<lilyp>civodul: as far as I can see it ought to be a couple of short fixes <KarlJoad`>Has anyone ever had their bootloader "disappear" on Guix System? <KarlJoad`>I use hot-swap bays on my desktop, and when I remove the Guix System drive, and boot from another, then go back to Guix, the Guix drive cannot be booted from. (The PC is shutoff when I switch them.) <sneek>flinner, you have 1 message! <mekeor[m]>wow, maximed has really exciting projects. one is using guix as a replacement for cargo. and they seem to work on guile bindings to gnunet, which i thought had fallen asleep. <PotentialUser-57>ok so i tried the package rollback command and went back one generation, now how do I go back to where I was lol? <lilyp>PotentialUser-57: switch-generation WHERE_I_WAS_LOL <lilyp>for guix-package, it's --switch-generation=WHERE_I_WAS_LOL <lilyp>style differs a little between system where it's a command and package where it's an option <mekeor[m]>btw, this should be tidied up, imo. guix-cli shouldn't end up as a mess, like nix did <PotentialUser-57>lilyp: thank you. that's a bit weird that the syntax is different but thanks for informing me <PotentialUser-57>mekeor[m] yeah i'll probably forget it by tomorrow and then i'll be confused next time i need to use it again :D <PotentialUser-57>by the way this might be unrelated so i dont know if im allowed to ask this but does someone know how i can stop xdg(?) from automatically recreating folders in my home folder that i dont want? I tried editing .config/user-dirs.dirs and set it to read only but it didnt stop it <lilyp>if you're using gnome, it actually sets that up to follow your $LANG <lilyp>or well, the language you specified for gnome, which is allowed to differ IIRC <PotentialUser-57>Unrelated, so updating/upgrading the base/services is completely separate from updating/upgrading manually installed packages if I understand it correctly. Is it possible to add single packages to the config.scm and have the whole thing act as one? For the cases where that would be desirable i mean, which is not always <lilyp>the operating-system config has a packages field and services can themselves also add packages <mekeor[m]>imho, you don't need to feel bad if you tried to find a solution for your issue yourself first, e.g. by using a web search engine <PotentialUser-57>mekeor[m]: thanks, yeah i tried something else first, someone said to delete the text in the file. I just commented it out instead. didn't work. set it to read only and still didnt work! but now it worked thanks to ur help :) *mekeor[m] hands lilyp a medal *vagrantc was hoping for metals <PotentialUser-57>mekeor[m]: I guess there is one other completely random question, and that is if you know if there is a way to change the default search in icecat without having to install a plugin. Not talking about the bookmark/letter searches <lilyp>I'm a little scared – how come my emacs-wide-int is already built on CI? <nckx>Probably to late to be useful, but unmatched-paren: I have never used GDM on Guix. <nckx>lilyp: How odd are we talking? <lilyp>"I submitted the patch 1 hour ago" odd. <lilyp>Unless someone quietly submitted it to CI, it should build locally <nckx>I used to do that a lot, not lately, but anyway it's not untinkable. <lilyp>oh, sure, it's thinkable, but even then I'm surprised by the speed <lilyp>btw. FWIW all of 55419 builds locally, but I'm leaving a small review window <nckx>*your patch didn't actually change the hash. <lilyp>it goes from broken quasiquote to supreme G-Exp <vagrantc>so, libxmlb accepted my proposal upstream to move xb-tool to bin (was in libexec) but hasn't yet released a version that supports it ... seems a smallish number of commits since the version in guix ... should i package the git version just to make guix's diffoscope coverage better? :) <vagrantc>seems a bit of hassle to just pull the patch from libxmlb upstream, but that might work too... <vagrantc>or i could hack the current libxmlb package to just install it in bin instead <nckx>I'd say sure, some purists would prefer release + cherry patch. If you want to ingratiate yourself with them and it's not too much trouble... <nckx>Bleh, took so long to type it's obsolete. <lilyp>definitely go with the cherry-pick over a plain hack *nckx hisses '...purist! Purissst!' <vagrantc>cherry-pick the patch and stick it in gnu/packages/patches/ and all the corresponding files? <littlebobeep>does anyone know of an analysis or critique of snap packaging from a security perspective? <mekeor[m]>lilyp: you worked on emacs-minimal, right? btw, did you know that --with-cairo is default since emacs 28.1 iirc? i.e. --without-cairo needs to be used explicitely. thus, iiuc, emacs-minimal should use --without-cairo config flag. <lilyp>I don't think with/without-cairo is meaningful when cairo is not found <lilyp>as for the flag in emacs-28, yes it could probably be dropped <lilyp>I don't think that'd be nice towards people travelling back in time right now <mekeor[m]>nckx: isn't "chad" a word originating in a very sexist and patriarchal part of the social platform web? <littlebobeep>mekeor[m]: I actually had to look it up, it mentioned something called the "manosphere" and I severely lost interest <lilyp>to be fair, it's also used to mock that very sphere <mekeor[m]>making jokes about something still spreads the idea of that something, sometimes. especially, when it's not recognizable as a joke, imo <PotentialUser-57>lillebobeep: isn't that the one that's centralized and all snaps are downloaded from a proprietary canonical server? <mekeor[m]>there is --with-input=PACKAGE=REPLACEMENT, but is there a --with-propagated-input, too? <lilyp>there should be no difference <lilyp>you can't add inputs to packages via command line anyway <mekeor[m]>lilyp: i don't wanna add a prop-input, i want to change the version of a prop-input <KarlJoad`>I am running into an issue when developing Guix. I am trying to access a service-type I define in Guix itself, but when I test it on a file using that new service-type, I get an error with suggesion about importing the module, but I _did_ import it. <KarlJoad`>Trying a full recompile of Guix now, to see if that changes anything. <dcunit3d>if i want to start this service as the user, but under a separate shepherd process, how do i go about doing that? <littlebobeep>PotentialUser-42: I don't know because other distros have snapd that are not Ubuntu, I haven't studied how it works, I am just curious about it's security <xelxebar>KarlJoad`: Your password is *************? <xelxebar>Is there a mechanism for adding xkb layout definitions without having to push them upstream to xkeyboard-config? <ulfvonbelow>today I learned that guix will complain about downgrades even if the commit being changed to is the channel introduction commit <KarlJoad`>I cannot seem to find an explanation for my unbound variable problem. The module is defined, and Guix is suggesting my module exactly as I have spelled it, but it still fails. <lilyp>where's home-msmtp-configuration in the latter? I don't see it <lilyp>neither home-msmtp-service-type <KarlJoad`>The 2nd is a testing configuration file. The first is the new mail module I am making for Guix Home. <KarlJoad`>The config file uses this new module to test my code. <lilyp>I get what you're trying, but you're missing half of the glue as far as I can see <KarlJoad`>What glue in particular? This is the first time working with the Guix codebase and Guile lang. <lilyp>A typical service instantiation reads like (service service-type (service-configuration (field value))) <lilyp>you have (home-environment (field value)) <KarlJoad`>I'm not quite sure I follow. In the example config, the service-type is instantiated using as a service in the list. Currently, the service-configuration is "defined" but is not used and does nothing. <lilyp>are you sure service-configuration does nothing or are you just typing things wrong? <KarlJoad`>I mean, I never use the configuration in the testing config, nor do I use it in any code in mail.scm. <KarlJoad`>I removed the service-configuration code from mail.scm, ran `make` and then ran `./pre-inst-env guix home build /tmp/test.scm`, and it still fails with the same error about home-msmtp-service-type being an unbound variable. <lilyp>because you're doing the exact opposite of what you ought have done <lilyp>your module doesn't look too wrong – the configuration record appears to be missing some important fields, but you'll see that soon enough <lilyp>your home-environment on the other hand specifies all the wrong fields and none of the right fields <KarlJoad`>Yeah, I have not really gotten the fields too far yet. I don't know what I want to expose to the user with home-msmtp-configuration. <KarlJoad`>The home-environment does not accept email addresses by default, I know. I added that as another change in an earlier commit. <lilyp>why add that to home-environment and not your service? <ulfvonbelow>to be clear, the error message is complaining that home-msmtp-service-type is unbound? <KarlJoad`>I am currently migrating Nix home-manager work I have done to Guix Home. home-manager allows defining a list of email addresses for your user, and using that for its equivalent of services. <ulfvonbelow>and if you comment out the (emails ...) field specifier temporarily, you get the same complaint, right? <lilyp>"I'm migrating my stuff" doesn't necessarily mean it's a good idea to create 1:1 copies <KarlJoad`>lilyp: I know, and it will change. Just wanted to wet my feet with a small change defining a list in the home-environment. <ulfvonbelow>then I'd say to run a 'guix repl' instance and try (use-modules (gnu home services mail)) and see what happens <KarlJoad`>ulfvonbelow: When I run `./pre-inst-env guix repl` and evaluate `%load-path`, I get a list that does include my development checkout of Guix. <ulfvonbelow>and the (use-modules (gnu home services mail)) succeeds, and lets you evaluate home-msmtp-service-type successfully? <KarlJoad`>It allows the first part (if I first do `(use-modules (gnu home))`), but the second part fails, again complaining about an unbound variable. <KarlJoad`>ulfvonbelow: The first command `,m ...` fails with `concatenate` being an unbound variable. <KarlJoad`>With the srfi-1 added, I now error on "home-msmtp-service-configuration: unbound variable" <ulfvonbelow>ah yeah, that should be home-msmtp-configuration, not home-msmtp-service-configuration <KarlJoad`>Now the `define-record-type*` is an unbound variable. <KarlJoad`>That did it. Now both of your REPL commands are working. <ulfvonbelow>hopefully that's all that needs fixing and guile's error reporting simply could use some improvement <KarlJoad`>Now I am running against the mixed-text-file function not being defined. I feel like that one is defined in a Guix module somewhere. <ulfvonbelow>on an unrelated note, our xonotic package seems to be terribly crashy <KarlJoad`>Thanks for your patience ulfvonbelow and lilyp. I would send those thanks actions if I knew how. <zacchae[m]>I see the chatty app for sms, but is there one in the repos for calls? All the ones I found seem to specify VoIP, which I don't think is what I want <zacchae[m]>(I can't be the only one trying to get guix to work on a cell phone) <the_tubular>Guix on a pinephone / librem is definitely something i've seen mention here multiple times <dirtcastle>lets say for example. i write a package definition for master branch( or upstream) if the new version comes do I have to write the definition again? or will it just upgrade automatically <kitty1>hmm, anyone else have any problems right now building new versions of emacs/emacs packages that have the dependency of emacs-buttercup? <kitty1>the_tubular: you've seen as a concept or seen people do? I'm very curious how well that could work out for people <the_tubular>Umm, all I remember was a channel created for phones. <the_tubular>kitty1 there's been big changes to emacs today, a few packages broke <jgart[m]>librem or pinephone pro would probably be the ideal computers to run Guix <kitty1>the_tubular: ah, if you find the channel feel free to get my attention uwu ; yeah, I am excited to mess around with the new versions of emacs-next when some of those packages unbreak lmao <jgart[m]>I tried Guix with Mobian on a regular pinephone and it was way to slow to be of practical use <kitty1>I don't really like a lot of choices those two groups make (librem and pine) , normal pine products honestly wouldn't be too bad if they focused a little more on modularity/hackability tbh <kitty1>just like, a *tiny bit* more; like how their laptop I don't even think has a replaceable wifi adapter <kitty1>the_tubular: even if its unmaintained if I ever try to mess with it seeing an example of what other people needed to do would make things smoother lmao <kitty1>yknow; what other small companies makes linux compatible phones? I've heard a lot of good things about fairphone but I assume thats just for android-derived systems or nah? <the_tubular>But yeah guix and emacs with a touch interface on a phone would be great lol <kitty1>the_tubular: honestly, that would be quite neat if done correctly <the_tubular>Yeap, and a small keyboard, for when you need something more specific <kitty1>it would probably require a good bit of hacking to make it reach its potential ; but suppose thats what emacs shines at <the_tubular>I'll try and google around for the channel, but no guarantee <kitty1>I really need to see what exists for <kitty1>minimal emacs-like functionality with scheme extension instead of elisp <kitty1>I think there is a dedicated project like that actually <kitty1>would be neat to mess with some time <kitty1>the_tubular: my dream would be running a minimal-defaults scheme extensible emacs-like thing and just building it up from the bare bones on to a guix/literally-any-free-os-but-linux system <the_tubular>guix/literally-any-free-os-but-linux system what do you mean by that ? <kitty1>and having that all like you said on every type of device <kitty1>the_tubular: some BSD instead of linux, or something more like Hurd or Plan9 if anything becomes a solid daily driver <kitty1>yeah, which is a bit of a shame tbh <apteryx>rekado: you may be interested in having a look at #55424, where a couple old bioinformatics (and others) Python 2 packages would be removed <kitty1>since Hurd isn't being developed seriously last I heard, and many BSDs seem to be quite elegant <apteryx>haven't you tried the child hurd VM? <kitty1>actually I have not, I really should some time. <the_tubular>I know apteryx, but having it ran in very precise Hardware / Virtualised. Won't probably help adoption <apteryx>you need to start somewhere. It can reuse most drivers from linux so I don't think the road would be necessary so long <kitty1>I want something I can actually throw on different devices <kitty1>even plan9 can be thrown on a wide array of devices <the_tubular>I'm not bashing hurd apteryx, just saying it's still far away from mass adoption. <kitty1>apteryx: Since you seem relatively familiar ; whats the current state of Hurd? know if anything of note has happened since the last hurd-related guix blogpost? <the_tubular>Didn't knew it can reuse drivers, that's actually pretty cool! <apteryx>sorry, I'm not that familiar. I think the hurd itself may have had some developments but on the Guix front (integrating it with the system) not much has moved since <apteryx>at least some drivers like audio and networking I tihnk <kitty1>its really a shame there isn't more active development there; if there is ever an even half functioning arm version of it I might throw it onto my laptop. <the_tubular>I might be wrong, but at this point I kinda gave up on hoping hurd will be usable in my lifetime <the_tubular>Definitely wanna try it, but if it's in like 50 years ... <apteryx>you can serve a web site on the Hurd today <the_tubular>HTTPS log balanced with centralized logs and everything ? <apteryx>the interface mostly appears just like a good old POSIX system, so everything just works, for the most part, which is a good part of the appeal of the Hurd <the_tubular>I mean if guix makesit very easy to switch that would be cool <apteryx>(i.e. you can the fancies of a microkernel while keeping the tools you know working) <the_tubular>2-3 lines in your system configuration. And you change OS <kitty1>think Hurd would be great as a temporary "it just works" situation if there was ever proper resources going into it <kitty1>really I want to see a nuclear swap kinda like what plan9 did and for a modern full system to be brought up from there lmao <kitty1>from the outside; the two biggest things hurd needs is <kitty1>1. more accessible and thourough documentation ; maybe even some video series <kitty1>2. just some actual resources put into it and get someone working on it full time <kitty1>and always having at least a couple people full time on it with no major gaps <kitty1>on the first point ; they really need to take a lot from guix on that because like <kitty1>usually guix isn't the worst if I want information about it even if im not using guix (which, I am, but I digress) meanwhile from the outside I literally don't know how to get a lot of semi-accurate information on hurd. <kitty1>but its like, not complete shit like it seems hurd is from the outside lmao <kitty1>I tried, and last I tried I couldn't find much at all. maybe I just missed some grand source of information but it seems all relatively undocumented. <kitty1>going to have to look at it closer when I try to mess with it in a vm some time but it just <kitty1>doesn't give me good vibes, a hardly actively developed for project that doesn't seem to have clear documentation that was easy for someone outside the project to find <bjc>not to rain on the parade, but i used the hurd back in the 90s <bjc>3-4 years is pretty optimistic <kitty1>I just want to see more variety than linux, especially in the guix/nix space where thats basically the only daily driver option as far as I am aware <kitty1>imagine if nix made a bsd fork and guix worked on a seperate bsd as well for the systems, and if hurd got more resources and documentation lmao <bjc>the hurd never had much traction, and when rms went whole hog on gnu/linux branding, it lost what little it had <bjc>it's kind of nice to see it popping up again. it's a neat idea, but i seriously doubt it'll ever fly. i'd love to be wrong, though <kitty1>yeah, same. Honestly, if we are talking about what we *really* want to see <kitty1>some effort really needs to be put into just building something simple and relatively portable up from scratch akin to plan9 only taking whatever new developments have been had since then, and just really, really, focusing on making it understandable and documented <the_tubular>it's kind of nice to see it popping up again. it's a neat idea, but i seriously doubt it'll ever fly. i'd love to be wrong, though, yeap this is also how I feel <bjc>the unix security model is deeply flawed, and bolting capability systems on to it isn't working very well <kitty1>I don't get why they didn't just build it on top of a BSD instead of linux <kitty1>oop looks like I didn't send a message there whoops <kitty1>meant to first say that I am only using linux because that is what guix supports, but I really don't want to be using linux compared to just freebsd or something <kitty1>or really any BSD, as long as it can run some IRC clients and emacs im honestly chill <kitty1>I just don't like how universal linux is <bjc>guix is a gnu project through and through. it's going to use as much gpl software as it can <kitty1>I like the gpl, and if that has that much influence on that ; thats terrible decision making. <bjc>that depends on the goal of the project <bjc>the gnu system is first and foremost about securing software freedom, and other things are sometimes sacrificed to promote that goal <kitty1>like, it shouldn't *not* be a factor ; but I think in the context of kernels linux is a huge issue, they are already dealing with things like WSL without acting, on top of that <kitty1>again, as someone who hasn't done shit in any of this <kitty1>if your going for the goals of having free software, its <bjc>it's more than just the kernel. another major component is libc, and freebsd doesn't use glibc <kitty1>kinda counter productive to go for the most bloated and least elegant system you can find, when a major part about free software should be understandability and documentation (which, I don't think is enough of a priority for a lot of people sadly) <bjc>you could make guix work on freebsd if you wanted to put the effort in, but it's just not a priority for most people <bjc>freebsd has its own libc <bjc>i'd hesitate to call linux bloated. it supports a *lot* of stuff. wayyyy more than freebsd <kitty1>when you say supports , in what level do you mean? <kitty1>like, what is it that it supports more of? <bjc>i'm not trying to be a linux booster, or anything. i've long been a fan of freebsd, and use it myself for some things, but it's not fair to linux to label it that way <bjc>like, just driver-wise. linux supports a lot of devices. linux also gets a lot more attention. freebsd is at the point of writing shims in its kernel in order to use linux drm graphics drivers <hnhx[m]>the_tubular: Ye this. I got 45~ MB of idle RAM usage once with Gentoo with Xorg and dwm running. <hnhx[m]>You can remove everything that you don't need from the Linux kernel. <bjc>virtualization works a lot better on linux than bsd, at least kvm vs bhyve <bjc>bhyve is very limited <kitty1>I have as well, I probably am talking too much for not messing around with more of these things in more depth myself; rather than having a more in depth understanding. <kitty1>bjc: also curious if you have any examples of in what way it is very limited? <bjc>you can't do gpu passthrough on bhyve. the pci passthrough can be kind of sketchy. trying to run non linux/bsd operating systems can be a trial <bjc>i think maybe gpu passthrough has been coerced into working by some people, but i don't believe it's at a place where it's considered supported yet <bjc>(i haven't checked in months, though) <bjc>it was only like 2 years ago that you could install windows in the normal graphical way. it's silly to try and compare bhyve to kvm for general purpose virtualization <the_tubular>BSD is probably something I should mess with more, but guix takes all my time :( <bjc>the thing is, in terms of software heterogeneity, guix is more important than bsd, imho <bjc>both linux and bsd look pretty similar from the userland. guix is a completely different way of organizing a system that veers wayyy off from standard unix <emacsomancer[m]>The potential separation (esp. with guix home) of base system and "userland" that's possible with guix is a bit BSDish. <kitty1>horizoninnovatio: heard about that before; from the outside it seems kinda interesting <lilyp>hmm, on the scheme side I don't see anything particularly wrong <lilyp>could there be a problem with auto-login for the hnhx user in particular? <hnhx[m]>lilyp: When I looked it up people mentioned that its an issue with pam. <hnhx[m]>however I couldnt edit the pam file since its read only on Guix : / <hnhx[m]>lilyp: do you think its a permission issue, or? <lilyp>I don't think the default PAM configuration ought to cause troubles <lilyp>there is one pam service to get gnome-keyring working, but that only applies to gdm <hnhx[m]><lilyp> "there is one pam service to..." <- I have gnome keyring installed <lilyp>no, if you don't have gdm, the service ought to do literally nothing <lilyp>though if you think it might be a problem you can delete it from your services <rekado>and now we’re at 233 connections. <rekado>this will die at 300 connections. <rekado>cuirass appears to continuously open new connections instead of actually reusing them. <efraim>I might've caused a few, restarted a couple of aarch64 builds <rekado>this shouldn’t happen, though, because cuirass is supposed to reuse existing connections in a dedicated thread. *the_tubular slaps efraim around a bit with a large trout *efraim shakes fist at trout <oriansj>efraim: just be glad it wasn't walleye <fum>Hello, are russian IPs banned from connecting to guix.gnu.org? <fum>If so, how long will this ban last? <nckx>fum: Yes, and we don't know. If anyone knows it's rekado but they're not responsible either. <fum>nckx: Ohh okay, thanks! <hnhx[m]>btw does Guix have mirrors running as a tor or i2p hidden service? <hnhx[m]>for an example void offers tor mirrors <nckx>Where would the HS descriptor be stored on the system? <nckx>and r4mdnnhdof4jvbm6bucdjdqkrc5uxcjnttixi5ltczzpernlsz6xjuyd.onion and <nckx>4zwzi66wwdaalbhgnix55ea3ab4pvvw66ll2ow53kjub6se4q2bclcyd.onion <PotentialUser-57>Is "hash guix" really necessary and what does it really do? The first time the manual tells you to update your system it tells you to use "guix pull" and then "guix sudo guix reconfigure /etc/config.scm". The second time it tells you how to update it says do "guix pull", then "guix hash", then "guix upgrade". Or maybe the first is for the base <PotentialUser-57>system and services or what you call them, and the second is for user installed packages? <cbaines>I'm guessing it says to run hash guix, rather than guix hash <cbaines>the first time you run guix pull, you'll be using the guix binary available to all users on the system <cbaines>once you've run guix pull, you probably want to switch to using that guix <PotentialUser-57>cbaines, it switches the guix program itself for the latest one you just downloaded with guix pull? <cbaines>not really, do you know how the $PATH works in bash/shells? <PotentialUser-57>it makes your terminal point to the latest guix version instead of the old one? I have a little understanding of $PATH, I've changed such paths a couple of times but I don't understand it on a deeper level <m4rk>I'm trying use PC with only free software, and I want to install Guix, but it requires linux-libre. What it is the best NVIDIA GPU for nouveau? I prefer NVIDIA over Intel by perfomance and over AMD because AMD requires blobs <PotentialUser-57>m4rk: you have to go back really far. I think like 200 series if you want it to be able to run, and like 9000 series if you want to actually be able to use your graphics cards performance. Nouveau website goes into it <PotentialUser-57>if you do a search of "nv40" and so on you can find what graphics cards are included in that <PotentialUser-57>last time i looked there was something about how all newer cards downthrottle to like 5% of performance and there is no way to stop it. I dunno if that has been improved or not, maybe it has <m4rk>kepler series isn't best for nouveau? 780 ti or titan x black I think. I asked here because I don't know where ask: #nouveau there aren't answers even asking it. I wanted it for D16 motherboard (amd64 mobo with libreboot) <m4rk>currently I'm using x200 laptop for libreboot, but I'm searching a good desktop environment (D16 is good desktop environment, but I need GPU, I have a lot of screens, so, I need good GPU) <PotentialUser-57>m4rk yeah my information might be outdated. but be aware of the downthrottle issue and do some research on it before you buy a graphics card. if you actually need the gpu for 3d rendering or gaming or whatnot <m4rk>I want the best GPU what works with only free software (except vBIOS, it doesn't look possible overwriting it with free software yet) <PotentialUser-57>yeah ure probably going to have to use whatever came on it from the factory <m4rk>I knew about downthrottle. GPU kepler series have this problem with nouveau, but I prefer that respect to propietary driver (NVIDIA driver) or blobs (AMD driver) <PotentialUser-57>if you're fine with the lack of peformance then kepler is a good choice probably <PotentialUser-57>if i remember correctly, kepler is indeed the last generation of cards that require neither a nonfree driver nor nonfree firmware <maximed>sneek: later tell f1refly: Add #:tests? #false to 'arguments'. <f1refly>I already figured it out eventually :) <sneek>f1refly, you have 1 message! <sneek>f1refly, maximed says: Add #:tests? #false to 'arguments'. <stfnbms>Guix upgrade gets stuck in the 'check' phase for emacs-deferred (not manually selected, presumably a dependency of the new Ekacs 28.1 package that I am trying to upgrade to). What to do? <efraim>actually stuck or just taking a while? How long has it been? <efraim>looks like no substitutes for emacs-deferred yet <lilyp>stfnbms: check upstream whether there's an emacs-28 specific bug <lilyp>if there's a deadlock or something and someone already submitted a pull request, adding that patch to the package ought to be simple <stfnbms>efraim: An hour or two, the little spinner on the left stopped spinning, and ther is no CPU load. Actually stuck. <efraim>ah, so looks like it's actually stuck and in need of some fixing up <stfnbms>Yes, I could wait for a substitute to become available. Just so tantalizing to finally switch to 28.1! <stfnbms>Does anybody know if native-comp is enabled in the regular emacs 28.1 package? There does not seem to be a separate native-comp package. <unmatched-paren>stfnbms: i don't think so, i saw lilyp saying no to this exact question earlier i think? <stfnbms>Ah, okay. Well, regular 28.1 will be exciting enough for now, and hopefully native-comp follows at some point. <mekeor[m]><jlicht> "hey guix! Long time no see" <- hi :) fun fact: did you know that "long time no see" is a word-by-word translation from mandarin/chinese? *littlebobeep just learned that Longsoon is making new MIPS64-based chips and upstreaming support in Linux for them *littlebobeep also remember's when rms's Longsoon netbook was stolen at some conference or university <littlebobeep>Etymology unknown. Attested US 1901, presented as pidgin English by a Native American. Possibly a calque of Cantonese 好耐冇見, comparable to no can do or chop-chop – if so, most likely US Chinatown origin, alternatively British Far East such as Hong Kong. Alternatively, native American origin, or native coinage as pidgin, particularly in cinematic portrayals of native Americans; compare language <lilyp>any news w.r.t. emacs-deferred? <mekeor[m]>a little off-topic, but talking about rms, i think it's very bad that he came back to fsf because he has very bad opinions on sex and sexualized violence (on children). i like copyleft and all but rms being one of the leaders of gnu and guix being part of this organization is a downside of guix for me <patched[m]>At this point I think any discussion about rms' personal views acts as a distraction and splinters our movement. rms is not gnu. <lilyp>As patched[m] says, GNU is an ideological movement, not a personality cult. <PotentialUser-57>RMS being part of FSF is a big upside to me, I don't think there is anyone else as stedfast as him, I don't think there is someone else that could fill his shoes <littlebobeep>mekeor[m]: rms wrote later they think sex with children is wrong you can read the website <littlebobeep>as far as GNU being an ideological movement, rms makes an absolutely excellent part of it, having founded the project and inventing copyleft and setting the standard for software freedoms that still matter today <PotentialUser-96>hi! is there a way to "use" a tree of modules? For example (use-module-tree '((guix (gexp utils)) (gnu (man texinfo)))) <PotentialUser-57>Although I think his ideas about sex and covid and vaccines are utterly insane I dont think its relevant to his advocacy for free software at all <mekeor[m]>patched: well rms is one of the leaders of fsf, which hold the copyright of all gnu code <jlicht>mekeor[m]: Not true in a general sense: I did not assign any of my guix-related copyright to neither GNU nor FSF/RMS <littlebobeep>mekeor[m]: not all GNU code actually, Guix code is not FSF-assigned, but Guile is for example <lilyp>PotentialUser-96: the (gnu) module has shorthands such as use-package-modules and use-service-modules <littlebobeep>mekeor[m]: have you looked at the source code of Guix and Guile? <mekeor[m]>littlebobeep: to this: "rms wrote later they think sex with children is wrong you can read the website" :) <mekeor[m]>i wasn't aware that gnu code doesn't necessarily have to be assigned to fsf, thank y'all :) <efraim>mekeor[m], PotentialUser-57 and potentially others: Please move discussions of rms and/or the fsf to guix-offtopic <mekeor[m]>and now that you told me, i remember that i also didn't assign my contributions to fsf :D <PotentialUser-57>mekeor[m] he changed his views on pedophilia many years ago, retracted what he had said, long before the corporate media tried to cancel him <mekeor[m]>actually i explained why this is not so off-topic for me <mekeor[m]>anyway, you can send me a link, and otherwise i don't need to continue this discussion <lilyp>"no off-topic for me" could still very well be off-topic to the rest of guix <nckx>Copyright is on topic, corporate media cancel conspiracies are not. <nckx>Luckily there's #guix-offtopic for civil chats. <PotentialUser-57>i already said ok, im not bringing it up further but then you spoke in a derogatory fashion to me for no reason <nckx>You imagined that. I didn't use either suffix. <jlicht>now for a technical q: the sources for our java-slf4j-api package seem to have been deleted (and that entire way of getting sources removed entirely): something similar (but not identical) can be downloaded from their git-repo; what do? <nckx>I was talking to mekeor[m]. <efraim>jlicht: any help from archive.org? or fossies? <jlicht>efraim: That's my (actual question), what do we usually do in these cases? I've so far been living life not having to care about such things :). <nckx>jlicht: In practice, rely on substitutes until the package can be updated. <efraim>To keep the original we normally try archive.org. Fossies will often have the latest tarball, which works for abandonware <oriansj>PotentialUser-57: please try to interpret the statements of others in a charitable light and in good faith as miscommunication is easy to occur in an environment such as this <efraim>but yeah, otherwise the best option is to try to update to a newer version <oriansj>clarifying questions in good faith usually are quite helpful. <nckx>efraim: Indeed, I was jumping to the conclusion that it wasn't archived anywhere. <nckx>Do check why it was removed, as some projects have the (questionable) habit of removing 'vulnerable' releases. (I dunno, Is see 'java' and '4j', I get flashbacks.) <toangladius>Hey, is this where I should ask for help with the Guix OS? I'm having trouble with system configurations <maximed>toangladius: Yes, though the ‘standard’ name is Guix System <efraim>looks like efivar isn't properly cross compiling <jlicht>nckx: good point, but in this case it seems they removed the entire concept of release tarballs from their website <maximed>Though that's efibootmgr and not efivar <efraim>guix gc --references $(guix build efivar --target=aarch64-linux-gnu) references gcc, not gcc-for-aarch64 or whatever its called <maximed>efraim: I guess it's using 'gcc' instead of 'TARGET-gcc' <efraim>I think I fixed more parts of it <efraim>now I get incompatable libraries for efivar and popt, so I'm missing a flag for efibootmgr <PotentialUser-96>mekeor[m]: do you have links with quotes that support your perception of rms's views? (sry/ I know this is off-topic, but mekeor isn't in #guix-offtopic and I can't send private messages) <efraim>ok, now I have efibootmgr and efibootmgr targeting aarch64 both building and targeting the correct libraries <toangladius>alright maximed - here's the thing. I'm trying to `guix system reconfigure`, but I end up with a build error on `webkitgtk-with-libsoup`. apparently this is a dependency of `lvci-gnome-shell` and `evolution-data-server`. I checked the log and I only see `Error .size expression for something something does not evaluate to a constant`. I'm not sure <toangladius>what to do about this, I would gladly remove both packages that depend on this but I have no idea how. <maximed>toaangladius: The build error in webkitgtk-with-libsoup is a bug (independent of guix system reconfigure) <Zelphir>I am using `guix shell` to run Emacs. I want to make a reproducible guide to getting org source blocks of Scheme code to work with GNU Guile. To do that, I would like to clear up anything created by the call to `guix shell`. I have tried using `guix gc`, but that does not seem to clean up everything. What command should I use to delete everything created by `guix shell ...`? <maximed>tangladius: question, what does "guix --version" output? <maximed>Did you see "webkitgtk-with-libsoup", or "webkitgtk-with-libsoup2"? <Zelphir>Ah sorry, replied to other message :D <toangladius>maximus: i'm running in a wm so I can't easily copy but I'll transcribe the relevant parts, first gets me `guix (GNU Guix) 0, copyright 2022` and then license, the other `/home/me/.config/guix/current/bin/guix` <maximed>Zelphir: you can leave "guix shell" session with control-D or restarting the terminal or whatever you use <maximed>toangladius: ok looks reasonable (it eliminates the ‘forgot to guix pull after an install’ potential cause) <maximed>toangladius: Could you run "guix build webkitgtk-with-libsoup2 --no-grafts", it will probably fail again, in the last few lines there should be a ‘log file at /var/log/...SOMEWHERE.gz’ message or so. <maximed>Make a copy of that log file and send it to bug-guix@gnu.org. <Zelphir>maximed: The command I run specifically already exits the shell, when Emacs is closed: `guix time-machine --channels="channels.scm" -- shell --cores=8 --check --pure --manifest="manifest.scm" -- emacs --no-init-file --no-site-file --no-site-lisp --no-splash --no-x-resources` But when I run it anew, I still see the same Emacs packages installed, inside Emacs (`list-pack`), that I installed manually. It is somehow not cleaned up. Perhaps the <Zelphir>packages are put in a location, where guix does not remove things? <toangladius>maximed: running it now. Either way, I'm assuming both `lvci-gnome-shell` and `evolution-data-server` are somewhere in either `%base-packages` or `%desktop-services`, do I just filter those or is there a more idiomatic way to go about things? I'll have to do this eventually anyway because if I'm planning to replace sudo with opendoas <maximed>Zelphir: WDYM with ‘installed manually’? <maximed>Do you mean with Emacs' interface to elpa or such? <Zelphir>maximed: Ah, sorry, I mean `M-x package-install RET geiser-guile RET` inside the Emacs that I run with the command mentioned above. <maximed>Zelphir: that's Emacs builtin-in package manager I guess, because "guix show geiser-guile" does not give any results. <Zelphir>Oh OK. I guess I will have to delete them somewhere, where Emacs puts them then. <maximed>To remove packages installed with Emacs built-in package manager, you probably have to use Emacs built-in package manager. <maximed>If you want to be able to temporarily install emacs packages with "guix shell", you'll have to package them and put them in the manifest.scm. <maximed>toangladius: I don't expect evolution-data-server to appear there but I don't actually know. <maximed>Unless you're using a gnome desktop environment maybe? <toangladius>maximed: no, that's the thing, I'm running a default ratpoison installation <maximed>You can have both on a single system (and then choose which one during login) <mekeor[m]>Zelphir: maybe you can start "guix shell --container --no-cwd emacs -- emacs" or so, so that emacs won't really run in your home-directory and won't mutate it <maximed>I don't know what's pulling in evolution. <maximed>Could I have a look at the system configuraiton? <toangladius>sure, but I'll have to copy it somehow, give me a sec <Zelphir>maximed: Yes, I would usually install the Emacs packages via installing guix packages in the manifest. However, in this case I want to exclude the possibility, that any bug might only be inside the guix package of an Emacs package (guile-geiser or geiser), to be able to report the bug to the author, for the original package. I feel installing the package via guix as well will move further from the author's usual setting, reducing the chances to <maximed>preferably not pastebin, it's full of ads <maximed>(though there are also other nice pastas out there) <bost>Is there some indication that there's a new linux kernel available (after `guix pull`) and I should run `guix system reconfigure` to activate it? <bost>IOW is there some better way than running `ls -la /gnu/store/*-linux-libre-*/bzImage` and comparing it with `uname -r` to get this information? <toangladius>maximed: sorry it took a while, pastebin is a mess, i don't think paste.debian has a simple api and paste.centos didin't work for some reason <nckx>bost: guix pull has the option to list new packages, including linux-libre. The ls you quote looks flawed to me: pulling alone won't build any kernels, so they won't appear in the store like that. <nckx>Is there a single command to get a yes/no answer: not that I know. You could likely script something 'good enough' ('does uname -r appear in guix package -I?' or so). <bost>nckx: That ls command shows me 11 bzImage files. And I don't remember building any kernel. What else than the `guix pull` builds the kernel? Hmm. *mekeor[m] runs 'guix pull --news -n -l' to see news of all past guix-pulls <PotentialUser-57>do you know of any web browsers that can save as mhtml, without plugins? usually i use falkon but it's not on guix yet <nckx>Epiphany (Gnome Web) should support in but I didn't test, PotentialUser-57. <PotentialUser-57>mekeor[m] thanks but preferably without plugins, as the link says its not actively monitored by mozilla and they advice you not to install it without trusting the developer <Zelphir>How would I install a shell for usage inside a guix container? Simply add for example `sh` or `bash` to my `manifest.scm`? (geiser-guile or maybe geiser in general need a shell to run source blocks). <mekeor[m]>PotentialUser-57: there might be other addons, too, i didn't check <nckx>TIL .mht(ml) == .eml, huh. <maximed>Though maybe you can do "sudo dmesg" first to see if it's caused by out-of-memory <Zelphir>adding `bash` in my `manifest.scm` worked : ) <PotentialUser-57>nckx that's very interesting. I couldnt find a way to save mhtml with gnome web but it can save to pdf which also suits my purposes (y) <toangladius>In the meantime I checked `%desktop-services` and there's a lot of gnome bloat in there, it's probably some of those things that it's building <nckx>Sorry to misinfo you but glad you found a workaround (though AFAIK FF supports PDF just as well through the print dialogue, or used to). <nckx>On second thought, that probably changes to 'print CSS' which might not be what you want. <Zelphir>Thanks for all the help! Learned a few new things about running things in a container. I hope this makes a good report and maybe I can get my literate programming in Guile in Emacs working again. <xelxebar>unmatched-paren: Unfortunately the debug output doesn't exist on very many packages. <xelxebar>You'll have to run `guix build --with-debug-info=... ...` <luishgh>hi guix, i was browsing the commit history and found "5f315e1d5f * gnu: emacs: Add support for socket activation.". i searched through the discussion on issues.guix.gnu.org but did't find an explanation of what this socket activation is. Could anyone explain it? Just asking for the sake of curiosity :D <unmatched-paren>hmm, apparently glibc:debug is already included in gcc-toolchain anyway... <luishgh>hmm, interesting. so with this new patch, we could activate the emacs-daemon through shepherd? <luishgh>i tried that in the past, but the problem was that it would use the emacs from the system profile instead of my custom profile that i activate through .bash_profile <shoshin>anyone around who can help me figure out how to use guix to install a new package to Emacs while it its running? everything works fine if i reboot emacs after installing the package, but i wasn't able to load it while Emacs was running <shoshin>like the load-path isn't getting updated <luishgh>i think using this function, not command, can help `guix-emacs--non-core-load-path' <shoshin>i don't seem to have that one, maybe i need to update guix.el <shoshin>yes i did. i'm only seeing two functions starting with guix-emacs <luishgh>try using `guix-emacs-autoload-packages' instead then. it should work as well i think <shoshin>i think the issue i'm having is that the newly installed package isn't in the emacs load path, so that `guix-emacs-autoload-packages' doesn't pick it up <luishgh>i don't know how EMACSLOADPATH could be updated <dirtcastle>I gave the target for bootloader as (target '("/dev/sda4")) <dirtcastle> I gave targets but looks like I need ' before (targets <dirtcastle>I gave targets not target. idk what the problem is unmatched-paren <bjc>bootloader sections should look like: (bootloader (bootloader-configuration (bootloader …) (targets '("/dev/sda4")))) <bjc>there's no quote before ‘targets’, and the quote before the device list is only being used to avoid typing more. in this case, '("/dev/sda4") is equivalent to (list "/dev/sda4") <unmatched-paren>dirtcastle: oh, right, i see. you mistyped here: "I gave the target for bootloader as (target '("/dev/sda4"))" <unmatched-paren>dirtcastle: not sure what the problem is; (targets '(...)) seems to be fine :) <nckx>And mount it relative to the new root, e.g., /mnt/boot/efi. <nckx>Eh, no EFI here but same diff. <PotentialUser-57>is there any reason why an installed application would not create a shortcut <roptat>not all of them, but we could improve the situation <roptat>there's a bit of code to create .desktop shortcuts in packages, but we have to write it. If you could take note of which packages need one and send a bug report, it would be great :) <dirtcastle>unmatched I mounted but the problem seems to be an syntax error or smtg. should I give /mnt/guix/boot/efi or /boot/efi or dev/sda5? <roptat>I mean, we have to add that bit of code to all packages that need a .desktop and don't provide one already <pashencija[m]>That's why MBR is created incorrectly for partitions. The code is just not implemented. <PotentialUser-57>unmatched-paren in this game it was a game called teeworlds that didn't populate my gnome menu ***ajarara is now known as Guest1814
<mekeor[m]>pashencija: what exactly is not implemented? <PotentialUser-57>Does anyone know why I keep getting the notice about GUIX_PROFILE environment thing? The manual says this should only appear if you install guix on another distro, but I'm using a full guix OS install and it shows up for me ***Xenguy_ is now known as Xenguy
<PotentialUser-57>I noticed gpg isn't installed and did a search for it. Plenty of packages. Which one is just regular gpg? gnupg? <pashencija[m]><mekeor[m]> "pashencija: what exactly is..." <- MBR entries according to file system <lilyp>PotentialUser-57: it shows up whenever you install something that causes an environment variable to change <lilyp>after a while there should not be any new packages changing it bar some special exceptions (emacs for instace did that before we patched it not to) <lilyp>well, "just ignoring it" might work, but if you open up an app from the shell and it doesn't find some important file, do remember to run the snippet that's reported <lilyp>on Guix System you don't need to do so manually after having logged out and back in, though, which is nice <bdju>I wish manifests supported partial upgrades and that I could blacklist packages that take an insane amount of time to build like qtwebengine and only pull them in when there are substitutes <bdju>how can I find out which package is making qtwebengine build? just skipping qtwebengine itself didn't work <bdju>my updates ran all night and it was only 60% done building qtwebengine so I canceled it <blake2b>just setup a remote build farm with cuirass and guix publish. gotta say, its one of the most straightforward/easy guix topics I've embarked on. imagined it would be more difficult tbh; thought I'd report because I remember during fossdem it sounded like a lot of folks haven't dug into it on their own yet <jackhill>blake2b: that's great to hear! I think making such things approachable is important to our mission <blake2b>jackhill: yeah I might do a cookbook entry on setting it up on linode if people would be interested in that. <blake2b>its pretty amazing actually, you get mumi and the whole package and it only took me like an hour <blake2b>but if i had a short recipe it could take 15 minutes <bdju>would rather not pay for a vps to solve the issue <bdju>speaking of issues, getting a backtrace as of a recent update when trying to do a guix system reconfigure. it says a dir in guix/checkouts is not owned by current user. did a recursive chown and nothing changed. also it looked like it was all owned by my user <euandreh>I'm having to source Bash completion files manually for packages installed via guix home, i.e. ". $HOME_ENVIRONMENT/profile/etc/bash_completion.d/*". Shouldn't this be dealt with by guix home somewhere? <bdju>can/should I just deleted the guix checkout that "is not ownd by current user"? <bdju>oh dear. guix --version show "0" instead of the usual hash <Kuschelyagi>also I believe since we're using sudo "current user" in this context might mean root, but chowning this directory to root might break some other thing so I'm hesitant to try that <ulfvonbelow>you can work around it for the moment using 'git config --global --add safe.directory /home/<you>/.cache/guix/checkouts/<thecheckout>' <bdju>oh, I was about to send in a bug report. should I not then? if it's a known issue I mean <KarlJoad`>There has to be a command for this, but I am not finding it. The command is given an output path in the store, and it returns the package that provides that output. <bdju>ulfvonbelow: thank you for the command, it seems to be working now <jackhill>ah, removing ~/.cache/guix fixes it, but I wonder how it got messed up 