IRC channel logs

2020-09-05.log

back to list of logs

***caleb_ is now known as KE0VVT
<raingloom>ccao001: well, i hadn't managed to update it yet, but now the errors are different. will investigate more tomorrow.
<dustyweb>hm
<dustyweb>what provides setfacl again?
<nckx>dustyweb: acl.
<dustyweb>nckx: aha, ty
<pkill9>this looks neat https://blogs.gnome.org/uraeus/2020/09/04/pipewire-late-summer-update-2020/
<joshuaBPMan>pkill9 yeah, that does look awesome.
<pkill9>apparently it will be ABI compatible with pulseaudio and jack, so it will be a drop-in replacement
<pkill9>and it says it works with carla and pavucontrol as well
<pkill9>well, you can test it now, so apparently it already works for that
<joshuaBPMan>I read that! That's got to be a huge amount of work to make that work reliably!
<pkill9>yea
<jgart[m]><rekado "one on how to make the dictionar"> rekado_: Is there a wishlist somewhere for things pending that could be worked on for the cookbook?
<joshuaBPMan>jgart[m] I was going to point you to some of my documentation, but I just realized that it has my username and password for some things....
<jgart[m]>joshuaBPMan: what type of stuff were you documenting?
<jgart[m]>would anybody be able to review this package for me? https://gitlab.com/libremiami/guix-packages/-/blob/master/vis-guava.scm
<jgart[m]>and tell me what I've missed
<joshuaBPMan>jgart[m] It's my guix cheatsheet document, but it accidentally has some passwords. I
<joshuaBPMan>'d rather not go into details....
<jgart[m]>It's a color theme for the vis text editor that I've packaged
<jgart[m]>joshuaBPMan: ok no worries
<joshuaBPMan>It's nothing bad. Like buying drugs or anything. Seriously! ARE YOU A COP@!!!??? SO PARANOID!!!
<joshuaBPMan>:)
<jgart[m]>If you get around to publishing what you think might be useful I'd be happy to get a copy for self-study
<jgart[m]>The vis theme above does not yet properly source the correct environment variables for vis to be able to use it
<jgart[m]>I'd like to fix that but I'm slightly stuck with it
<jgart[m]>any help is greatly appreciated
<jgart[m]>I packaged it by first following the template for vim-lua https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/vim.scm#n403
<jgart[m]>which is a vim editor theme
<joshuaBPMan>jgart[m] I'll write down your irc info and get back to you.
***caleb_ is now known as KE0VVT
<jgart[m]>joshuaBPMan: or if you'd like you can also email me. (my matrix username at dismail dot de)
<joshuaBPMan>jgart[m] I've got a dismail account too!
<joshuaBPMan>OHHHH!!!! Are you talking to me right now via matrix?
<joshuaBPMan>that's probably what the [m] is on your name
<bdju>re: earlier icecat issues, I've been using qutebrowser, but I just clicked a link in my chat program which defaulted to opening in icecat and it used all my RAM and hanged my machine for a couple minutes. I had to pkill icecat via ssh and then it came back after a bit
<bdju>so I may have to really downgrade it or something, seems very serious
<apteryx>bdju: what's that icecat issue? I'm using a fairly recent one without issues
***catonano_ is now known as catonano
<pkill9>bdju: there's an earlyoom service which runs the earlyoom daemon which automaticlaly klils processes before they hang your computer
<pkill9>earlyoom = early out-of-memory
<bdju>apteryx: basically just that when I woke up this morning, my icecat wasn't running, and when I restored it, it maxed my cpu, kicked up my fans, and the 4 tabs I had would never finish loading, so I gave up and killed it after several minutes. but earlier it didn't hang my computer like this.
<bdju>although now with qutebrowser open maybe I had less free memory than earlier. also worth noting I actually closed my 4 pinned tabs hoping an icecat session without tabs wouldn't lag so much, but that was before the hang just now, so it must not have helped
<bdju>pkill9: thanks for the tip
<bdju>okay, got the earloom service enabled now. that does make me feel better
<jgart[m]><joshuaBPMan "OHHHH!!!! Are you talking to me"> yes, I am
<jgart[m]>via the bridge
<joshuaBPMan>jgart[m] very cool. I used to think that matrix would replace irc, but I guess it's not quite as federated as they would like...
<apteryx>bdju: your next best ally for out of RAM situations may be ZRAM. There's also a service for it. Helped my 8 GiB system a lot.
<joshuaBPMan>bdju what's the differance between zram an ram?
<joshuaBPMan>and*
<apteryx>zram compresses parts of your RAM and uses this as swap
<joshuaBPMan>apteryx cool!
<apteryx>if using zstd compression, it can often pack 4 times more data in RAM (at the cost of some CPU uses)
<apteryx>I was able to build a webkit-gtk, which wants about 12 GiB to build, on my 8 GiB system with it
<pkill9>nice
<joshuaBPMan>apteryx I should probably give that a try...though I am not typically running into low ram usage with sway...at least I don't think so.
<apteryx>browsers...
<joshuaBPMan>apteryx I suppose...I'll have to check my browser usage...and ram usage more often...But I don't think firefox uses nearly as much memory as webkit based ones.
<apteryx>My WM is currently using 4.6 MiB of resident memory
<joshuaBPMan>apteryx That's 1/2 a gig right?
<apteryx>no, that's less than 5 mebibytes :-)
<joshuaBPMan>That sounds like next to nothing...
<joshuaBPMan>Oh... your point being that web browsers are memory hogs. gotcha
<apteryx>yeah, sorry for not being clear, eh
<joshuaBPMan>No worries. I figured it out eventually. :)
<bdju>apteryx: I have 16GB of RAM and do not normally use much at all, this was probably some weird memory leak scenario
<jgart[m]><joshuaBPMan "jgart very cool. I used to thi"> the only pleasant non-bloat client at the moment for me is gomuks.
<jgart[m]>I hope to work on a package for gomuks soon *jgart adds gomuks to bucketlist
<jgart[m]> https://github.com/tulir/gomuks
<jgart[m]>I haven't tried this one yet: https://github.com/8go/matrix-commander
<jgart[m]>should this be packaged using the `copy-build-system` https://github.com/8go/matrix-commander given that it is not on pypi?
<roptat>jgart[m]: you probably want to use the python-build-system anyway, because it provides some interesting phases. You could replace the build phase to copy the files into place
<vits-test>Guten Tag.
***caleb_ is now known as KE0VVT
<jgart[m]>roptat: thank you! I'll try that. Looks like I would just have to package matrix-nio and whatever it depends on in order to try matrix-commander
<jgart[m]>roptat: Do you have any interest in packaging reasonml? I remember you mentioning that you were interested in packaging things for guix from the java and ocaml ecosystems.
<jgart[m]>I imagine the rescript-compiler would be a bear of a mess to package though
<vits-test>Any chances that `guix weather` supports -L flag (now or in future)?
<jgart[m]>vits-test: what does -L do?
<vits-test>jgart[m]: "-L, --load-path=DIR prepend DIR to the package module search path"
<vits-test>Like: "i want to see if that makeshift definition has a substitute".
<jgart[m]>right! just saw it now by running `guix help build`
<jgart[m]>vits-test: that would be nice!
<vits-test>BTW, if someone know: when finally a emacs-w3m will be unable to hang whole Emacs? Is Emacs lacks multi-threading (or whatsnot)?
<vits-test>It just horrible. Of course i can press C-g multiple times..
<vits-test>Still bad, though. "Suboptimal".
<vits-test>Maybe there some magic configure knob that Guix didn't use by default?
<vits-test>
<vits-test>Some less off-topic: Is that ususal to mcron be zombie-process?
<str1ngs>vits-test: what does guix describe output
<str1ngs>the hash.
<vits-test>str1ngs: guix 4546c0d
<str1ngs>vits-test: you need to use the channel
<vits-test>str1ngs: OK. Isn't a Overhead :) ?
<str1ngs>add http://paste.debian.net/1162722 to ~/.config/guix/channels.scm do not add it to your x86_64. use it only for the rockpro
<str1ngs>since it's pinned at 4546c0d plus one commit. this will provide substitutes upto webkitgtk. I need to work on the rest of the depends for nomad.
<vits-test>str1ngs: Of course. BTW, i need to found out how to make a (guix-just-load-all-the-modules-for-my-makeshift-trash-now!)
<str1ngs>for what do you need that?
<str1ngs>oh one sec I know what you need
<vits-test>str1ngs: just grab some package from it's module and edit it (and not track deps).
<str1ngs>export GUIX_PACKAGE_PATH=$HOME/config/guix
<str1ngs>then add stuff in $HOME/config/guix/gnu/packages
<str1ngs>vits-test: make the files in $HOME/config/guix/gnu/packages modules. guix will find them anyways.
<vits-test>str1ngs: Thanks. Off-topic: Did You knew that emacs-w3m takes 7min to render a guile.html, but w3m takes 13sec? Horrible. BTW, emacs not so cool. Pity i can't help now.
<vits-test>eww takes 46 sec
<vits-test>And thus not kicks me out of IRC :D
<str1ngs>emacs is great, just all the browers for it are not great :)
<str1ngs>also the terminals/shells are not great speed wise. eshell is not bad.
<str1ngs>one day I will convert to eshell!
<vits-test>str1ngs: hehe.
<str1ngs>vits-test: probably you need this until we can move forward in hashs. http://paste.debian.net/1162724
<str1ngs>vits-test: btw I got sway bluetooth and emacs working on the rock pro. don't know what to do about screen resolution
<vits-test>str1ngs: You mean, there is not troubles?
<vits-test>with resolution.
<str1ngs>well the resolution is kinda low
<str1ngs>I'm not use to wayland
<vits-test>str1ngs: IS X doen't work for You too?
<str1ngs>vits-test: have not tried, sway was the lowest barrier to entry. lol
*vits-test i'll use eww, not w3m, in Emacs, then. In separate instance :D
<vits-test>str1ngs: Yes, that's cool to just start it from TTY and not care.
<str1ngs>vits-test: let me know when you have the webkigtk substitute. I'll start looking at the nomad depends
<vits-test>str1ngs: If You remember some acceptable resolutions, try like `output HDMI-A-1 resolution --custom 1440x900` in .config/sway/config (default is in .guix-profile/etc/sway).
<str1ngs>vits-test: thanks I appreciate it.
<vits-test>str1ngs: BTW, i can guix pull to separate profile, using --channel (or how it called) flag!
<vits-test>Until webkitgtk fixed.
<vits-test><fat>Containers! Blockchain!</fat>
<str1ngs>vits-test: just use the channel it's full proof
<str1ngs>you can use --url with --commit but it's tedious to do that every time
<vits-test>str1ngs: Wait.. looking up..
<vits-test>str1ngs: guix pull --channels=FILE --profile=PROFILE.
<vits-test>then guix package -p nomad -i webkitgtk
<vits-test>\o/
<vits-test>+, of course the --substitute-urls
<vits-test>alias mikeup=smth
*vits-test one need to be careful using M-x term in line-mode.
<str1ngs>vits-test: do guix build webkitgtk -n
<str1ngs>always build before you install if you are not sure about substitutes
<str1ngs>how do I change the gcc for package again?
<str1ngs>wait I remember you just add it to native inputs... I think.
<joshuaBPMan>somehow gnus auto-recognized that I have a local running dovecot server running.
<joshuaBPMan>Gnus is SOOO fast now. Also, I don't know how I set up my locally running dovecot server...
<joshuaBPMan>like I legitately am really confused right now.
<vits-test>joshuaBPMan: SOOO fast is lake a boy running _to_ a history class?
<joshuaBPMan>I'm seriously confused, if anyone has any idea how I miracuously did this, I would love to know.
<joshuaBPMan>vits-test: like a boy running after the ice cream truck!
<joshuaBPMan>The only slight issue is that searching is not working...
<joshuaBPMan>it is trying to find the notmuch executible, but can't find it.
<joshuaBPMan>I suppose I could try to install notmuch again.
<peanutbutterandc>Hey guix
<peanutbutterandc>Where can I find the logs for something that is built as /gnu/store/hash-package?
<peanutbutterandc>I looked at /var/logs/guix/drvs but can't find the logs there
<peanutbutterandc>I `find . -iname '*package*'` too, and while it did return a few results, none of them matched the hash
<peanutbutterandc>and the directory that makes the first 2 letters of the /gnu/store/hash-package, does not even contain the logs
<peanutbutterandc>but maybe that is because I ran the build with --verbosity=2 and therefore it didn't save a log since everything was given in stdout
<peanutbutterandc>Oh wow. There is a --log-file option. And, the log file's hash does not match the build directory's hash.... cool
<vits-test>peanutbutterandc: probably --log-file without any options returns a newest log?
<vits-test>ls /gnu/store|grep .drv
<vits-test>* sorry
<vits-test>sneek: later please tell peanutbutterandc: example, give it a filename: `for x in /gnu/store/*linux*; do guix build --log-file $x; done`
<sneek>peanutbutterandc:, vits-test says: example, give it a filename: `for x in /gnu/store/*linux*; do guix build --log-file $x; done`
<vits-test>sneek: later tell peanutbutterandc: example, give it a filename: `for x in /gnu/store/*linux*; do guix build --log-file $x; done`
<sneek>Got it.
<vits-test>sneek: later tell peanutbutterandc: err, rather /gnu/store/*linux*.drv !
<sneek>Okay.
<zjgkkn>hi! what is a output field in packages?
<vits-test>zjgkkn: glibc, for example has three outputs: the default "out", a "debug", and a "static". It's to save the "precious Megabytes".
<vits-test>Also encountered a "doc".
<zjgkkn>vits-test: substitules exists only for "out"?
<vits-test>zjgkkn: The Manual (https://guix.gnu.org/manual/devel/en), 4.4 "Packages with Multiple Outputs".
<vits-test>zjgkkn: I think anything can be substituted
<vits-test>zjgkkn: even `guix build --log-file X` can use substitutes!
<zjgkkn>vits-test: thx
***apteryx is now known as Guest98226
***apteryx_ is now known as apteryx
<bhartrihari>Hello, is there any way to install some font without packaging it? I would need it to be listed in the output of fc-list.
<bhartrihari>Nevermind
<vits-test>bhartrihari: Seems that some ~/ dir i searched for fonts. Try man fc-cache?
<vits-test>I'd forget.
<bhartrihari>Thanks vits-test. I got it.
<peanutbutterandc>Is anybody here currently?
<sneek>peanutbutterandc, you have 2 messages!
<sneek>peanutbutterandc, vits-test says: example, give it a filename: `for x in /gnu/store/*linux*; do guix build --log-file $x; done`
<sneek>peanutbutterandc, vits-test says: err, rather /gnu/store/*linux*.drv !
<peanutbutterandc>vits-test, Oooohhhhhh.... that seems like a neat idea.... I'm giving it a shot right away just a sec
<vits-test>peanutbutterandc: No. It's all a lie. The Guix is lie. The sneek is lie. It's all a masons conspiration. Use Debian O.o
<peanutbutterandc>Noooooo!!!!!!!!!!!!!!!!!!!!!!!!
<vits-test>Don't tell em it's me told You. Just ran. RAAAAN.
<peanutbutterandc>1forest1
<peanutbutterandc>vits-test, Nope... I just `./pre-install-env guix build package-i-am-working-on` and because I'd already built it, it just gave me /gnu/store/hash-package
<peanutbutterandc>.... and it's that particular build that I want to log for
<peanutbutterandc>so how do I know which .drv file is associated with this particular build?
<peanutbutterandc>I'm confused
<peanutbutterandc>I want to see what the ./configure phase was saying for this particular change in the package
<peanutbutterandc>i'm sure it must be possible somehow
<peanutbutterandc>vits-test, Also, I have another question: and this is the more pressing one right now --- what is the use of #:modules AND #:imported-modules in (arguments)?
<peanutbutterandc>case in point `guix edit openshot`. I've seen a few packages with definitions like that.... what is going on there?
<peanutbutterandc>what does #:modules do? and #:imported-modules? and why do some packages only use #:modules?
<vits-test>peanutbutterandc: IDK.
<peanutbutterandc>vits-test, I figured the first issues out: `guix build --derivations package` to figure the package out. Then, `guix build --log-file the-returned.drv` file... but alas, the log file does not have what I wanted to see (the log of the build).. just "grafting..."
<peanutbutterandc>I see... I guess I'll have to wait for somebody else to explain that
<vits-test>peanutbutterandc: nckx said that (arguments do not use the modules from (use-modules on the top of the file. So there is imports inside of (arguments. Or so He said.
<vits-test>Or so am remember.
<peanutbutterandc>Yes... I understand that #:modules is imports inside arguments.... I get that much... the arguments section is sent to the daemon... so it does make sense...
<peanutbutterandc>.....
<peanutbutterandc>what I don't understand is, if #:modules and #:imported-modules.... I'm not sure what each of them do...
<leoprikler>i think #:modules adds them to the closure whereas #:imported-modules also generates a (use-modules)
<peanutbutterandc>leoprikler, Oh hey there! Hmmmm.... why does `guix edit openshot` (the package definition, I mean) also import the build system itself? and also (guix build utils)?
<peanutbutterandc>Take this working package, for eg: I use (wrap-program) which is supplied by (guix build utils) but without any call to #:modules....
<peanutbutterandc> https://github.com/peanutbutterandcrackers/guix-packages/blob/master/imglapse.scm
<peanutbutterandc>It's not only that, `guix edit kdenlive` uses a wrap-program too (from 'guix build utils', again) but does not use #:modules
<peanutbutterandc>I'm confused
<leoprikler>guix build utils is IIRC already imported by the build system
<leoprikler>openshot needs qt-utils apparently
<leoprikler>hence it being added to both #:modules and #:imported-modules even when the build system itself is just a python-build-system (i.e. without qt)
<peanutbutterandc>so.... #:modules just lists the modules to import, and #:imported-modules goes on to import the modules or something? o.O Kinda' seems redundant
<peanutbutterandc>I'm sure it isn't... and serves a purpose... guix is well-designed...
<leoprikler>You can think of #:modules as the transitive closure of #:imported-modules and the imported modules of imported modules.
<leoprikler>I'm not sure if there's a way of automatically figuring that out
<leoprikler>(especially not without parsing all those files)
<peanutbutterandc>leoprikler, I have to confess, I don't know what transitive closures are.... I do know about closures though
<nly>./pre-inst-env guix build bsd-games
<nly>/gnu/store/s17xc4g72dlndic6jbl9zlg01fpf7xlc-bsd-games-2.17.0
<nly>what's the hold-up on 'bsd-games': http://issues.guix.gnu.org/issue/40818
<nly>can someone take a look, thanks :)
<brendyyn>just review backlog i think
<vits-test>nly: TLDR some apps fail to run within this messy patch :P
<brendyyn>rekado_: was this the typescript transpiler you were talking about? https://github.com/Brooooooklyn/swc-node/
<rekado_>brendyyn: https://github.com/swc-project/swc/
<brendyyn>ty
<brendyyn>kinda confused on how to start building the thing
*rekado_ doesn’t know
<rekado_>I’ve never packaged a Rust thing
<brendyyn>it seems to be on npm.js.com
<brendyyn>and it wants me to curl | sh the nightly version of rust
<brendyyn>perhaps packageing babel is easier
***caleb_ is now known as KE0VVT
<raghavgururajan>Hello Guix!
<vits-test>Hallo raghavgururajan.
<raghavgururajan>leoprikler: I wanted to ask you something. On current wip-desktop branch on savannah, to test gdm, when I spawn VM with the following config and diff, gdm fails with the error https://bin.disroot.org/?a84a6786dd09b3e6#9KM4ZYB81dMtpCrhTzCQHGESWoVan5vou8QFypdZZobn
<raghavgururajan> https://disroot.org/upload/PQyvxTwkgAbFQ6G6/gdm2.diff
<raghavgururajan> https://disroot.org/upload/eUxoVKGxcFhfux8o/vm.scm
<raghavgururajan>leoprikler: ^ Do you have any clues?
<raghavgururajan>leoprikler: I think that, "gdm: GdmDisplay: Session never registered, failing", is the fatal one.
*raghavgururajan literally typed the error log in pastebin, from the VM.
<lukashevich>Hello everyone! :) I installed Guix on my computer recently and it's so much fun. However, I do enjoy using graphical software, like browsers. And I I've been usin gnu icecat for a couple of days. However, there appears to be some problem with characters -- some of them are just missing. I mean, a page get loaded but there are lots of whitespaces instead of characters that should be displayed. I figured maybe it's somehow connected with the way I installed
<lukashevich>the browser and someone might've faced it also. Thank you!
<leoprikler>I'll have a look at it, wait a bit
<leoprikler>lukashevich: seems like you're missing fonts
<vits-test>lukashevich: see https://guix.gnu.org/manual/devel/en/html_node/Application-Setup.html#Application-Setup
<raghavgururajan>leoprikler: Cool! :-)
<leoprikler>raghavgururajan: I find user "(null)" highly suspicious. Try to see what user should exist there
<vits-test>lukashevich: 2.6.3 X11 Fonts
<lukashevich>Oh, I thought about that! :) However, I haven't tried installing them yet. Thank you leoprikler and vits-test
<vits-test>Да незачто.
<raghavgururajan>leoprikler: The errors/warnings other than "gdm: GdmDisplay: Session never registered, failing" and "gdm: Child process -191 was already dead.", appears for gdm in master as well.
<leoprikler>really?
<raghavgururajan>Yeah, 3.34.1 also gave those.
<raghavgururajan>and gdm started.
<raghavgururajan>But not 3.36.3 fails with those unqiue errors.
<lukashevich>vits-test, Ha that was a funny one!
<leoprikler>when was 3.34.1 added? A few days ago or a few months ago?
<raghavgururajan>leoprikler: GDM on master is 3.34.1 and on wip-desktop is 3.36.3. The former was updated on 2020-04-18
<leoprikler>In that case I don't think the behaviour on master is sane either, because `grep null /var/log/*` shows no results
<leoprikler>trying `sudo grep -R` as we speak
<raghavgururajan>I see.
<raghavgururajan>I doubt wether something has to be changed in gdm-service-type.
<raghavgururajan>*whether
<vits-test>Hell. I almost ended writing an advert of keyboard-driven browsers for lukashevich.
<raghavgururajan>leoprikler: https://gitlab.gnome.org/GNOME/gdm/-/blob/3.36.3/daemon/gdm-session-worker.c#L397
<leoprikler>please use cons* instead of append-list
<leoprikler>also add gdm-service-type before elogind
<leoprikler>try adding accountsservice-service as well as a service of polkit-service-type
<leoprikler>probably also dbus-service
<leoprikler>your config is too minimalistic, I doubt gdm is supposed to work without those
<raghavgururajan>Hmm. But gdm-service-type should work on its own right? It extends them.
<leoprikler>extends does not mean, that it instantiates them
<leoprikler>in fact, using gdm-service-type without having one of them is a bug in your config
<leoprikler>[extends is like "if you have a service of that type, add X to it"]
<raghavgururajan>leoprikler: I see. With same config, when I tried 3.34.1, gdm starts. But I wonder why 3.36.3 just fails.
<raghavgururajan>Should accountsservice-service, elogind-service and dbus-service come after gdm-service-type in the list?
*raghavgururajan trying %desktop-services to test gdm
<rekado_>sneek: later tell brendyyn babel will most definitely not be easier to package. It’s written in JavaScript and has lots of unpackaged dependencies.
<sneek>Will do.
<raghavgururajan>leoprikler: Now with %desktop-services, all other errors of gdm disappeared. Outstanding ones are "gdm: GdmDisplay: Session never registered, failing" and "gdm: Child process -191 was already dead."
<raghavgururajan>Those two repeated six times.
<raghavgururajan>With different -NNN
<rekado_>raghavgururajan: is there any left-over state?
<rekado_>raghavgururajan: e.g. in /var/lib/gdm or /home/<you>/.local
<raghavgururajan>rekado_: I'll check
<leoprikler>hmm, that's sadly not much output still
<leoprikler>I think you can make it output more using https://help.gnome.org/admin/gdm/stable/troubleshooting.html.en
<leoprikler>question is how you get the messages out of the vm
<leoprikler>for the record, what options do you pass to the VM?
<raghavgururajan>Just -m 4096
<leoprikler>try -m 2G
<raghavgururajan>You want me to reduce from 4G to 2G?
<leoprikler>wait 4096 is 4G?
<leoprikler>try 8G then
<raghavgururajan>yeah
<raghavgururajan>LoL. All I have is 8G.
<raghavgururajan>I am afraid my host will crash
<leoprikler>hmm
<raghavgururajan>rekado_: /var/lib/gdm is empty
<leoprikler>btw. this is the line you're reading: https://gitlab.gnome.org/GNOME/gdm/-/blob/3.36.3/common/gdm-common.c#L196
<leoprikler>so not really much info from that
<str1ngs>hello nckx where you able to look at that php build issue on aarch64 I mentioned yesterday?
<leoprikler>raghavgururajan: btw. tracing it back there is only one line that would turn pids negative at https://gitlab.gnome.org/GNOME/gdm/-/blob/3.36.3/daemon/gdm-launch-environment.c#L329
<vits-test>sneek later tell nckx: str1ngs, "The php build issue": #43221
<sneek>Got it.
<vits-test>sneek: srsly? I didn't. Well..
<str1ngs>.
<str1ngs>nckx: I think your version bump to php fixed the build. will let you know.
<str1ngs>build on aarch64 to be specific. probably though you are sleeping :)
<raingloom>if ccao001 is here: i just sent the patch to update picard, idk when it will get merged.
<hulten>I have an Antminer (a BeagleBone Black of which the UART interface is severed, as is the S2 button I think) on which I'd like to install GuixSD.
<hulten>Has anyone tried this, i.e. without being able to connect an USB–TTL cable? Is it possible at all?
<joshuaBPMan>Heyo #guixoids!
<lfam>Howdy
<str1ngs>joshuaBPMan: its geeks! as in guixs
<joshuaBPMan>str1ngs ahhh! That's clever.
<str1ngs>joshuaBPMan: see footnote https://guix.gnu.org/manual/en/html_node/Introduction.html#FOOT1
<str1ngs>so tecnically you can use hello guix or hello geeks
<joshuaBPMan>str1ngs nice. That's totally a strings thing to say.
<joshuaBPMan>I'm actually looking forward to setting up guile-syntax highlight on my blog. The Haunt site works just fine, but it would look soo cooler with syntax highlighting
<str1ngs>I hear ya man
<str1ngs>how my making out with haunt?
<joshuaBPMan>I am actually liking haunt! gnucode.me
<joshuaBPMan>It works fabulously well. I've been coding my experience...
<str1ngs>joshuaBPMan: http://git.savannah.nongnu.org/cgit/nomad.git/tree/scheme/nomad/views.scm?h=devel and help-mode.scm could use some love if you are looking for a project.
<str1ngs>joshuaBPMan: nomad is a extensible web browser written in guile. I use haunts html.scm to present webviews. but they are not pretty yet.
***slyfox_ is now known as slyfox
<joshuaBPMan>str1ngs That is awesome! Are you saying the website for nomad is built in haunt? Also does nomad run inside emacs?
<str1ngs>joshuaBPMan Nomad's internal views are handled by haunt html. aka sxml . Nomad is a standalone program. but you can connect to it via geiser from emacs.
<joshuaBPMan>str1ngs That's pretty awesome. I'm assuming it's using webkit
<str1ngs>the reference implementation used webkit. later it can use others
<str1ngs>uses*
<joshuaBPMan>str1ngs that's cool. I might have to look into playing around with that!
<str1ngs>aye, and we code use someone that is good with html namely haunt
<joshuaBPMan>str1ngs Thanks for the invite. Time permitting, I may have to look into that.
<str1ngs>joshuaBPMan: understandable
<joshuaBPMan>str1ngs my main goal is to find a way to generate income AND do awesome coding projects at the same time.
<str1ngs>nice
<joshuaBPMan>str1ngs It's a good goal, but if you've got any tips, I'd love to hear them.
<str1ngs>open source projects are a good place to gain experience for sure.
<joshuaBPMan>str1ngs That's true. That experience could be useful elsewhere.
<str1ngs>vits-test: with guix describe 81ea278e05986f9ccee078bd00d4d7fc309dd19c there are aarch64 webkitgtk substitutes on bufio.org.
<str1ngs>vits-test: now building linux-libre-arm-generic
<matijja>Hello Guix!v
<wigust>matijja: hello
<matijja>wigust: hi
<str1ngs>vits-test: make run with nomad-aarch64 should work. I need to look at why jack2 is failing.
<matijja>After long time I try to run guix on my Arch machine and it throw me next error: https://pastebin.com/NbBdTJWe
<wigust>matijja: How did you install Guix?
<matijja>With installer script (https://git.savannah.gnu.org/cgit/guix.git/plain/etc/guix-install.sh)
<roptat>I'm trying to import/export, so. "guix archive --export linux-libre > linux.pack", but then "guix archive -t < linux.pack" says it's corrupted...
<andreoss>`guix refresh` shows that emacs can be upgraded to 27.1, but `guix upgrade` installs 26.3. What am I missing? I did `guix pull -u` twice already and I'm on a foreign distro
<roptat>and trying to load a substitute (after decompressing it): guix archive: error: input doesn't look like something created by `nix-store --export'
<wigust>matijja: Is your Arch up to date? Seems libc is not compatible.
<joshuaBPMan>andreoss Why guix pull -u ? I do a guix pull
<lfam>I'm not sure there is a `guix pull -u` command. Can you double-check what you tried, andreoss?
<joshuaBPMan>Can you try a reboot?
<joshuaBPMan>Also, some guix users are noticing some issues with upgrading to the latest version of emacs. Me included :)
<andreoss>Oh, sorry. I did `guix pull` indeed, it was `guix package -u`
<lfam>andreoss: I think you may have understood what `guix refresh` does. If you want to see the current version of emacs, you can do `guix pull && guix show emacs`
<joshuaBPMan>andreos guix refresh is really intended only for guix developers.
<lfam>`guix refresh` is a tool for working on package maintenance. You would use it to update a local copy of the Guix source code, test the new version, and then submit it for inclusion in Guix
<lfam>andreoss: Is there another GNU/Linux distro that you're familiar with? We can translate the commands to Guix for you :)
<rekado_>if “guix upgrade” does not do what you think, then you should check that you’re actually using the recently pulled “guix” and not something older.
<rekado_>check with “type guix” or “which guix”
<rekado_>it should be ~/.config/guix/current/bin/guix
<andreoss>`guix show emacs` gives me 26.3, but I'm unable to find any traces of "26.3" in scm files (probably a wrong place to look)
<andreoss>rekado_: it's /usr/local/bin/guix for me
<andreoss>Is that the issue?
<andreoss>rekado_: Thanks. It's now upgrading. I have no clue how guix ended up in /usr/local/bin
<lfam>It's put there by the binary installer script so that there is a `guix` command on $PATH before the first time that `guix pull` is run
<matijja>wigust: I will try to upgrade and reboot the system.
<lfam>I've noticed a few instances of people being confused by this /usr/local/bin/guix recently. Perhaps we should try to rethink it
<efraim>I have a /gnu/store/993vkza7734p654s9nfkzq49zv1j79r7-efl-1.24.3/share/gdb/auto-load/usr/lib/libeo.so.1.24.3-gdb.py and i feel like the path is wrong
<efraim>judging from /gnu/store/xa1vfhfc42x655hi7vxqmbyvwldnz7r0-glib-2.62.6/share/gdb/auto-load/gnu/store/xa1vfhfc42x655hi7vxqmbyvwldnz7r0-glib-2.62.6/lib/libgobject-2.0.so.0.6200.6-gdb.py, also meson build, looks like 'usr' should be 'out'
<efraim>basically same contents of the file too
<efraim>8 hours for the automake test suite on mips64el
<sunova>Hellol. IS there any specific tool for integration between zfs snapshots and guix?
<sunova>guix derivations, I suppose.
<joshuaBPMan>sunova that's a hard no. :(
<joshuaBPMan>licenses issues
<sunova>So sad :( Not even any efforts from the community?
<joshuaBPMan>sunova congratulations! You just became the project lead for integrating zfs snapshots and guix!
<rekado_>AFAIK you just can’t distribute binaries of ZFS with the rest of the system
<rekado_>so you can use it in a system declaration just fine
<joshuaBPMan>sunova The tiny baby little hiccup is that as rekado_ said, guix won't be able to integrate zfs snapshots. :(
<rekado_>see also the zfs package in (gnu packages file-systems)
<rekado_>I don’t know what you mean by “integrate”
<rekado_>you can use ZFS with Guix System.
<rekado_>derivations are not related to any of this
<vagrantc>it basically means you'll never get substitutes for it?
<vagrantc>e.g. you always have to rebuild zfs yourself?
<rekado_>not from our servers, right
*rekado_ —> zzzZ
<joshuaBPMan>rekado_ I didn't even know that we had zfs packaged!
<vagrantc>it's mostly an issue with oracle ... they're really the only entity that could fix the licensing incompatibilities
<joshuaBPMan>What about bcachefs?
<sunova>joshuaBPMan: I like that encouragement. Btw I didn't mean official integartion or something like that, I mean some sort of efforts to leverage snapshots to manage derivations. Words are tricky
<joshuaBPMan>joshuaBPMan frantically opens up a new tab and starts searching for the answer.
<joshuaBPMan>Oh sweet icy carrot tops! WE HAVE BCACHEFS TOOLS PACKAGED!
***jess is now known as meow
<lfam>sunova: Depending exactly on what you mean, Guix already manages things that are created from derivations in a snapshot-y way. ZFS or btrfs or other snapshot-capable file systems don't have much to add
<vagrantc>probably more elegant than the hardlink deduplication of /gnu/store
<joshuaBPMan>vagrantc good point.
<vagrantc>it seems like people who like zfs or btrfs really get excited about a lot of the features :)
<lfam>How would it be more elegant? Is that a desirable quality for software?
<vagrantc>lfam: have the filesystem handle de-duplication rather than messing around with manually checking for matching files and hardlinking them?
<lfam>I don't know about zfs, but btrfs basically does the same thing (manually checks for matches and then does something similar to hardlinking)
<vagrantc>ah.
<lfam>Anyways, the nice thing about hard links is they work across file systems, so we don't have to pick one. Eventually it may be fine to settle on a required FS, though
<vagrantc>i would've assumed it would be done intelligently, since they have fancy checksum hashes and stuff
<lfam>I mean, I think it's intelligent enough :)
<vagrantc>e.g. they already know the content before having to commit it to disk
<vagrantc>but sure, userspace hardlinking has it's advantages
<lfam>I do think we should be excited about these new filesystems. Unix is due for a total redesign of how storage is done
<sunova>lfam: I got your point. Thanks for the explanation. Is there any reference for guix base layout? I have a vague question in my mind, but I first need to know if it's valid
<lfam>The current design is suboptimal at best
<joshuaBPMan>lfam What makes you say it is suboptimal?
<lfam>The current design of storage as a concept on Unix
<lfam>sunova: Do you mean the filesystem layout?
<sunova>Yes
<lfam>joshuaBPMan: It's so simplistic, to the point that we can't reliably innovate
<lfam>Take backups, for example. There are a number of half-baked solutions for efficient backups
<lfam>It could be handled at the OS-level, with complete integration with the storage subsystems. And you could abstract physical disks away completely, for transparent remote storage.
<lfam>All these things exist, but only halfway
<lfam>You can turn them on or off and there are a million little problems
<vagrantc>all you need is gitfs :)
*vagrantc made it up, but someone's surely tried
<joshuaBPMan>lfam Are you eager for something like the Hurd?
<sunova>vagrantc: Or maybe ostree
<lfam>I wouldn't say I'm eager. I don't know will the solution will come from
<lfam>Where the solution will come from
<lfam>Probably from Apple
<lfam>Something iOS is the future of operating systems, in the sense that it's highly advanced and already widely used
<lfam>Something like iOS
<lfam>Sorry for all the typos. I'm getting distracted
<lfam>It could come from Guix, too, if we keep growing :)
<joshuaBPMan>I hope we do :)
<lfam>sunova: I don't know a hyperlink off-hand. Maybe this talk, "An introduction to functional package management with GNU Guix": https://guix.gnu.org/en/blog/2017/back-from-fosdem-2017/
<lfam>That type of Guix presentation usually includes a visual depiction of how Guix uses the filesystem
<lfam>If not, there is surely another one here: https://guix.gnu.org/blog/tags/talks/
<lfam>I think it could come from Guix because we've shown that we are willing to break old conventions when it's convenient. But a redesign of storage on Unix-derived systems will be a really bold step
<lfam>People tend to hate change
<lfam>On the other hand, on iOS, there is no concept of "files" for users or applications. Storage is transparently moved between the device and the cloud. I don't know how their backups are presented to the user, but I've never known an iOS user who "lost" something
<lfam>People with iphones have message history going back a decade, from device to device. It happens automatically when they get a new phone
<lfam>It's inspirational, IMO
<lfam>There's nothing about it that has anything to do with software licensing.
<lfam>Btrfs depudlication: https://btrfs.wiki.kernel.org/index.php/Deduplication
<lfam>My impression is that it's not practical to do live deduplication
<joshuaBPMan>lfam: I didn't know that iOS could do that.
<lfam>Some people view these aspects of iOS as too restrictive (no files?) But people seem to like using it
<nckx>str1ngs: Yes, I was asleep 🙂 However both Overdrives built php@7.4.9 just fine on the commit you mentioned.
<sneek>Welcome back nckx, you have 1 message!
<sneek>nckx, vits-test says: str1ngs, "The php build issue": #43221