<gabber>has anyone created a `emacs --bg-daemon` service yet? <bjc>i have an fg version <podiki[m]>I have yet to turn all my guix config into one literate org-mode file, as I've done for a few dotfiles <gabber>bjc: would you mind to share that? i've tried a couple of different configurations (bg-daemon, fg-daemon) yesterday but failed.. i'm new to writing my own services <bjc>then in your config: (service home-shepherd-service-type (home-shepherd-configuration (services (list emacs-shepherd-service)))) <gabber>is it possible the definition of emacs-service-type is missing? <bjc>i don't define emacs-service-type *gabber wonders why this isn't part of the source tree <bjc>it seems like it should be; feel free to submit a patch <bjc>i don't feel like i have the bandwidth to do it myself rn <gabber>(how) would you want your authorship dignified? <phodina[m]>Any idea which package provides KF5::ConfigQml? Tried kconfig, but the cmake still complains <bjc>sneek: later tell gabber: no, i don't need credit. also, you'll probably need to make changes to it anyway <dcunit3d>i understand that in guix i shouldn't make changes to /etc or other system directories... if i just generated an ld.so.cache, will this cause problems? <dcunit3d>i'm running into Julia package compilation issues and i'm trying to pin down what guix packages i'm missing. <jmes>Hey folks, I wanted to try Gnome+Wayland but I'm struggling to get just a bare-minimum setup. I am too dumb to deciper this backtrace (also includes the services in my config) (https://bpa.st/BGFQ). Any help would be appreciated. <benjamin1>any chance of running guix with an encrypted boot partition? <benjamin1>I set up my laptop with one a long time ago, before I realized how much of a pain it was going to be <benjamin1>plan was to have guix manage the grub config for booting both systems (potentially with chainloading?) <jackhill>jmes: perhaps try changing cons* to list. cons* expects a list as the final argument. For the services field this is usually something like %desktop-services or %default-services <jackhill>jmes: you also might have a better gnome experience using gdm over sddm since I think gnome has some integration there. Its default is X-only for now, but there is a wayland? field. <jackhill>benjamin1: I'm runing a Guix system with an encrypted boot (well all of mine /boot is on the same encrpted volume as the rest of /, which might matter by default the grub config loads the kernel and initramfs from /gnu/store and not /boot). I believe that it is still the case that our grub can only unlock LUKS-1 volumes and not LUCKS-2 <jackhill>The good news is, that I haven't found it to be too much of a pain! The gotcha is that I have to enter my password twice, once for grub, and a second time for linux. <jmes>jackhill: Oh whoops, that should be the silly mistake, thank you! And yes, I've been successfully using the default X+gnome via gdm but I need to test some stuff on Wayland. <jmes>I'd like to use Xmonad as I do on my other machines but I haven't gotten around to learning guix enough to get that working. <benjamin1>jackhill: thanks! very glad to know that this is possible. currently my whole disk is LUKS-1, so that shouldn't be an issue <jackhill>jmes and benjamin1 cool, glad I could help <benjamin1>do you have an example 'bootloader-configuration' I could take a look at? *jmes appreciates jackhill ***LispyLights is now known as Aurora_v_kosmose
<benjamin1>'guix build nss-certs' broken for anybody else? <benjamin1>> error: corrupt input while restoring archive from #<closed: file 7f5d705dbe70> <benjamin1>this seems like it might be an issue with the substitute server, but I'm really not sure <benjamin1>hmm, might actually be a locale issue on my end <benjamin1>it's failing on a cert that has a nonascii filename I think <dcunit3d_>i'm running into issues executing binaries external to guix. i think it's because they don't inherit the environment variables from the profiles. <Aurora_v_kosmose>It might be useful to source the guix profile's /etc/profile into your shell prior to calling the programs. <dcunit3d_>in other words, there's a ELF binary and i try to run it from bash. when i do, i get "bash: .bin/elf-bin.file: No such file or directory" <dcunit3d_>i do, but those variables aren't passed to the spawned processes <dcunit3d_>also, when i try `exec ...` anything, the shell crashes <dcunit3d_>it's just really hard to google. i tried strace and no dice. one line and it just can't open the file. it happened with another program the other day. i tried to connect with it with GDB just to see the same thing pretty much. <dcunit3d>what do you mean? i've got an old textbook open and i'm trying to clarify some things regarding processes/exec/spawn/fork <Aurora_v_kosmose>Oh. I had been led to believe by the quit message that you had just left. <dcunit3d>i accidentally opened a second IRC context <Aurora_v_kosmose>Ah, well, sorry about that. In any case, I'm going to go afk for a while, so you'll have to see if someone else knows. In the event that it takes a while, I would suggest also making a post in the guix-users mailing list. <Aurora_v_kosmose>Right, on that last note, seems like help-guix rather than guix-users is the naming scheme that was used. <phodina[m]><jackhill> "The good news is, that I haven't..." <- I also have encrypted rootfs and boot. Is there a plan to store the secret somewhere so it does not have to be typed twice as the other distros already have this? Either by storing the password or using a keyfile? <dcunit3d>if you know how that works, i have some files i need to recover lol <dcunit3d>i have a bus pirate, but i guess that's a bit off topic. ping me if you know <maximed>unmatched-paren: I'm not seeing a cycle in the Cargo.toml (in the git repo?) <maximed>-raw doesn't depend on -macros or tinystr itself <maximed>in that case, it's probably a test-only dependency. <maximed>Currently, cargo-build-system somehow sorts out the cycle automatically (if thins are put in #:cargo-inputs / #:cargo-development-inputs) <maximed>unmatched-paren: FWIW, sometimes the version requirements are overly picking. <maximed>So doing some substitutions to the Cargo.toml can be sufficient <unmatched-paren>Although the dependency on tinystr-macros in tinystr is optional, so I guess I'll create a bootstrap package that substitutes out that tinystr-macros dependency <nckx>dcunit3d: It's probably because âexternalâ binaries hard-code the interpreter needed to run them, which is the actualy file that's not found (/lib/ld-linux-x86-64.so.2 or equivalent). If so, no environment variable will save you, you'll have to patch the binary with a tool called patchelf (in Guix). <dcunit3d>i'm working on getting that set up in my environment, someone from nonguix helped me out, but thanks! <dcunit3d>i'm learning a lot more about linux than when i was using arch <maximed>Today I did "guix pull --rollback" / "guix pull --switch-generation=..." for the first time. It was pretty useful. <roptat>unmatched-paren, you could use pk and look at the logs <nckx>(pk whatever) ; prints and returns whatever. <nckx>unmatched-paren: Your fennel is almost certainly printing nothing, I can't imagine the display not working⊠<nckx>You can verify that with (display (pk output) file). <roptat>unmatched-paren, you can also put that in a file and use "guix build -f ..." to quickly try things <roptat>depending on what %conf-dir is, fennel might not be able to access it from the build chroot <roptat>when building a derivation, you can only access a clean /tmp and the dependencies from /gnu/store <nckx>You need to run fennel on the host side, or carry your nvim.fnl over to the build directory. <unmatched-paren>how would I turn (local-file blahblah) into a store file path inside the gexp? #$file doesn't work <roptat>#$(local-file blahblah) should work <nckx>local-file takes a string, not a gexp. <unmatched-paren>I'm using (config-file "nvim.fnl"), which is defined as equivalent to (config-file (string-append (dirname (current-filename)) "/" "nvim.fnl")) <unmatched-paren>and I want to pass the store path as an argument to (string-append #$(file-append fennel "/bin/fennel") " --compile " FILENAME) <nckx>What is âconfig-fileâ? <roptat>unmatched-paren, how does it fail? <nckx>Sharing the entire file would help too. <nckx>unmatched-paren: Thanks. <roptat>/gnu/store/zb5wpl6a65cjajblan3wx78ywzv is probably the builder. You can have a look inside to see that the local-file got expanded correctly <roptat>mh, I think it's an issue with the return value of the gexp in mixed-text-file <roptat>mixed-text-file expects a list of strings, appends them and puts them into the file <unmatched-paren>Ah, I was using computed-file before, but changed it while I was trying to fix the problem. <nckx>qtwebkit's âc++â command lines are so long that I need to zoom out twice to see one â[percentage%] BuildingâŠâ at the top of my full-screen terminal window. <nckx>unmatched-paren: Nice. Looks cleaner. <silicius[m]>I wonder how to get the guix home's profile to show in the output of guix package --list-profiles <jmes>I reconfigured my system into a state with no networking :P Now I can't update substitutes when trying to reconfigure it again. Do I have to rollback or is there another way out? <jmes>Oh, rolling back requires network too? <jetomit>jmes: you could configure the network manually, but that might take longer than just rebooting into a working configuration <nckx>jmes: Can't you âjustâ select an older generation from the GRUB boot menu? <nckx>Then roll back or forwards from there. <jmes>jetomit & nckx: Yeah I think that's the fastest way. I was trying to roll back without network just now <jetomit>with a wired connection you can probably just `dhclient <interface>`, wireless is a bit more involved without a running networkmanager <jmes>Yeah it's a troublesome laptop with no ethernet port. I should really buy a usb to ethernet adapter <jmes>Hm, I just tried booting into the previous 2 derivations but they both get stuck at the TTY - I can't even switch to another TTY. It would normally load GDM at this point. Both of these versions were working fine before. <jmes>There is some output past the login prompt: localhost elogind[347]: elogind is already running as PID 336 <jmes>and typing credentials does not work, no <jmes>I'm not sure my keyboard input is doing anything (the CAPSLOCK key has an LED light but is not toggling on press) <jmes>though it worked in the grub menu <nckx>The elogind message sounds like a Shepherd bug that was recently fixed. Which is of no use to you, with an older Guix, of course. <jmes>and for typing LUKS decryption phrase <jmes>nckx: darn, so it is a bug that somehow outwits booting into a previous version <jmes>I'll check it out, thanks <nckx>Which implies that a few reboots might âwinâ the race, but eh, fun :-/ <jmes>Hah, I'll give that a couple of goes <jmes>I just need to reconfigure once successfully so I get the bugfix :P <nckx>I don't want to get your hopes up but you *might* get lucky, who knows. <nckx>Otherwise we'll have to figure out how to get you networked after all. <jmes>Also hopefully I don't have to rebuild the kernel again. I dunno how system reconfigure works, but I am very much hoping I can just add the networking to the latest derivation without wasting more cycles. <jmes>Yeah, I don't mind a manually network config. But of course I'd rather avoid it <jmes>Haha, I was able to log in by racing the elogind message at the TTY login. But once it showed up it eats all my input. Manual configuration it is. <nckx>Oh, I have it in my .guile. <nckx>You can replace pretty-print â no brackets â with (@ (ice-9 pretty-print) pretty-print) â brackets â or leave it out entirely, but it's hard to read then. <nckx>guix repl <<<'((@ (ice-9 pretty-print) pretty-print) ((@ (guix store roots) gc-roots)))' <nckx>Short & quick debug hint turned out to be neither. <nckx>I guess they don't contain a /manifest at the top level? *nckx really doesn't want to run âguix homeâ and deal with the fall-out, builds a VM. <nckx>It also doesn't create links in /var. <roptat>ah right, I didn't follow the issue, I just came back <nckx>I know there's a /profile/manifest in the âguix home buildâ output, but does it get symlinked to /manifest or /profile/manifest, is the question. <nckx>roptat: No problem, maybe that answer is obvious to someone familiar with guix home as well. *nckx is eyeing (@ (guix profiles) generation-profile). <phodina[m]>Does anybody now how to enable widgets in qtwebengine package? I get errors from cmake saying Qt5WebEngineWidgets are not provided. <silicius[m]>I noticed today that xscreensaver is both outdated by a major release and lacks a dependency which causes it to not compile one of its binaries <silicius[m]>I packaged gdk-pixbuf-xlib because it wasn't in guix at all and it works <silicius[m]>Though now if I wanted to contribute it, should I send two separate patches, or a patchset? It's both a new lib being added and a package being updated to use that lib <jeandudey>Hi guys! I've made a patch for `ppp' fixing cross compilation, e.g.: making an installation image from x86-64 for aarch64 is broken right now. #56151 tries to fix it <nckx>Open a new bug & send both patches there, silicius[m]. <nckx>& thanks to all you patchin' folk. <silicius[m]>So it would be two emails, one sent to guix-patches [patch 1/2] and the other to NNNN [patch 2/2], right? <nckx>Yes. If you use git send-email, you'd use --to=nnn@debbugs.gnu.org -2 to send the two most recent patches, for example. <silicius[m]>thanks. I'm gonna sort out my email client situation first because doing it through roundcube last time was not pleasant <unmatched-paren>silicius[m]: would you mind testing my aerc patchset, if you're looking for an email client? would be appreciated :) <silicius[m]>unmatched-paren: I was planning to use emacs for email, probably with notmuch <unmatched-paren>Since aerc is a vi-like email client, you probably wouldn't be interested in that case :) <phodina[m]>Does anybody know how to build this? It's missing the QtWebenginewidgets pkg <phodina[m]>And when I check the pkg qtwebengine is in the inputs. There's no magical switch to enable widgets <phodina[m]>Tried to substitute the name to just qtwebengine which also didn't work <nckx>unmatched-paren, silicius[m]: I think there's a poorly-documented way to get all (say) v3 patches at once in mbox format, but it's poorly documented. <rekado>what does âbroken in curlâ mean? <rekado>applying these patches should work by piping to âgit am --patch-format=mboxâ or something like that. <nckx>You could try it, but I'll spoil the surprise by saying the number 500. <rekado>I did try it and I donât get 500 <nckx>Same with -LO, it's not the HEAD. <nckx>Same if you actually paste the actual URL I used above after copying it, because copy & past is a hard problem and it's normal that it's not 100% reliable. <rekado>guess Iâm using curl so wrong that itâs right again. <nckx>For my sanity I wish you were. <nckx>I guess I'm being geoblocked for my edgy opinions on Wayland clipboards. Probably fair. <nckx>(Is there a typo I'm stubbornly missing?) <nckx>unmatched-paren: âEveryone likes surprises, pasting should never be boring and predictableâ. <nckx>I dunno who's to blame (Sway? Something else?), but once in a while I select something and it doesn't go to the primary selection. Sometimes it does. <nckx>It is not consistently based on client application if that's what you think. That would make some sense. <nckx>Whooptidoo. I bet you can reliably curl things, too. I clearly stepped on a grave this week. <nckx>Offended a wizard at Lidl. *unmatched-paren usually just uses wget but maybe this is something that can't be done with wget? <nckx>curl is generally shorter and more featureful for fetching (not mirroring), so I use it. No holy wars. <nckx>Sure, but I got it with -LO too. <nckx>-LI is⊠not good, and a bug, but understandable if Mumi doesn't handle HTTP HEADs reliably. <nckx>* Mishandling -LI is⊠not good. *unmatched-paren opens curl(1) to see what -LI does <nckx>With -LO I get an empty file. <nckx>unmatched-paren: -L follow redirects; -I perform a HEAD (headers-only) fetch. <nckx>It's a great debugging tool until a broken server decides to handle HEADs differently and usually incorrectly. âș <nckx>There's an option to GET with headers but I forgot what it is. <nckx>It was user error, kind of (I'm not sure â rekado?): /1 only works if there's more than one /patch-set/⊠I think? <rekado>yes, mumi doesnât have a fallback handler for requests it doesnât know about. <Aurora_v_kosmose>dcunit3d_: For just file recovery, you might be better off just booting the system with a livecd/liveusb, decrypting the drive using the device mapper and mounting that to copy onto another system or portable storage device. <nckx>Why does /0 or /1 (I assume it's 1-indexed but tried both just in case) not always work. <rekado>Iâm slow; Iâm still talking about HEAD <rekado>re /0 or /1: it depends on the message subject <rekado>itâs just naively matching on the [PATCH] or [PATCH v123] subject <rekado>so for /1 to work thereâd have to be a [PATCH v1] subject <rekado>this thing isnât documented, because it was nothing more than a silly little hack that I hoped someone would shape into a real feature <rekado>would be nice to make it consistent and remove surprise <rekado>but that would be against my work ethos. Apparently. <nckx>My bad, really. I shouldn't have treated it as a feature. <nckx>If âyeah, it sometimes worksâ is expected, I guess it's not much of a bug report to point it out. <lilyp>speaking about mumi is there a way we can get merge'd bugs actually merged? <lilyp>If I send a "merge NNNN MMMM" to mumi, it should show both NNNN and MMMM under the merged IDs <lilyp>reason being that people still end up creating 20 bug numbers in sequence :P <rekado>it currently shows whatever debbugs says the current bug is merged with. <lilyp>oh, that tiny details field? <phodina[m]><phodina[m]> "Does anybody know how to build..." <- nckx: Any insightful tip? Last time it solved the problem :-) <silicius[m]>hmm. The last patch version (v3) I sent received to replies for a few days, but I also don't see the new package in guix yet. What should I expect? <bjc>it can take some time. i've had a minor patch waiting for months <bjc>mostly, though, it's not more than a couple weeks ***caleb__ is now known as KE0VVT
<silicius[m]>I expected someone would yell at me for the description being just one long line - either a quirk of the website or my client sent a different email than it showed me <lilyp>a few things: your indentation is off, as is the description formatting <lilyp>your version comment also does not put up the air of professionalism, you might want to reword it a little <lilyp>As far as I can see, it is not seen as one long line, but as several lines which are longer than fill-column. <lilyp>maybe it was, maybe it wasn't, I can't tell <silicius[m]>I see I'll have to send the patch again, the proper way. <nckx>Ere I bugger off a report: is the âpâ key in htop@3.2.1 broken for anyone else? <nckx>silicius[m]: I think the version comment's fine đ€· Maybe "release" â release, though (it's a release or it isn't, no?) <silicius[m]>I forgot why exactly I put that in. I'll make sure to correct that <maximed>nckx: What's the p key supposed to do? <nckx>Ah, so it works for you? <maximed>cron -f <-> /usr/sbin/cron -f (on a Debian system) <nckx>It's more useful (and was added for) Nix/Guix :) <maximed>wait that's the foreign distro's htop <nckx>OK, thanks for confirming. <nckx>They've been doing some heavy well-intentioned refactoring and probably broke it. <maximed>so somewhere in between 3.0.5 an 3.2.1 <nckx>Surprised I didn't notice this sooner (or when updating), I use it a lot. <rekado>lilyp: yes, that little details thing. <rekado>lilyp: Iâd be happy to apply patches to make mumi more useful or prettier or both. <unmatched-paren>maximed: When do you think you'll be able to merge antioxidant, assuming you hit no significant hurdles? I presume it'd be merged into core-updates, right? <maximed>unmatched-paren: librsvg is IIRC core-updates, and also rust, so antioxidant would be too <maximed>eergh it builds fine locally but fails on the build farm ... *maximed compares derivations *maximed runs "guix gc --verify=contents" <lilyp>rekado: My idea was more or less a "merged" view, where all messages of all merged bugs are shown in temporal order <lilyp>that'd be super helpful for those patches where the second message is the merge to control@debbugs <maximed>WDYT of adding -L and -m option to "guix challenge"? *the_tubular Collects all the flags <nckx>I have a completely subjective but worrying feeling that Azure RTOS GUIX is ranking higher in search results for âguixâ than it did several years ago. <nckx>Familiar, or my imagination? <the_tubular>I don't have a baseline on this nckx, and the meta seach engine I use, also indexes bing, So I guess Microsoft can also push their own programs <the_tubular>But yeah, I noticed the Microsoft guix is always up there when I google something about guix <the_tubular>As I said above I wouldn't be suprised if whitelisting bing will bring M$ guix higher <lilyp>well, in DuckDuckGo it's the second entry right below the correct Guix, so you're probably right <nckx>I mostly use DDG, but I check Google once in a while just to stay hip with the kids (and make sure I'm at least aware of what trends there). <nckx>So TBC this isn't based on data, just something I feel like I noticed lately. <unmatched-paren>After disabling Google, I see the wikipedia page for Guix, then the guix.gnu.org site, then... several completely unrelated results. <nckx>I didn't track it, wish I had. <nckx>unmatched-paren: Nothing Renesas? <unmatched-paren>The difference in search result quality between Google and Bing is really striking... <nckx>It used to be theirs before MS bought it. <nckx>unmatched-paren: In which direction? <the_tubular>Let me make that piece of shit search engine your default though <nckx>In other news, _somebody_ is installing Guix into an Alpine Linux VM bridged to IRC: https://www.tobias.gr/irckvm.png This will end extremely well and reflect very well on Guix đ :-/ <nckx>On the bright side, the VM is connected over Tor so it will probably never finish. <lilyp>the_tubular: Don't you see it's #pissnet? <nckx>the_tubular: Different network, name is in the screenshot, but it's⊠a special place. Bewarned. <lilyp>lemme guess the CoC doesn't apply there? <lilyp>or even worse; no code of conduct can be made to apply <nckx>No, but not because it's Naziville central. It's a joke network. Hence why someone is installing Guix into an IRC bot. <unmatched-paren>"Network rules: (1) You must do exactly the opposite of what codes of conduct tell you to do." <nckx>Kind of? But like with literally less nazis than Libera because nobody bothers to troll pissnet? Anyway, this is going off-topic, I just wanted to share the on-topic part. *vagrantc wonders if nckx is available for a marketing gig <the_tubular>I mean maybe I could learn by watching someone configuring guix live in a irc <the_tubular>trying to edit text in an irc must be a challenge though ... <nckx>They've quieted the bot whilst it downloads the release tarball over Tor for the next 2 hours, so it's pretty quiet for now. <the_tubular>Why do you think it would bring such a bad name for a "small" community ? <nckx>the_tubular: Finally, ed's time to shine! <nckx>This is literally what it was made for, so⊠<nckx>vagrantc: I'm game for anything that pays, and most things that don't. <nckx>unmatched-paren, lilyp, the_tubular, and anyone I forgot: Thanks for sharing your SERP samples. ***rgherdt_ is now known as rgherdt
<nckx>Sorry, missed: the_tubular: <Why do you> You're probably right I care too much. But one doesn't want Guix itself to become the joke. <mbkamble>I am using "guix home" to configure my home env and launching emacs as daemon using home-emacs-service-type. I also have separated my installed pacakges into ~/.guix-extra-profiles as described in the Guix Cookbook. But it looks likes emacs daemon does not inherit the env variables from the login shell (I source the <profile>/etc/profile in <mbkamble>~/.bash_profile). But rather only those inherited from the user shepherd which seems to source only /etc/profile. As a result eshell PATH contains only the PATHs setup from /etc/profile but not the ones from my guix-extra-profiles. Can anyone provide some hints on how I can source etc/profile of my profiles inside ~/.guix-extra-profiles before <mbkamble>services from my home-environment are launched <rekado>lilyp: I donât know how to provide a single merged issue view that wouldnât feel terribly disjointed and confusing. *yewscion is back (gone for 19:16.32) <civodul>nckx: wait, you're using an IRC-to-tty bridge?... <nckx>Apologies for the confusion, I'm not respo^Wdoing any of this. <nckx>Also, the VM reconnected after being put in the background (it was âspammingâ download progress messages for >1h) and guix-install.sh had died. No way to tell why, and I might have accidentally killed the VM for now :3 <nckx>civodul: But yes, though not much of a TTY I'd say. <nckx>I think this was all Noisytoot's doing. <maximed>(I hope I) found the issue with rust-nu-command: rust-shadow-rs had irreproducible output <maximed>(patched out locally for now, upstream later) <maximed>I think someone noticed something like that some time ago>?