IRC channel logs

2020-07-31.log

back to list of logs

<guix_user>Is there any way to change the store location to something other than /gnu?
<NieDzejkob>guix_user: yes, but you won't be able to use the project's substitute servers
<NieDzejkob>you'll have to compile everything yourself
<apteryx>sorry about the previous IRC connect/disconnection noise!
<apteryx>I've got the situation under control now, and improved my setup to reduce the chance of it occuring again.
<apteryx>using a bit of Guile scripting with WeeChat :-) https://paste.debian.net/1158377/
<lfam>nckx: I did consider it but decided against it. It's my residential ISP and the uplink is slow. I run the mirror because I have 3 Guix installations on my home network
<lfam>Almost all the traffic from the mirror stays local
<MtotheM>my biggest dislike with guix is that it's not very usable offline. which is the reality for me most of the time, unlike other distros where you can just clone the repo. that's not really possible here. since it pulls directly from source code on git repos.
<wdkrnls>Is there a guix command I'm missing to search all distinct historical versions of a package?
<lfam>No wdkrnls
<NieDzejkob>git log
<NieDzejkob>;)
<wdkrnls>That's what I was thinking :) e.g. git log -S texlive | grep texlive | grep 'pdate to'
<PotentialUser-84>Hello, back again about the UEFI install, busy making sure I have correctly reproduced the problem, should be another 30 mins and I'll be back
<nefix>compiling is slow :(
<VigilanceTech>Hi everyone. Due to being burned out in the California wildfires I have very limited internet/electricity. I've tried a couple times to install (and burned up all my tethering data) and it's never been able to complete. The last time I had to leave my laptop running overnight while I slept and when I got up its battery was dead and again all it does is give me a black screen with blinking cursor when I try to boot. The battery was pretty full wh
<PotentialUser-84>so, i've attempted the UEFI boot install again, and have failed to get a boot. I made sure to write out all the commands I used in the exact order I used them in the manual installer, about 25 commands in total. I can list them out all at once or whoever is available can just ask what they need like parted commands, fs commands, etc
<NieDzejkob>PotentialUser-84: you could put them on paste.debian.net, that way you don't spam the channel but we still get to think through all the info
<PotentialUser-84>sure thing, just posted as potuser84
<guix_user>I'm trying to setup my bananapi m2u to run the guix system. I feel like I have more control over guix than armbian.
<guix_user>I erased the onboard mmc, and have been having trouble getting anything back on there :(
*raghavgururajan added *Stemming* support to *Tracker*. File search should be even better in *GNOME*.
<VigilanceTech>Oh I just found the manual installation section in the manual. NVM
***catonano_ is now known as catonano
<PotentialUser-84>Updated it, as there was a typo http://paste.debian.net/1158389/
<lfam>The speedup from the Intel i5 in my x230 to the Intel i5 in my x260 is remarkable. It only takes ~5 seconds to build python-xyz.scm now
<lfam>That could have taken a minute previously
<lfam>Okay, it takes 10 seconds
<lfam>It's really a lot faster
***terpri__ is now known as terpri
<PotentialUser-84>just a side note, I do actually get a successful install message from guix system, and that the target /boot/efi has been successfully installed, just no ability to boot
<vivasvat>nckx, I did indeed see your message about europe vs american speeds
<vivasvat>like I guess it is what it is
<vivasvat>does anyone know if there are plans to eventually expand infrastructure/mirrors to other loctions?
<terpri>dunno, but hcoop.net might be able to host a US-based mirror depending on bandwidth requirements
<terpri>might be possible to use bittorrent (i'm sure it's been discussed) or ipfs for distributed mirroring?
<vivasvat>thats an interesting proposition
<vivasvat>but wouldn't it be hard to implement
<vivasvat>oh wait nvm I misunderstood
<vivasvat>wait nvm nvm
***sm2n_ is now known as sm2n
<bavier[m]1>there's been some actual work on ipfs and gnunet integration
<bavier[m]1>not bittorrent, afaik
<vivasvat>oh cool
<vivasvat>but I bet lack of peers would be a problem
<bavier[m]1>I don't think it would be a problem ;)
<vivasvat>idk ipfs, but if its anything like bittorrent, wouldn't it?
<vivasvat>like if you need to install a package no one is seeding
<vivasvat>or something like that
<bavier[m]1>for a short time, probably, but people pick up packages pretty quickly
<hendursaga>I'd more than likely 'seed' once it's out.
<hendursaga>Question: seeing how much disk writes updating Guix is taking, I'm almost wondering if I should split my partition into two: one encrypted, for my data, and then an unencrypted one for all Guix-related stuff. Anyone done this yet?
<vivasvat>can someone explain the difference between guix pull
<vivasvat>and guix upgrade
<Noclip>I've run "guix upgrade" two times within a few minutes and it made two times the same update (according to the version numbers) but wrote them to different files in the store. I didn't run "guix pull" between the two upgrades:
*Noclip sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/hQWGQQYflqkHdwjarDMovBbi >
*Noclip sent a long message: < https://matrix.org/_matrix/media/r0/download/matrix.org/KdoSsGihoMMdVmRnDDCXgxfq >
<Noclip>Running guix upgrade more than those two times didn't anymore do something (that's how it should be). Why did it perform the same upgrade two times?
<Noclip>vivasvat: Do you know APT from debian or ubuntu?
<vivasvat>I come from pacman
<vivasvat>but yeah I know the basics of APT
<Noclip>ok
<vivasvat>guix pull -> apt update
<vivasvat>guix upgrade -> guix upgrade
<Noclip>jup
<vivasvat>*apt upgrade
<vivasvat>oh ok
<Noclip>There is a little difference:
<vivasvat>so essentially guix pull changes the local copy of package versions or whatever
<vivasvat>like it syncs your system with most up to date packages
<vivasvat>and upgrade actually does the upgrade
<Noclip>guix pull also updates guix (the package manager) itself and not just the package information.
<vivasvat>ohhh ok
<vivasvat>thats cool
<vivasvat>makes sense
<vivasvat>thanks for the help!!
<Noclip>As I'm not using guix system I don't know wether guix pull also upgrades the whole system or just the package manager.
<pkill9>it doesn't upgrade the system on guix system
<pkill9>you have to run `guix system reocnfigure` to upgrade the system
<Noclip>will reconfigure also upgrade the kernel?
<vivasvat>yeah I'm also not running guixSD rn
<vivasvat>im using guix on top of pacman
<vivasvat>tho I can see the power of doing system management in guile
<vivasvat>when doing guix install
<vivasvat>how do I know when the package will be installed by source
<vivasvat>and when the package will be installed via substitues
<apteryx>there's no clear telling, but 'guix weather' can gaging the coverage
<apteryx>it depends on whether the build farm has already managed to build the package or not; when not, it falls back to building on your machine (or an offload machine, if you set up one).
<kelsoo1>Distrotube "My Thoughts On GNU Guix After Three Days" https://invidio.us/watch?v=MV8SOJ2hlWQ
<mroh>kelsoo1: interesting, ty for sharing!
<kelsoo1>welcome
<Noclip>This time it's also synced over to LBRY.
<mroh>he has fun -> mission accomplished ^^
***apteryx_ is now known as apteryx
<nefix>morning!
<nefix>right now, my telegram definition looks like this: http://dpaste.com/64RLWPSAH, but it's failing to some CPP error. I'm using https://github.com/NixOS/nixpkgs/blob/master/pkgs/applications/networking/instant-messengers/telegram/tdesktop/default.nix and https://git.archlinux.org/svntogit/community.git/tree/trunk/PKGBUILD?h=packages/telegram-desktop as base, but I can't see what's the thing that makes
<nefix>it break :/
<nefix>the error is the following: https://dpaste.com/2BKDH4MYD
<mroh>nefix: I can reproduce this, qtimageformats doesnt build on master, with the error you pasted.
<nefix>:/
<nefix>at least it's good to know that I'm doing progress :)
<nefix>mroh: so what would you suggest? Should I make an override to continue?
<mroh>maybe you can try building your pkg w/o qtimageformats?
<mroh>here is a quick fix: http://paste.debian.net/1158403 for qtimageformats
<mroh>nefix: just posted a patch, maybe its pushed soonish.
<nefix>mroh: thanks! I'll try again later
<nefix>see ya!
***sputny1 is now known as sputny
<NieDzejkob>mroh: huh, this is weird. Any idea why this patch is necessary? (i.e. why the package got shipped without the include in the first place)
<mroh>NieDzejkob: no, idk. maybe some versions of qt had a <cmath> include somewhere, I guess, because qview (our only consumer) did build at some point, but no clue
<NieDzejkob>mroh: qtimageformats is a part of qt, though, and the versions are matching
***dddddd_ is now known as dddddd
<NieDzejkob>uh oh, looks like there's going to be a Rust 1.45.2 on Monday
<mfg>#Building-from-Git in the guix manual has a error, the git log command needs double dashes before the filename
<mfg>at least git complains here about it
<mfg>and it works when they are added
***jonsger1 is now known as jonsger
<NieDzejkob>mfg: you might be looking at an out-of-date manual, the section doesn't use any git log command in `info guix' (which is updated on pulling)
<Formbi>hi
<Formbi>I managed to package Node 14.6, but the thing I have is kinda ugly: https://0x0.st/ix4o.txt
<Formbi>could you help me clean this up a bit?
<NieDzejkob>Formbi: at this point it's probably cleaner to integrate the configure-flags into the phase itself, it could simplify the code a lot
<Formbi>how to do that?
<Formbi>I don't have almost any knowledge of those mechanisms
<dannym>Formbi: Hmm, the package recipe looks pretty good to me
<Formbi>I heard some deeper things can be inherited
<Formbi>so if it would be possible, the recipe would be less repetitive
<dannym>Formbi: Oooooh
<dannym>Formbi: Yeah, if configure and path-npm-shebang are unchanged, can definitely inherit those
<dannym>Formbi: The syntax to do that is "interesting", though
<dannym>Formbi: Example I took from gnu/packages/gnome.scm :
<dannym>Formbi: (arguments
<dannym> (substitute-keyword-arguments (package-arguments libsoup)
<dannym> ((#:configure-flags configure-flags)
<dannym> `(delete "-Dgtk_doc=true" ,configure-flags))
<dannym> ((#:phases phases)
<dannym> `(modify-phases ,phases
<dannym> (delete 'move-doc)))))
<dannym>In your case, "node" instead of "libsoup". Then put "replace 'patch-files [... your stuff ...]" instead of "delete 'move-doc"
<Formbi><3
<NieDzejkob>Formbi: oh, I didn't realize you copied most of that from the previous versions
<Formbi>ok, but I'm now stuck on phases :(
<Formbi>how to remove the «(add-before 'configure 'patch-files …» in this?
<Formbi>and how to replace «#:test-target "test-ci-js"» with «#:tests #f»?
<NieDzejkob>what have you got now?
<Formbi>hmm?
<NieDzejkob>like, what changes have you already made?
<Formbi> https://0x0.st/ix4o.txt
<NieDzejkob>have you seen danny's example of substitute-keyword-arguments?
<Formbi>ah, I see
<Formbi>or not…
<Formbi>I put the configure flags
<Formbi>but I don't understand how to remove the add-before
<dannym>Formbi: What it's doing is just give you the inherited phases in ",phases", not the instructions how to build them
<dannym>Formbi: So you have a list of phases which there is a phase 'patch-files
<dannym>Formbi: inherited from the super-record
<dannym>Formbi: So if you want to replace it, just (replace 'patch-files ...)
<dannym>Formbi: (if not, (delete 'patch-files)
<dannym>)
<Formbi>ah, so the thing is just named patch-files
<Formbi>ok, so what about the tests?
<dannym>Formbi: Try ((#:tests? _ #f) #f) ; note: TWO underscores
<dannym>whoops
<dannym>Formbi: Try ((#:tests? _ #f) #f) ; note: TWO parameters after #:tests?
<Formbi>meybe there is some misunderstanding
<Formbi>originally there was «#:test-target "test-ci-js"», I removed it and added «#:tests #f»
<Formbi>does the thing you said do that?
<dannym>Formbi: Yes, so in order to add #:tests #f, you have to use substitute-keyword-arguments, add in there ((#:tests? _ #f) #f)
<Formbi>ok
<dannym>Formbi: #:test-target you can't remove, but just replace it by the correct test target
<dannym>Formbi: Or leave it be--it won't be used when #:tests? #f anyway
<Formbi>k
<raghavgururajan>Hello Guix!
<mroh>hello raghavgururajan!
<Formbi>hello
<str1ngs>hello, this command does seems to build way more then just nomad which is unexpected. guix build nomad-git --with-commit=nomad-git=518ffe5faf6aa95520594d5aeb3aa2f0a4f9c92b . for example one if it's inputs is webkitgtk which I know substitutes exists for but it still wants to rebuild webkitgtk. is my understanding of --with-commit wrong?
<mfg>NieDzejkob: you're right, just checked the info manual :)
<Formbi> https://0x0.st/ixJW.txt what the flying f?
<Formbi> https://0x0.st/ixJy.txt this is the recipe
<bavier[m]1>Formbi: there's some quoting missing
<Formbi>hmm
<Formbi>should I double-quote the list?
<bavier[m]1>single-quote, I think, would be fine
<Formbi>I mean I had '( before and changing to (list didn't do anything
<Formbi>so maybe I could just do ''(
<Formbi>aha
<Formbi>it's doing something
<bavier[m]1>I usually need to look at other `substitute-keyword-arguments` uses to remind myself
<bavier[m]1>code-staging! yay!
<rndd>hi everyone! i'm trying to pack cl-liballegro https://paste.debian.net/1158445/ but got this error https://paste.debian.net/1158448/ . could anyone give any tips?
<Formbi>«(assoc-ref inputs "allegfro")»
<Formbi>is it supposed to be like that?
<Formbi>maybe is tries to string-append an error or something
<bavier[m]1>fix the spelling of "allegfro"
<bavier[m]1>otherwise the assoc-ref returns `#f`, which cannot be string-append'd
<rndd>hmmmm
<rndd>didn't got
<rndd>ohhhh
<rndd>i seee
<rndd>damn
<rndd>thanks 0_0
<rndd>ok, now it is https://paste.debian.net/1158453/ and the error is https://paste.debian.net/1158452/
<PotentialUser-84>Hello, back with the UEFI install, I was able to get the graphical install with a guided partitioning successfully installed, but I am still unable to boot. I am wondering if there are any known issues with the thinkpad t420 hardware/firmware? I have UEFI only startup enabled, but it still takes me to the intel boot agent after install
<guix-vits>Hello Guix.
***sneek_ is now known as sneek
<Formbi>rndd: «Package ( does not exist.»
<Formbi>unbalanced parentheses?
<Formbi>PotentialUser-84: I don't think so
<Formbi>I have a Thinkpad T240s and it's a worse model
<Formbi>ok, Node was built ^^
<PotentialUser-84>interesting, I have been able to install other gnu/linux systems even a few weeks ago, and they booted, I guess I'll try one of them just to test and make sure it's still possible and come back in a little bit
<rndd>Formbi: yee, not it is https://paste.debian.net/1158463/ and error is https://paste.debian.net/1158464/
<rndd>*now
<Formbi>you seem to have screwed up path specification
<Formbi>SBCL thinks it is some symbol
<nikita`>dustyweb: it's too warm to think about a right time to chat tomorrow and I am a bit jetlagged from 5 hours sleep, so I'll just be on tomorrow and you can contact me? my plan is to do job applications tomorrow morning and gsoc work in the evening
<nikita`>I just had to be in the comfy 37°C outside for a while :|
<guix-vits>+1 Yeah, sometimes one can just dream to move to Yakutia, or smth.
<dustyweb>hi nikita`
<dustyweb>yeah let's talk then
<dustyweb>get some rest! :)
<nikita`>ok :)
<PotentialUser-84>back with good news on the UEFI install. I relabeled the partition as ms-dos and installed opensuse, then switched back to uefi only, and installed GUIX and got it to boot with a gpt table. Does anyone have any ideas as to why simply deleting the partitions wasn't working the several times I tried to install it without relabeling the table? I would
<PotentialUser-84>n't think a relabel would be necessary, but I'm new to guix
<apteryx>owe have Guile native grep definitions in dub-build-system.scm; perhaps we should move those to (guix build utils) ?
<apteryx>seems they would be useful in conjunction with find-files.
<rndd>hi everyone! what is the name of zlib licence for packaging???
<rndd>* Zlib
<bavier[m]1>"zlib"
<bavier[m]1>it's actually one of the reasons the `(guix licenses)` module is usually imported with the "license:" prefix
<roptat>license:zlib if you import (guix licenses) with a prefix
<bavier[m]1>or maybe the reason
<roptat>"expat" too
<bavier[m]1>ah, yup
<rndd>hmmmmm
<rndd>thanks -_0
<rndd>hmmm, "(exception unbound-variable (value #f) (value "Unbound variable: ~S") (value (licence:zlib)) (value #f))"
<rndd>ohh
<rndd>wait
<rndd>my bad
<blackbeard[m]>If you have two distributions installed, how do you configure grub to find the new kernel every time you update the not-guix distribution
<kmicu>Does chainloader really need an explicent pointer to the kernel?
<rndd>does anybody know how to call BSD-2-Clause License in guix?
<apteryx>license:bsd-2
<rndd>apteryx: thanks
<rndd>what is the difference between inputs and native-inputs?
<apteryx>native inputs are inputs that are to be used at build time, e.g., run by the build machine directly. inputs could be libraries to be linked with, that would match the target architecture.
<rndd>apteryx: thanks
<apteryx>without taking into account cross-compilation, you can remember native-inputs as inputs only required at build time, while inputs dependencies which will be typically used at run time.
<apteryx>propagated inputs are like inputs, except they are brought into the profile, which is often undesirable as it can lead to conflicts in the profile.
<apteryx>example of native-inputs: test dependencies, linters, code generators, compilers, etc.
<rndd>hmmmm
<rndd>okay
<rndd>oh, i packed one cl package. sbcl stoped to see shared library in check phase. so i (deleted 'check ) . is it right?
<pkill9>i think propagated inputs may also affect the build itself
<roptat>rndd[m], to disable tests, usually we use #:tests? #f
<pkill9>certbot hasn't renewed my lets encrypt certificate
<pkill9>how do i do it manually?
<pkill9>and why has it not?
<pkill9>hmm
<pkill9>m cron says it will do it tomorrow
<pkill9>so who kknows
<ryanprior>the m is for mañana
***terpri__ is now known as terpri
<hendursaga>How do I check if my package definition update doesn't break dependent packages? Rebuild under ./pre-inst-env every single one of them and see if they succeed?
<Formbi>I would just call the variable some-package-21.37
<Formbi>I doubt dependent packages would break then
<Formbi>OTOH that 21.37 could become the default some day…
<raghavgururajan>Folks, how to send the `guix build foo` process to the background?