IRC channel logs

2021-05-22.log

back to list of logs

***lispmacs[work] is now known as forthmacs[work]
<drakonis>ok nvm
<drakonis>the pip version is too old
<drakonis>i feel dumb now, but i can finally engage in the migration.
<drakonis>it is finally time...
<rekado>maximed: I tried to fix the polyglossia issue multiple times, but it’s really quite difficult. Needs lots of packaging of weird software that doesn’t like to be built from source.
<rekado>(seems like everyone else just copies the generated files and calls it a day)
<drakonis>ArneBab: when can we use wisp with guix?
<ArneBab>drakonis: right after `guix environment --ad-hoc guile guile-wisp -- wisp` finishes :-) — that’s the guix wisp instaREPL :-)
<ArneBab>drakonis: sounds great that you got it out of the way!
<ArneBab>(orjson)
<Guest85>hi
<Guest85>the guix install shell script was interrupted.
<ArneBab>Guest85: for a foreign-distro install?
<Guest85>when I re-do, I get A previous Guix installation was found. Refusing to overwrite.
<Guest85>ArneBab yeah
<ArneBab>Guest85: does it help too rename /gnu to /gnu-broken?
<drakonis>noice
<acksys0>I'm trying to get multi-monitor working in my new guix libvirt vm. Having some trouble. I have spice agent service enabled and xf86-video-gxl driver package installed. I see in xorg log that the qxl driver fails to load.
<acksys0>[ 25.169] (EE) Failed to load module "qxl" (module does not exist, 0)
<drakonis>i got it to work on a vm tho
<drakonis>now let's see if i can do the same in here
<Guest85>ArneBab didn;t work
<hwpplayer1>finally running gnome on real hardware
<drakonis>hmm so weird, getting some other error, thanks python
<drakonis>worked fine before
<drakonis>hmm, right, works fine if i invoke it through python
<drakonis>fun, that.
<ArneBab>Guest85: do you have /gnu ?
<ArneBab>Guest85: what’s the exact error you see?
<drakonis>i'll be honest now, i managed to make it work on the vm
<drakonis>but not in the host
<acksys0>I ran ps and looked into the xserver.conf in /gnu/store, and the module path for the qxl isn't there. Do I need to add the location to the xorg configuration myself? Installing the xf86-video-qxl package doesn't take care of that?
<drakonis>oh, there is a newer pip version in the repos
<Guest78>ArneBab: After restart, running guix-install.sh, I get "[ FAIL ] Could not obtain list of Guix releases."
<drakonis>ah jeeze its finally dealt with
<drakonis>is there a way to hide packages from showing on guix install?
<ArneBab>Guest78: can you run that script with bash -x guix-install.sh?
<ArneBab>that should give you information about the last command that succeeds
<ArneBab>(and about all other commands that run)
<drakonis>problem's finally solved, time to get moving
<Guest78>ArneBab: https://paste.debian.net/1198390/
<Guest78>deleting /var/guix
<hwpplayer1>cool
<drakonis>i'm highly pleased at this outcome
<hwpplayer1>Do you test a virtual machine on a bare metal installation ?
<hwpplayer1>It says you need to enable virtualization extension on BIOS
<cbaines>civodul, regarding mapping derivations to commits, the closest thing is maybe the package history pages, but the entry point there is the package, and it gives a range of commits
<cbaines>if you just want a bunch of commits, then such an endpoint could be added
<drakonis>hmm
<drakonis>i have a friend that'd like to use guix with cloud-init
<lfam>It would be awesome to have cloud-init for Guix
<Xe>is there cloud-init in guix?
<drakonis>^ there we go
<Xe>i'd like to plug guix into my VM stuff but i really would prefer having cloud-init at some level of the stack so i can make things a bit more homogenous
<drakonis>oh
<drakonis>there is
<drakonis>ish
<Xe>the package search has cloud-utils but not cloud-init
<drakonis>yeah so it does
<lfam>IIUC, cloud-init would need to "learn Guix" to some degree, right?
<Xe>also is there an easy way to install guix on nixos without breaking all of the everything?
<dstolfa>lfam: screw it, i'm DIYing something that can run linux-libre with some tape and a usb-c hub
<dstolfa>i don't want to use a non-free wifi chip, it's a nightmare
<Xe>oh and can i use a normal kernel with the non-free blobs?
<ytc`>hello. i've installed gnu guix to my system yesterday. but i cannot use emacs' guix interface package. every time i try to do something, this error comes out: "Error in evaluating guile expression: While executing meta-command: error: package/output-sexps: unbound variable."
<drakonis>Xe: yes
<drakonis>we have that
<drakonis>but not in the default repos
<lfam>Xe: Yes, you can use any kernel you like but only linux-libre is included in GNU Guix proper
<lfam>Check this blog post: https://guix.gnu.org/blog/2019/creating-and-using-a-custom-linux-kernel-on-guix-system/
<drakonis>gosh the best thing is how easy it is to do that
<lfam>This is "freedom zero" of the "four software freedoms": The freedom to use Guix as you see fit
<drakonis>i love that
<Xe>is there a guix package mirror that has more connection speed?
<lfam>ytc`: Sorry about that :( I guess that emacs-guix needs to be updated
<rndd>hi everyone!
<lfam>Xe: It's slow for you?
<Xe>i have gigabit but it only downloads at like 200KB/sec
<lfam>It's usually pretty fast but sometimes, usually late at night here in the New York time zone, it gets slow
<lfam>Hm
<lfam>There is a mirror for users in China
<lfam> https://guix.org.cn/blog/sjtug-guix-mirror-is-online-now/
<Xe>i'm in Ottawa
<lfam>Otherwise, I usually get several megabytes per second
<ytc`>lfam: do you have that problem either?
<lfam>ytc`: I'm not using Emacs or emacs-guix so no
<lfam>emacs-guix used to be part of Guix and that avoided these problems but the author decided to move emacs-guix into its own project :/
<lfam> https://emacs-guix.gitlab.io/website/
<lfam>I would report it there, or you could report it to Guix at <bug-guix@gnu.org>
<Xe>how big should i make the guix zvol?
<Xe>for the VM
<rndd>lfam: thanks for your help
<lfam>Xe: That's like the root filesystem?
<lfam>Or partition or whatever?
<Xe>like the entire disk for a VM
<Xe>how many gigabytes should it be
<lfam>Guix is not very space-efficient. I recommend at least 10 GB for whatever partition holds /gnu/store.
<lfam>But it depends on what the VM is for. I would make it larger if you will have a desktoip
<Xe>just a headless server thing
<lfam>I would do 20 or 25 GB to be comfortable
<lfam>It will give you a few "generations" of rollbacks
<lfam>Plus space for your application / user data
<Xe>looks like the vm image is 30G
<lfam>That can't be wrong :)
<drakonis>lfam: i should say that nix is somehow less space efficient
<drakonis>sadly
<lfam>Heh
<drakonis>because it tends to engage in mass rebuilds too much
<lfam>Well storage is incredibly cheap now, compared to when the old-school distros like Debian were created
<drakonis>my nix store is humongous because every often i need to get 15 gigs because a dependency higher up in the chain got modified
<lfam>When time is short, it's hard to prioritize shrinking the packages. But we do value saving space, and even have the `guix size` command to profile space usage
<lfam>Storage has gotten dramatically more expensive in the last 2 months, though. Even the used 4TB drives I was buying for $50 are now selling for $70
<lfam> https://www.ebay.com/itm/154146421394
<lfam>Kind of crazy
<lfam>But it's still very cheap compared to 20 years ago
<rndd>why storage prices increased?
<lfam>Overall semiconductor shortage and, for some reason, cryptocurrencies have evolved to use storage as well as compute and memory
<Xe>rndd: memecoin people made a memecoin that wastes disk space to kill the environment
<lfam>"Chia" cryptocurrency
<Xe>i wish i was kidding lol
<lfam>It completes the circle of ransomware
<rndd>meh
<Xe>actually, how would i put a random go program by arbitrary git repo into guix?
<rndd>this is kinda pain
<rndd>check (gnu packages go)
<rndd>as i remember
<lfam> https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/golang.scm
<rndd>your program should be names as package
<rndd>to binary to be build
<rndd>or something
<Xe>i have the vm up, what's the password?
<lfam>Then test based on the instructions in Building from Git and Running Guix Before It Is Installed: https://guix.gnu.org/manual/devel/en/html_node/Contributing.html
<lfam>If I understand correctly, the root user has an empty password Xe
<lfam>Same for the 'guest' user
<Church>Hmm is there a list of service modules?
<lfam>The VM image is based on this file: https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/system/examples/vm-image.tmpl
<Xe>how do i remove the gui
<rndd>Xe: remove dsktop services, i think
<rndd>also there was an example
<rndd>of headless
<rndd>Church: hi, check sources
<lfam>Church: Documentation of services: https://guix.gnu.org/manual/en/html_node/Services.html
<vagrantc>i've toyed with the idea of setting up caching proxies in north america for ci.guix.gnu.org ...
<Xe>is there a way to generate the base system config file?
<lfam>vagrantc: It's pretty easy to set up an nginx mirror based on mirror.conf in maintenance.git
<Church>Ah saw that already.
<lfam>Works great
<rndd>guix gnuservices
<Church>Guess there's no jellyfin service
<Church>Oh wait doy
<rndd>Church: guix/gnu/services
<drakonis>imagine having good docs and imagine how that makes you feel
<rndd>drakonis: meh, soures documented
<lfam>good night!
<rndd>kinda
<vagrantc>lfam: i've got a local mirror on my network, but i was more thinking about setting something up to share with the rest of my continent :)
<drakonis>rndd: i mean, when you're going from nix to guix, the differences are screaming at you
<rndd>drakonis: sooooooo
<rndd>which docs better?
<drakonis>guix's, obv
<rndd>meh
<drakonis>nix's might as well not exist
<rndd>i like nixos unofficial wiki
<drakonis>its not official and it has grown increasingly out of date in a few regards
<rndd>i personally found nix language kinda lame
<rndd>why not write nix module in ocaml or haskell
<rndd>meh, let's write another dsl
<drakonis>yes sure
<drakonis>dstolfa: someone else that feels the same way as you :V ^
<dstolfa>hah
<dstolfa>yeah, they just wrote a bad ocaml
<rndd>or really bad haskell
<dstolfa>but at least we have guix
<rndd>guix is awesome
<dstolfa>i second that
<rndd>computing is fun again
<dstolfa>so, for anyone interested about my AX201 adventures, i went on a little bit too much of an adventure and decided to DIY the laptop by getting a hardcase and gluing two usb-c hubs along the length of the back lid and then sticking any dongle necessary to use linux-libre
<Xe> https://i.ibb.co/YBDmFqt/Screenshot-higana-2021-05-21-19-55-58.png the installer is broken
<drakonis>are you using 1.3.0 or the development images?
<Xe>1.3.0 in libvirtd with a virtio disk and booting with a BIOS
<Church>Hmm. Seems I'll have to write a jellyfin module
<Church>And modules for my dns servers of choice
<drakonis>oh i see
<ix>drakonis: do you miss flakes
<ix>Or did you never use them much
<Xe>i just gave up for now, the vaccine shot i got is making my arm hurt and i can't focus
***acksys0 is now known as acksys
<ix>Mine only just stopped hurting
<ix>But, second shot done :)
<acksys>I did end up getting this working. I did have to manually add the xf86-video-qxl module to the xorg modules path in my config.scm in order for xorg to load it.
<drakonis>ix: i really dont miss them
***terpri is now known as robin
<drakonis>they existed as a bandaid to nix having non existent multiple source repo management
<ytc>i am trying to change my gtk theme with lxappearance but changes don't take effect. what am i missing?
<drakonis>it adds a layer of complexity that doesnt really improve the usability
<drakonis>flakes died in the vine really fast
<ytc>also i need to add theme packages to my config.scm file because otherwise themes don't show up.
<drakonis>ix: i genuinely do not miss it
<hwpplayer1>Hi people!
<hwpplayer1>I am searching for gpg signed commit
<hwpplayer1>Is there any trick to sign commit ?
<drakonis>ix: have you tried the docs?
<drakonis>oops wrong one
<drakonis>hwpplayer1: check the docs
<hwpplayer1>Yes I know
<drakonis>ix: where should i reply to that question tho?
<hwpplayer1>I tried but it failed
<hwpplayer1>git config commit.gpgsign true and git config user.signingkey CABBA6EA1DC0FF33
<hwpplayer1>is it the same article ?
<vagrantc>hwpplayer1: you have a secret key for CABBA6EA1DC0FF33 ?
<hwpplayer1>no it is on the article not mine
<hwpplayer1>vagrantc:
<hwpplayer1>hmm understood
<hwpplayer1>I have my secret key on external disk
<hwpplayer1>Is it about that ?
<vagrantc>you'll need to use the public key fingerprint instead of cabbageandcoffee :)
<hwpplayer1>fingerprint ?
<hwpplayer1>cool
<vagrantc>the public key fingerprint corresponding to your openpgp key
<vagrantc>if you only have one key, you m ight be able to omit specifying that explicitly
<vagrantc>if gnupg is set with a default-key or such
<hwpplayer1>I did not set default key
<hwpplayer1>Let me fix it
<vagrantc>or if it only has one, you might be able to omit it
<hwpplayer1>No it is configured
<vagrantc>e.g. no need to specify user.signingkey in your git config
<hwpplayer1>I can post my config here is it okay ?
<vagrantc>probably use https://paste.debian.net
<hwpplayer1>Okay
<hwpplayer1> https://paste.debian.net/1198393
<hwpplayer1>vagrantc: ? are you there ?
<vagrantc>hwpplayer1: not sure if many of those values are ok to have empty
<vagrantc>hwpplayer1: notably user.email and user.name
<hwpplayer1>i deleted them not to show publicly
<hwpplayer1>there are values of course vagrantc
<vagrantc>well, presuming your deleted values are correct, it looks like a valid git configuration file
<hwpplayer1>It was working with other distros
<vagrantc>for the value in signingkey ... does gpg --list-key YOURPUBLICKEYFINGERPRINT ... show the key you expect it to
<vagrantc>?
<hwpplayer1>yes
<hwpplayer1>Same gpg key
<hwpplayer1>I am aware of that
<vagrantc>can you paste the error message that didn't work?
<hwpplayer1>yes
<hwpplayer1>it is short also
<hwpplayer1>hwpplayer1@masscollabs ~/test$ git commit -S -m "mert" error: gpg failed to sign the data fatal: failed to write commit object
<hwpplayer1>
<vagrantc>try to replace with some text rather than just delete things you want to keep secret
<hwpplayer1>A command -S -m and gpg failed a general issue on any system
<vagrantc>do you have a gnupg agent running?
<hwpplayer1>that is it maybe
<hwpplayer1>that should be
<hwpplayer1>ssh is running for example
<hwpplayer1>how to run it on Guix ?
<vagrantc>admittedly, i sign my git commits on a foreign distro and haven't used guix for that :)
<hwpplayer1>cool
<roptat>hwpplayer1, you need to install pinentry, and put something like this in your ~/.gnupg/gpg-agent.conf: pinentry-program /home/hwpplayer1/.guix-profile/bin/pinentry
<hwpplayer1>cool
<hwpplayer1>I did thanks vagrantc and roptat and all people!
<roptat>yw, did it work?
<hwpplayer1>Yep It works
<hwpplayer1>I forgot to install it
<hwpplayer1>usually it comes with gpg on Ubuntu as i remember
<hwpplayer1>maybe that is why
<apteryx>roptat: I think civodul had done something to make the pinentry-program line not necessary in gpg-agent.conf; although I've never tested it
<roptat>ah, thanks for letting me know :)
<MarsIronPI> https://youtu.be/T0NYZDGTA5g the l0de radio hour is live! IRC's FINEST HOUR, the late night call in show where you are the star! Call in live, tonight's episl0de: A Small White Dog
<ix>drakonis: here is fine
<ix>Dr
<ix>drakonis: fair enough. I saw the docs. Nice, but nothing flake-like present
<ix>So upsetting
<drakonis>what do you even do with flakes though?
<Eloquence> https://youtu.be/T0NYZDGTA5g the l0de radio hour is live! IRC's FINEST HOUR, the late night call in show where you are the star! Call in live, tonight's episl0de: A Small White Dog
<alextee>anyone know what the GNU project's stand is on the freenode thing?
<alextee>is there an "official" discussion going on somewhere?
***jetomit_ is now known as jetomit
<pinoaffe1>alextee: while I don't think there's an official position as of yet, I can't imagine it being anything but "switch to libera" once it comes out
<robin>individual channels are switching to libera. either #guile or #guix (maybe both but i only checked one) is only logging from libera now
<robin>i didn't think gnu had much of an irc policy but apparently there are some official channels: https://www.gnu.org/server/irc-rules.html
<robin>(the links are to irc.gnu.org but the text below implies it's an alias for irc.freenode.net)
<dav1d>/!\ THIS CHANNEL HAS MOVED TO IRC.LIBERA.CHAT #HAMRADIO /!\
<dav1d>/!\ JOIN #HAMRADIO TODAY. THIS CHANNEL HAS MOVED TO IRC.LIBERA.CHAT #HAMRADIO /!\
<robin>even #gnu has 1/3 as many members as on freenode (and ofc many people are still on freenode as channels migrate, so it's probably closer to a 50/50 split in terms of apparent network preference (and ofc not everyone knows what's happening with freenode at this point, it's barely been a week since there was any public indication that something might be wrong)
<robin>)
<robin>ah, yes, irc.gnu.org is just a CNAME for irc.freenode.org
<albanc>/!\ THIS CHANNEL HAS MOVED TO IRC.LIBERA.CHAT #HAMRADIO /!\
<xmx>Why is Haskell GHC package is build form tarball rather than git source ?
<rekado_>we generally use published release tarballs
<rekado_>GHC is no different
***rekado_ is now known as rekado
<xmx>rekado_ : Whats the normal source git or tarball when creating new packages ?
<leoprikler>Whatever upstream uses
<cda__>/!\ THIS CHANNEL HAS MOVED TO IRC.LIBERA.CHAT #HAMRADIO /!\
<cda__>/!\ JOIN #HAMRADIO TODAY. THIS CHANNEL HAS MOVED TO IRC.LIBERA.CHAT #HAMRADIO /!\
<xmx>leoprikler: So can it be tarball releases ?
<xmx>Auto genetated ones ?
<leoprikler>Wdym auto-generated ones?
<leoprikler>the tarballs generated by Github/Gitlab are not usable, they're unstable
<xmx>Okay. Thanks.
<hwpplayer1>hi people!
<vivien>The problem with tarballs is they encode modification dates, and so if you re-genenrate the tarball these dates will change and as a result the tarball too
<xace_>/!\ THIS CHANNEL HAS MOVED TO IRC.LIBERA.CHAT #HAMRADIO /!\
<leoprikler>vivien is there such a thing as `tar --january-first-1970`?
<vivien>I don’t know
<vivien>There’s a dist-package function in guix
<vivien>It runs make dist (or distcheck, I don’t remember)
<vivien>But since the result is a tarball with modification dates in it, it is not reproducible
<vivien>I wish it were
<vivien>But then, you also need to use the exact same versions of the autotools to re-generate the tarball
<vivien>So yeah generating the tarball reproducibly is not something one might want to do
<leoprikler>would be great if future autotools simply had an option to do that
<jpoiret>is golang software hard to package in guix?
<xxn>Does any one use haskell language server setup on Guix destro for Emacs ?
<leoprikler>jpoiret: somewhat cumbersome from what I can tell
<leoprikler>I still haven't configured lsp-mode :P
<leoprikler>nvm I don't even have it installed, hah
<xxn>Im tryed to set it up but failing.
***jonsger1 is now known as jonsger
***dekenevs is now known as mitzman
***mitzman is now known as dekenevs
***dekenevs is now known as mitzman
***mitzman is now known as dekenevs
***dekenevs is now known as mitzman
***mitzman is now known as dekenevs
<lfam>Hello all
<lfam>I won't be able to do the kernel updates this weekend. If anyone would like to do it before Tuesday, feel free!
***qyliss is now known as ^_^
***^_^ is now known as qyliss
*roptat translating :)
*oriansj strapping them boots soo hard
***Noisytoot is now known as [[
<xmx>Any tips on making new package defs for Haskell libs and programes ? Because depens of 'guix import hackage "packagename"' given is bit different than the depens showed in hackage site.
<oriansj>xmx: that is because guix doesn't have hidden dependencies.
<xmx>Also it is different from guix package that is already there. For ex. take 'ghc-vector' in all the three cases there is only one or two depens are common.
<xmx>Compare 'guix import hackage vector' vs 'guix edit ghc-vector'. Why such difference ?
<oriansj>xmx: one is the software's best quess of what is needed and the other was written and tested by a person
<tissevert>hi guix
<roptat>hi tissevert :)
<tissevert>hey
<tissevert>so it was a matter of locale after all for the website's issue
<roptat>tissevert, yes it seems so, I pushed a patch, so the readme is more correct now
<tissevert>great !
<tissevert>I missed that, I've spent a little time trying to read guix-patch and didn't even see it
<tissevert>the input flow is way too much for me ^^'
<roptat>tissevert, I didn't send it there, I pushed directly
<roptat>should have answered though, let me do that
<tissevert>I'm still amazed by the flow of the mailing lists
<roptat>yeah, it's becoming hard to follow, even if I ignore most of it
<maddo>hello, is there a proper channel to report outdated/not working packages?
***[[ is now known as Noisytoot
***MidAutumnMoon9 is now known as MidAutumnMoon
***jakesyl_ is now known as jakesyl
<ytc>hello. how can i change my gtk theme? lxappearance changes don't take effect and i am not using any desktop environment.
<zap>maddo: I think its okay to report it here but there is a dedicated email for bug reporting: bug-guix@gnu.org
<maddo>thank you zap
***maddo` is now known as maddo
<zap>hey guix, what do you do if you want gexp to take arguments? One thing I thougt of is to insert paremeters as gexp inputs
<zap>but having #$parameters all over the placy quickly becomes tedious to manage
***jakesyl_ is now known as jakesyl
<maddo>guess they're still coping with sudden influx of traffic
<maddo>hope they solve it soon
<zap>maddo: I think they are in unususal situation :) probably so much exitement about a new IRC network is rare these days
<raghavgururajan>Hello Guix!
<zap>raghavgururajan: hello!
***anonymous is now known as river
<nckx>Hullo Raghav, Guix.
***terpri_ is now known as robin
<raghavgururajan>hmm. things built on bayfront aren't made available as substitutes.
***sneek_ is now known as sneek
<raghavgururajan>maddo: If any of them are gnome packages, let me know. :)
<nckx>...it's not caching, it can't be caching, it was caching?
<nckx>AFAIK that ‘can't happen’ otherwise.
<maddo>raghavgururajan: amule and emacs-telega actually XD
<nckx>Which substitute in particular, raghavgururajan?
<raghavgururajan>You mean caching on my machine? I also tried `sudo rm -rf /var/guix/substitute/cache/*`.
<raghavgururajan>tracker and its dependencies.
<raghavgururajan>wip-gnome
<nckx>There, or in guix-publish, or in ngixn (I don't know offhead if bayfront uses a proxy).
<raghavgururajan>maddo: I see.
<nckx>A store path if possible?
<nckx>I assume you compared hashes to make sure you're actually requesting the exact same thing.
<raghavgururajan> /gnu/store/ak7if97imnhkfp2zi3wbxi1g4mgbq7kz-tracker-3.1.1.drv
<raghavgururajan>My local git checkout and checkout on bayfront are on same commit.
<raghavgururajan> /gnu/store/0zyp4dmibl3y57hi8zdgj3ipla8xkdgf-gnutls-3.6.15.drv
<raghavgururajan>hashes are same.
<xmx>Why are those cabale revision are present in haskell package definitions as arguments ?
<nckx>Yes, I can reproduce with cURL. Buttehm: herd: service 'guix-publish' could not be found.
***drakonis1 is now known as drakonis
*raghavgururajan brb
<nckx>Is this something that changed because of the Guix Build Coordinator?
*nckx wonders if cbaines has a ping set up for those words :)
***dongcarl2 is now known as dongcarl
***ix is now known as Guest4079
***xkapastel_ is now known as xkapastel
***Guest4079 is now known as ix
***ix is now known as Guest880
***Exagone314 is now known as Exagone313
***TheCreeper1 is now known as TheCreeper
<raghavgururajan>nckx: Ah.
***leoprikler_ is now known as leoprikler
***Exagone314 is now known as Exagone313
***dongcarl9 is now known as dongcarl
***TheCreeper1 is now known as TheCreeper
***xkapastel_ is now known as xkapastel
<dstolfa>i assume mattermost and element are not in guix because the evaluation of how free they are is still going on?
<dstolfa>namely, electron?
<raghavgururajan>dstolfa: and zulip
<dstolfa>yeah, anything electron really
<raghavgururajan>Yep!
<jackhill>dstolfa: As I understand it, it's not about how free they are in a nominal sense (they probably are free), but in a practical one: untangling the knot of javascript dependencies is a lot of work, and it has not been done yet.
<jackhill>For Zulip there is at least the zulip-term terminal client in Guix.
<dstolfa>jackhill: yeah, that matches with what i've heard about it. is there an ETA on when they might be untangled or is just a case of "wait an see"?
<raghavgururajan>👆️
*raghavgururajan 's message came late
<jackhill>dstolfa: I don't think we know enough abotu how close we are to a tipping point. plus as a volunteer effort, it goes in fits and starts. I haven't personally worked on it, so don't have a better sense than that. Jullien L (I think) has been doing a lot of that work.
<dstolfa>jackhill: alright, that sounds reasonable. thanks for sharing
<jackhill>I do think it would be a great boon to Guix and free software folks when we do work it out.
<xmmx>How does :cabal-revision work in haskell build system ?
<jackhill>I'm someone hopeful that some people from the land of javascript are finding value in untying the knot with projects like esbuild (https://esbuild.github.io/ in Guix) and rome (https://rome.tools/ not in guix yet), and that will help greatly.
<jackhill>s/someone/somewhat/
<merazi>greetings
<jackhill>greetings merazi!
<merazi>:-)
<merazi>How are you?
<jackhill>merazi: I'm well, and you?
<jackhill>xmmx: I don't know, hopefully someone else does
***lispmacs is now known as forthmacs
<solene>I made a git clone of guix repository, followed https://guix.gnu.org/manual/en/guix.html#Building-from-Git but at "make check" I have a failure on "FAIL: tests/channels.scm"
***forthmacs is now known as lispmacs
<merazi>jackhill: I'm fine as well
<raghavgururajan>dstolfa: You might be interested in node-importer work. https://git.savannah.gnu.org/cgit/guix.git/log/?h=wip-node-importer
<solene>is it expected?
<dstolfa>oh hm, yeah that does look interesting. thanks raghavgururajan
***drakonis1 is now known as drakonis
<bdju>do appimages work on guix? never used one before but was just trying one and afaict nothing is happening
<dstolfa>bdju: yes, but you may need to put the linker in the global path
<bdju>linker? what would that process involve exactly?
<leoprikler>patchelf is your friend when dealing with appimages
<bdju>alright
<dstolfa>bdju: it would involde finding your glibc's ld-linux-x86-64 and symlinking it to /lib
<dstolfa>involve*
<dstolfa>or patchelf, as leoprikler said
<dstolfa>works too
***Noisytoot is now known as [[
***xkapastel_ is now known as xkapastel
***[[ is now known as Noisytoot
<pkill9_>does this work without setuid? https://github.com/tsgates/mbox
***Guest880 is now known as ix
***ix is now known as Guest444
<pkill9_>nvm
***Guest444 is now known as ix
<jackhill>pkill9_: mbox looks like interesting software, does it work without setuid?
<jackhill>I guess it hasn't been updated recently. I wonder what that means (probably that "better" sandboxes have been developed (a la gvisor))
***jackhill is now known as KM4MBG
***KM4MBG is now known as jackhill
<pkill9_>that's what i wanna know, but tbh there needs to be a better solution for sandboxing
<solene>I don't understand https://guix.gnu.org/manual/en/guix.html#Sending-a-Patch-Series , should I send a single patch file to both emails addresses?
<solene>or is guix-patches@gnu.org enough?
<vagrantc>solene: if you have updates to the patch or multiple patches in a series, then you should send them to the appropriate bug number after the initial submission to guix-patches@gnu.or
<vagrantc>g
<solene>ok I see
<vagrantc>the first mail to guix-patches will give you a bugnumber, and then follow-ups should go to the corresponding BUGNUMBER@debbugs.gnu.org
<vagrantc>though the important thing is to contribute; if you get something small wrong it's no big deal, we can always improve it later :)
<solene>I'm taking the opportunity to check the accuracy of the documentation ;)
<solene>first patch sent \o/
<vagrantc>solene: also good to help improve the documentation :)
<solene>I'm relying on https://guix.gnu.org/manual/en/guix.html for everything, it's not always easy but I've been able to manage so far :D
<solene>I had to setup a wireguard tunnel to be able to send the stupid patch, took me a while to understand how to mix %desktop-services with the vpn service :D
<vagrantc>yeah, sometimes it makes assumptions that can be hard to follow ... though it's more well documented than many projects
<solene>am I doing right in regards to the process to work locally? I have a term in the guix git repo with "guix environment guix --pure --ad-hoc help2man git strace less" and I use ./pre-inst-env guix {build,install} something , and from another shell I make changes to scm files
<vagrantc>sounds reasonable
<vagrantc>you might want to use a separate profile for installing things ... that's a judgement call
<vagrantc>e.g. i try to keep my default profile just with stuff from guix proper, experiments in separate profiles
<solene>I'll try later maybe, I'm really struggling. I daily used guix like 4 years ago but it seems I forgot everything :D
<solene>Kudos to the people involved in the installer, I've been very surprised to see the installer with full disk encryption also encrypted /boot , it's the only distribution I know doing that.
<vagrantc>the encrpyted /boot part is only relatively recently possible with grub ... though i supposed i've been using that for quite some years now :)
<vagrantc>it's technically possible with debian-installer, but you have to set it up manually ...
<vagrantc>though i think it still requires using luks 1 format, rather than luks 2 ... unless grub learned how to do luks 2 :)
<vagrantc> May 02 help-debbugs@gnu.org ├─►bug#48172: Acknowledgement (support split /boot partition) (debian inbox)
<vagrantc> oops ...
<vagrantc>meant to link to https://issues.guix.gnu.org/48172
<vagrantc>as that would allow luks1 for /boot and luks2 for /
<solene>should I expect a mail from guix-patches@gnu.org in return to mine?
<vagrantc>yeah, it may take a while ... not sure what a reasonable delay to expect
<vagrantc>but it should eventually reply with a bug number and a little bit of information about what to do with it :)
<civodul>the first message to the list can take a few hours to reach it; subsequent messages go through instantaneously
<vagrantc>same with debbugs.gnu.org ?
<civodul>that applies to all the lists at lists.gnu.org
<civodul>and debbugs sits behind them
<civodul>so yes
<vagrantc>ah, didn't realize debbugs was *behind* the list interface
<civodul>kinda annoying, but i guess it helps reduce spam
<civodul>well, i think it's behind it
<civodul>i don't know the plumbing details actually :-)
<vagrantc>pay not attention to the bug report behind the curtain
<solene>looks like greylisting ?
<solene>I wrote a little something to remember how to setup wireguard :) https://dataswamp.org/~solene/2021-05-22-guix-wireguard.html
<vagrantc>in addition might need moderator approval for the initial post
<civodul>solene: nice! that could be a useful addition to the manual or cookbook
<solene>it seems the wireguard module lacks a setting for keep alive (this is optional in wireguard), that's a nice introduction to a service ^^
<solene>civodul: which cookbook? I only found the manual
<civodul>solene: https://guix.gnu.org/cookbook/en/guix-cookbook.html
<civodul>(linked from <https://guix.gnu.org/en/help/>)
<civodul>it's meant to be more informal and "application-oriented" than the manual
<solene>I like thos
<vagrantc>i must admit, my first habit when looking at documentation is searching for examples :)
<ruffni>`guix install mes` fails on my x86_64 debian host. is there anything special i need to do/consider or is it a bug?
<civodul>ruffni: looks like it fails here too: https://ci.guix.gnu.org/build/286577/details
<civodul>could you report it to bug-guix@gnu.org?
<ruffni>sure thing
<civodul>from the page above it might be related to the upgrade to mes 0.23
<ruffni>it fails like this: "test/test3/hello.sh: line 23: GET_MACHINE_FLAGS: unbound variable"
<civodul>indeed
***john2gb3 is now known as betelgeuse
<ecbrown>apteryx: i made a little progress on qt6. i have my work in https://github.com/ecbrown/guix.git on branch qt6-updates. basically i tried to follow your example for qtbase, moving qt5 to *-5 and then adding qt6 upgrade. (i still have to work on the new modules that don't have a qt5 analogue)
<ruffni>civodul: is there a template on what to submit? or should i just describe the bug in a couple of words and add the link to ci you posted before?
<daviid>hello! fwiw, g-golf now works with guile >= 3.0.7, does not depend on gtk+-3.0 any more ('since' commit 40d0756) and works with gdk4/gsk4/gtk4 - it would be nice if someone could update the package 'accordingly' and point to the latest f04756ea967281b0b1c34a9043e64542e034196e commit ... tx!
<civodul>ruffni: yes, what you write; you can also paste the last few lines of the build log for convenience
<civodul>TIA!