I'm referencing in the config.scm file?
<sneek_>Welcome back xavierm02_, you have 2 messages.
<sneek_>xavierm02_, rekado_ says: You can use (operating-system (inherit my-other-config) (hostname "different") …) to override individual fields of an operating-system configuration.
<sneek_>xavierm02_, rekado_ says: Pretty printing can be done with the “pretty-print” procedure that’s part of the “(ice-9 pretty-print)” module.
<Marlin[m]>try /run/current-system/profile/bin/guix
<guix-guest4>how can one use a grub-bootloader in a system configuration with an "empty" installer, that is, to override the installer field of the bootloader record with an empty procedure?
<bricewge>Following this tutorial http://guix.info/blog/2018/a-packaging-tutorial-for-guix/, I'm trying to build with "direct checkout hacking" a package I patched. But executing `./pre-inst-env guix build exim` keep producing the old output.
<bricewge>W\hat am I missing here?
<ryanprior>I want to get started building a basic python package using guix. Does anybody use python+guix who can point me to a basic introductory example and describe what your workflow is like? I've been searching but haven't had much luck yet, so I want to write-up my experience once I get my footing.
<cbaines>cbaines, regarding graphs, that's definately something I'd like to add at some point.
<cbaines>or rather civodul (but they're absent)
<cbaines>I'd also really like to continue the work I started on tracking build servers (Cuirass instances) https://prototype-guix-data-service.cbaines.net/builds
<cbaines>bavier, I'll put the slides/notes for the guile-days presentation up somewhere once I've written them :D
<enderby>latest mpv break for anyone? getting "mpv: ../mesa-19.0.4/src/compiler/glsl/serialize.cpp:555: void read_uniforms(blob_reader*, gl_shader_program*): Assertion `vec_size + prog->data->UniformStorage[i].storage <= data + prog->data->NumUniformDataSlots' failed."
<enderby>using guix on debian
<kmicu>Great work cbaines!
<bavier>enderby: it's a caching failure; earlier today someone said doing a `rm -r .cache/mesa_shader_cache/` fixes the issue
<enderby>bavier: awesome, works thanks!