<slimjim>lfam: that seems to be doing the trick! <slimjim>and now failing (as expected) because --TEST and --foo aren't valid configure flags for gdb <slimjim>lfam: is there a reference for how to use substitute-keyword-arguments ? <lfam>It's like, if I knew Scheme well, I could read the implmentation, go "duh" and just use it <lfam>I basically find packages that do what I want to do and figure out how to copy them. It's not very admirable <lfam>I even have a copy of The Little Lisper <lfam>My job does not involve computer programming <slimjim>yeah that's what I was trying - but just didn't quite get it right - so THANK YOU for finding the right incantation :) <flatwhatson>it's one of the really awesome things about guix, it comes with a comprehensive set of examples! <lfam>Happy to help you stumble a way to a solution ;) <lfam>Oh I also have a copy of How To Design Programs <lfam>So maybe one day I'll work through these books <lfam>And learn enough to get a better job <slimjim>adding configure flags to a package seems like a fairly common use case - along with adding make-flags (like for hardward-specific optimizations in CFLAGS) <slimjim>I think maybe there's been some discussion about that recently on guix-devel <slimjim>seems like a nice candidate for a high-level package transformation <lfam>I guess that substitute-keyword-arguments is somewhat high-level <lfam>But there could be a replace-configure-flags, append-configure-flags, etc <lfam>Maybe I should do a coding bootcamp, I don't know <slimjim>yeah I guess there are so many variations it'd be hard to please everyone with a --with-configure-flags vs --with-extra-configure-flags vs --with-extra-make-flags or whatever <lfam>Well... we aim to please! <lfam>I love this part of the recent blog post The Big Change: "Our goal though is to empower users by giving them abstractions they can comprehend, hopefully providing a smooth path towards programming without noticing." <slimjim>minor note... are you using a guix or scheme-specific emacs package to handle indentation? When I write my definitions, emacs wants to indent the ((#:configure-flags. etc way over to the right, aligned with (package-arguments gdb) <the_tubular>lfam do you know what's happening with the installer ? <the_tubular>lfam I fixed my network problem of yesterday and i'm in a rabbit hole of errors <slimjim>oo a new blog post! Hadn't seen that one yet <lfam>I don't know the_tubular <lfam>It will help to share the errors, though <lfam>In what context do you see this message? <lfam>Did you try searching the web for this message yet? <the_tubular>I'm at the very end of the installer, after I edited the config <podiki[m]>slimjim: nckx and I were just discussing this indentation going wrong the other day <flatwhatson>specifically all the scheme-indent-function stuff fixes the problem with substitute-keyword-arguments <lfam>Usually it means that a file or directory could not be deleted because it is still in use <flatwhatson>well... (put 'substitute-keyword-arguments 'scheme-indent-function 1) fixes that one, but there are a bunch more in there <slimjim>flatwhatson: oo thanks, I just need to copy that into my custom packages directory <podiki[m]>flatwhatson: has the indentation been working for you, no change recently? <flatwhatson>as usual with emacs there can be many moving parts :) <podiki[m]>was just trying loading a guix file and finding a #tests? #f line, pressing enter after, it indents wrong <podiki[m]>but maybe that's cause of the #f? (rather than an sexp) <the_tubular>I tested both 'latest' that I downloaded yesterday and stable, and both are not working <Kolev>Maybe I need to install Fedora again... Wait, i can't write ISOs because my computers are down. <Kolev>Oh! I can use a live system! <nckx>slimjim: What's the last package expression that failed to work? I haven't actually looked at it yet. <podiki[m]>sounded like the lack of configure flags to modify and/or inherit/package? (no idea what that does vs regular inherit) <nckx>podiki[m]: Something occurred to me: if you write new-style goodthink (arguments (list …)), you don't need those keyword exceptions, everything lines up fine. So either they were removed quite proactively (unlikely) or, at least, it will matter less & less as time marches on. <podiki[m]>slimjim: did you find out what package/inherit does exactly? i wonder if that came into play <nckx>It's super counter-intuitive if you're not comfortable with Scheme-the-programming-language. (Am I? Some days. ☕ helps.) <slimjim>yeah and I'm not. I tried looking at the def of substitute-keyword-arguments but define-syntax mystifies me <nckx>Do note that you're not inheriting flags anymore, just substituting your own. I don't know if it matters. Probably not. <the_tubular>I reran the installer once again and got the same error as before : gcc-10-cross-environement-variables.patch not found ... <podiki[m]>nckx: well I get the indentation wrong with the #:tests? #f (press enter), maybe that needs a rule? I tend to see indentation problems when doing a package argument stuff <slimjim>yeah it works because gdb doesn't actually have any configure flags to inherit <slimjim>so I've somewhat punted understanding the real problem because it doesn't affect me right now <nckx>podiki[m]: But if you write (list #:tests? #f<RET>, as you will probably do more & more in future, it works out fine; see what I mean? <podiki[m]>the_tubular: if I remember the discussion earlier, a file needed to be modified to let guix (the installer?) include the patch; maybe it wasn't pushed/built yet <nckx>Maybe I'm just an optimist looking for reason in a mad world. <slimjim>the end result works, which is what I was going for: a multi-architecture gdb that can debug targets of all supported architectures <podiki[m]>nckx: oh, I may have missed that in the new guix style, that's the new way? <podiki[m]>(I'll have to grep for better examples to copy...) <nckx>podiki[m]: It's more natural when writing in the gexp style. <podiki[m]>the_tubular: if you want it quicker (maybe), you could do a local git checkout of guix (assuming you have something else to run it on), make the change, build the installer image <podiki[m]>nckx: ah, yes thank you. I have that open in a tab to read still... <the_tubular>podiki[m] can you tell when was the last time the installer was updated ? <nckx>But then I'd never write #~`("foo" where Ludo' does. So eh, I guess de gustibuses still disputandstuff. *podiki[m] still needs to learn more/any gexp *nckx like processing list. <podiki[m]>I'm all about lists, putting the L back in Lisp <podiki[m]>the_tubular: what change were you looking for that someone mentioned? you can see if it has been made <the_tubular>From a working install to not be able to install guix ... <podiki[m]>time machine back to pre-big merge and build the installer from that? <podiki[m]>(btw this would be good to post to bug, devel and or help, since the installer needs testing before a new release) <the_tubular>Yes, that's what I'm trying to figure out, jpoiret told me not to use the 1.3 installer, but every installer I use is borked ... <the_tubular>Did try, I had no network access, then I switched to latest <the_tubular>But the network problem wasn't related to the ISO, so I'd have to give it another shot ... <podiki[m]>if I remember the earlier conversation, the patch is missing because it is not listed in gnu/local.mk? is that why the installer image doen't have it? <podiki[m]>me neither, but I know files get listed in it.... <Kolev>Installing Sway, WeeChat, and Emacs on a live system. <nckx>the_tubular: Did you paste your current error somewhere? <nckx>I'm making some assumptions here, like ‘there is a well-defined error’. <the_tubular>Patch not found : gcc-10-cross-environement-variables.patch <podiki[m]>commit 4796b5d6d18dcc0f1b6c8e134977ebdb575c5741 <nckx>‘The guix package also needs to be updated to point to a commit that includes this patch.’ Now I can't even add anything 😉 Thanks! <podiki[m]>is where the patch was introduced, not added to local.mk Does that determine what gets included on the installer image (as a list of files Guix needs?) <nckx>I'll push those patches. <slimjim>ahh nckx, I think I've hit on the issue: `(substitute-keyword-arguments (package-arguments gdb) ((#:configure-flags flags '()) `(append ,flags '("-TEST"))))` generates an exception on the `,flags` part when gdb has no existing configuration-flags, so flags is set to '() <the_tubular>I can't run guix pull, it also errors, so it would be cool to push a new ISO with those errors fixed <slimjim>so that's why generating my own list, and never referencing the original `flags`, works <podiki[m]>sounds like once that patch is pushed and the guix version is updated, the next installer image built by the CI will be fixed (of that problem) <podiki[m]>click on the evaluation number to see the commit it was built from <nckx>One that doesn't exist yet. <nckx>I'm testing the updated Guix package. <nckx>Then I'll push it, then the CI can build a new ISO image. <podiki[m]>the link I posted will search for the latest installer image, you can see there how fresh it is and from what commit <slimjim>nckx: so what would be the proper way to append configure flags to an inherited package in a generic way where you don't know if the inherited package actually had any configure flags? E.G. you don't know if flags will be '("Something") or '() <slimjim>becuase ,flags breaks when flags is '() <the_tubular>Got it podiki[m] I'll report back with the "stable" installer <nckx>I'll get back to you on that if nobody else answers, slimjim :) *nckx overstressed, sorry. <slimjim>no worries - I got my current problem solved. I'm still tugging on the thread because this seems like an area for improvement - either to make substitute-keyword-arguments more robust, or to improve my own scheme understanding if I'm just not getting some basic syntactical thing <KE0VVT>Wee! On live system. Can't get Sway running, though. XDG_RUNTIME_DIR not set, it says. <podiki[m]>slimjim: from my basic understanding not sure why that should fail, you are appending to '() or a list, either way that should be fine. but maybe that's an edge case? (since if you are inheriting you can know beforehand) <tschilptschilp23>there way to tell pytest where to look for the tests? I think I saw it somewhere, but cannot find this definition anymore... <mroh>tschilptschilp23: try something like `(add-before 'check 'change-directory (lambda _ (chdir "test-dir") #t))` in your modify-phases. <podiki[m]>in the meantime you helped fixed a bug for the next version at least :) <the_tubular>Umm, I got to the very end and got : /mnt/boot/fie doesn't look like an EFi partition ... ? <podiki[m]>someone here will know better, but was it formatted correctly? (fat?) <nckx>Well, your firmware was unable to find a boot partition IIRC. <nckx>So maybe the file system really is hosed. *nckx FAIL: tests/cran.scm, ugh. I'll try building guix on another machine. <nckx>Although this seems to have become less urgent which is a relief. <podiki[m]>I remember the 1.3.0 installer is a bit flakey with partitioning in my experience, especially if there were any partitions before (whether it created them or not) <nckx>sudo lsblk -f might be of interest. <nckx>Oy. That would not combine well with a partition that we're not sure is 100%. <KE0VVT>I'm using a TTY on a live system because I cannot log in due to elogind. <the_tubular>podiki[m] I used the graphical installer and didn't touch anything <podiki[m]>I had (this summer) better luck with starting with a wiped disk, though later just did everything manually <podiki[m]>sorry, gotta run, but hope we get you set up soon! (I know it is frustrating, let's hope this means we'll get 1.4.0 in better shape) <kwjc> thanks dev team for your hard work! <KE0VVT>Thanks for not posting emoji. I cannot see them in this TTY. <nckx>I mean, someone had to say it. <singpolyma>Dang, my emoji keyboard doesn't have all those "bold letter" ones to make while words <nckx>KE0VVT: You've suffered long enough; they were just spaces. <nckx>The only failing test turned to be a Cran description parsing XFAIL (that… didn't?) so push we go. <singpolyma>I enjoy that #guix whiplash between very serious work, bikeshedding, and shitposting <nckx>the_tubular: Serious question, are you cursed? :( <KE0VVT>the_tubular: Are you stuck in a TTY like me? Because elogind. <nckx>I genuinely feel bad for you. I've had weekends like this, best case you learn a lot, but it's not fun. <the_tubular>Also the installer is defaulting the EFI partition to btrfs for some reason ... <the_tubular>So I changed the efi partition to fat32, and now I can'T mkfs the other partition <the_tubular>Installer jsut crashes and forces me to start from the beggining every damn time ... <nckx>Are you familiar with installing Linux (mostly) by hand? Partitioning etc.? <nckx>Walking you through it over IRC is a bad idea but it's what I'd do at this point. <nckx>I admire the work put into the installer but I wouldn't use it myself TBH. <tschilptschilp23>mroh: ha -- I got properly fooled by the pypi tarball! It simply does not carry the tests, and I was looking at the github source ;) <nckx>No, I've never seen that. <nckx>This makes me suspect that it's not Guix's ‘fault’ at some level. <nckx>the_tubular: Anything weird in dmesg? <nckx>Maybe your binaries (more likely: a shared, shared library) are corrupted by a bad USB write? <nckx>Don't get me wrong, it's Guix's ‘fault’ that it's not robust against deviation from the norm, but mkfs returning 1 is not anything I've ever seen or can explain through Guix. <the_tubular>When I try to manually mkfs it I get : cannot open /dev/sda2, ressource might be busy ... <nckx>I would repartition at this point — really wipe the table and start anew, and wipefs -a each partition for good measure — before you continue. <nckx>the_tubular: That just sounds like it's mounted? <nckx>If you started the installer but it failed/crashed, that would do it. <nckx>Is the data on the target drive precious? <nckx>Bind mounts can be deceptive. <the_tubular>Also why is the isntaller defaulting /dev/sda1 (efi partition) to btrfs ? <nckx>Yeah that's just buggy and unforgivable. <nckx>I don't know why though. <KE0VVT>In general. Haven't tried it in Guix. <nckx>Oh, the CI ‘images’ specification only evaluates every 24h. 20h to go. That's not good enough. *nckx wonders if they can cheat. *nckx leaves to get wood. *the_tubular retires again <the_tubular>my /efi partition is 500mb so that's obviously not it <the_tubular>I have can,t show output as I'm in the installer, and no way to retrieve it <blackbeard>beware I have never used the graphical installer <blackbeard>I am not sure if the installer has more tty's that you can use <the_tubular>As I said above, they are all pretty much empty, let me triple check <blackbeard>it seems like there is, as you said, a problem with grub and efi <the_tubular>When I ran df -h, not a single partition is more than 5% <blackbeard>if you are just installing, maybe try wiping the hard drive and starting over? <nckx>It's your NVRAM (again, this happens… a lot?). <nckx>That would work around it! <nckx>I meant that in a good way. <nckx>the_tubular: Apparently, deleting (at least) /sys/firmware/efi/efivars/dump-* is safe. <nckx>And should free up enough space. <the_tubular>Ok ok, is there something else that I can also do so the problem won't come back ? <nckx>I say ‘apparently’ to cover my 🍑 and because I don't have any EFI systems myself. <the_tubular>I did as you said, and rolled back my bios to default last time <nckx>Oh, you were one of the people who had this problem before? <nckx>Sorry for not remembering. <blackbeard>nckx: hahahaha seems like you are used to that good old corporate life <nckx>I don't know who (Linux, the firmware, other nefarious gremlins) actually creates those dumps. <nckx>I am quite sure that those are safe, but you get only an internet stranger warranty. <nckx><I don't know who> so I can't say how to stop it for good. <nckx>I think that's a weird quirk of efivarfs: it returns EINVAL for EEXIST. <nckx>So there are no dump files and it's still full? <nckx>You can try to delete unused boot entries with ‘efibootmgr -b XXXX -B’ but if we've been here before, you already knew that. And that's the only solution I know. <nckx>Because it always sufficed for me. <blackbeard>the_tubular: those commands are in a link from the arch linux forum I sent <nckx>Where XXXX is an index from ‘efibootmgr’. <nckx>Oh, follow blackbeard's guide then, I missed that. <nckx>It will give you more context than I can type in this little text field. <nckx>You can certainly delete anything Guix since you'll be re-adding it anyway ☺ <nckx>‘copying GRUB's .efi loader to $ESP/efi/boot/bootx64.efi’ is a good tip. I did that one one buggy machine without knowing it was a known trick. <nckx>I'd rather not. Just ‘guix install efibootmgr’. <Michal-Atlas>Hello, sorry to interject, but could you please tell me where to find how to 'enable distributed builds'? Can't find what config or flag it is, would be grateful for some pointers, been at it for hours. <nckx>I think Windows uses bootx64.efi. Which would explain why anything else tends to elicit firmware bugs, because nobody ever tests anything but Windows. <nckx>Your system is in an unknown state right now. <nckx>Normall, you can install stuff in the installer just fine. <nckx>As well as pull, but don't, it would just waste time here. You don't need a cutting-edge efibootmgr. <the_tubular>And then when I run the command, it says command not found <nckx>Michal-Atlas: Do you mean offloading? <nckx>Did it make a difference. <nckx>It has a --help option if you really care 😉 But I don't think bash does negative caching anyway, let's forget it. <Michal-Atlas>I tried running guix install ... -M0 to see if my offloading really works since it passed the guix offload test. but it replied "either increase --max-jobs or enable distributed builds" <nckx>Just hit C-d to log out & back in, the_tubular. <nckx>Michal-Atlas: Ooh, that is some odd non-Guix terminology… and indeed it only occurs in the nix daemon code. <nckx>Michal-Atlas: Try that, e.g., adding --max-jobs=2 to the guix command. <the_tubular>I try re-running guix install efibootmgr, the output says : the following package will be installed : efibootmgr ? <nckx>And typing efibootmgr doesn't work? <nckx>I'm starting to feel as you must at this point. <nckx>/root/.guix-profile/{,s}bin <nckx>Or try running /root/.guix-profile/sbin/efibootmgr directly I guess. <nckx>If PATH is buggered, let it be. <Michal-Atlas>That seems to just build everything locally, I'm sorry am I making some really dumb mistakes here? I tried following the Docs as closely as possible <nckx>I don't think so Michal-Atlas. I find offloading hard to debug as well. It's great when it works… <nckx>Just ls /gnu/store/*/sbin/efibootmgr <nckx>Run that. At least you can tab-complete (we're not animals), right? <nckx>If that doesn't return anything I'm stumped. <nckx>And we need to reboot and start from a known state. <nckx>Show you all /sbin/efibootmgrs in the store. <nckx>No matter if they're in a profile or not. <nckx>There should be only one, but who knows anything at this point. <nckx>Then you can invoke it with that absolute file name. <the_tubular>I'm not sure I follow, what is the * in that command ? <nckx>Michal-Atlas: I'm reading the docs to see if I can find any missing gotchas I've noticed myself . <the_tubular>ls /gnu/store/*/sbin/efibootmgr, no such file or directory <nckx>Here, for example, it returns (amongst some profile entries) /gnu/store/j89vi2a3xkzp1msbqcfplk2767rwh9ax-efibootmgr-17/sbin/efibootmgr. But it seems it's really not in your store. <nckx>You should really reboot since we're debugging something in your current ramdisk that doesn't, in the end, matter. <the_tubular>Totally agree, but I've rebooted more time than I can count at this point <nckx>I know it's tedious to go through the installer again but you'll have to anyway, and this way we can immediately switch to a tty, install efibootmgr, delete some crud, and hopefully switch back to the installer to finish this. <nckx>the_tubular: You'll restart with a clean live system. <nckx>Nothing will change in your NVRAM but you should be able to install packages now. <Michal-Atlas>That's a shame, it's one of the things I really was looking forward to, having a mid-range older laptop and a beefy server to hook into. Anyways, thanks for the motivation, knowing it's not supposed to work first try kinda lifted my spirits.'=D But I tried a different package (this time from the main channel) and it works flawlessly. SO the <Michal-Atlas>problem was in using an alternate channel, my bad, should've checked that first <nckx>Michal-Atlas: I'm glad it worked! …but that's not a known limitation. Not expected. <nckx>Michal-Atlas: Some *types* of derivations aren't offloaded, though. Ones that are assumed to be cheaper to simply perform locally. <nckx>But the channel they're from is not itself a factor in that decision. <the_tubular>I got a tty nckx, so just guix install bootmgr and hope ? <nckx>I assume it will work fine, because it should. Then follow the Arch guide which I haven't read to delete stuff. <Michal-Atlas>Oh, that makes sense, but is quite strange, the issue seems to be in qtbase 5 required by the package I'm building, but qtbase 6 gets offloaded just fine. Anyways, thanks for the help, if base guix works then that's 90% of packages, thanks a bunch for your time. <nckx>Any time (which is currently 3am so forgive my sluggishness). <the_tubular>but the output made more sense than before i rebooted <nckx>This might be solved by a logout→in. <nckx>Guix can't easily change the PATH of the shell that invoked it. <nckx>Well, this is where (we were supposed to arrive here quite a while ago :) you'll have to do most of the thinking, because I can't say what's obviously unnecessary. I vaguely remember deleting stuff like booting from floppy, Intel Something Manager, that kind of useless stuff. <nckx>This machine is weird to me. <the_tubular>Yeah, well there is 4 network, probably cause I got 4 nics <nckx>Also, why is it running out of flash if there's hardly anything in it. <nckx>In a chip on the motherboard. <nckx>The one that keeps your ‘BIOS settings’. <nckx>This is very different from what I've seen. <nckx>IME there was always plenty of vendor/legacy stuff that was never ever used. <nckx>Was booting in BIOS mode not an option? My earlier ‘it's a work-around’ was misinterpreted as ‘not a good fix’ but I didn't mean it that way at all. <nckx>Because of course, because you ran over a witch's cat as a child, obviously. <nckx>Unless someone else shows up, trying to fix UEFI that's vastly different from the ones I've used, if that's even possible, does not seem to be a good investment of our time at this point. <nckx>A slightly buggy BIOS might be the least worst option. <nckx>(I share your disappointment.) <the_tubular>I can't remember, I dropped BIOS a long time ago ... <nckx>Quite possibly. I'm not very fresh. <the_tubular>What kind of entry am I suppose to see when I run efibootmgr -v ? <nckx>Not strictly hardware, ‘boot options’, including stuff like ‘Lenovo Recovery’ in that example. <nckx>Which relies on the original OEM (Windows) recovery partition so would be an example of something safe to delete if you had it. <nckx>That's beyond my experience. <nckx>I first learnt of iDRAC less than a week ago when berlin went down. <nckx>It's not been a great week, has it ☺ <nckx>the_tubular: It can do some pretty scary stuff for almost-certainly-vulnerable firmware. <nckx>Wait, new-new, as in #3? <nckx>(Do you know if it's been patched in Guix? Also not really my world, enterprisey logging software…) <nckx>Maybe I shouldn't ask someone without access to current Guix. <the_tubular>Umm, well. my guix machine is down since yesterday, so I don't know about #2.5 <nckx>This a a good case for CVE numbers existing. ‘#3? Well, kind of. #2.5. Oh, so the thing with the stuff but now the stuff is worse? Yes.’ <nckx>Update to 2.16.0: This is a follow-up release because of an incorrect fix to CVE-2021-45046. <nckx>Oh, I thought #2 was just the 3 upgraded to a 9 so people suddenly had to rush to fix it. <nckx>One vuln a week, please, people. <the_tubular>Didn't follow this closely today as I was fighting with Guix all day ... <nckx>If you used one of the ‘discredited mitigations’ you're screwed, but 2.16.0 is not discredited? That's how I read it anyway. <nckx>Who knows at this point. <the_tubular>Fitting that my server is down with no way (that I know of) of getting it back up ... *nckx almost crying: I actually use Guix everywhere and it works fine I don't understand it just works I'm so sorry I swear it does. <nckx>Nah, it's not a real file system and you'd have to know a good bit about the implementation to even guess. <raingloom>if i want to run a gexped script in my shepherd-service, i don't need make-forkexec-constructor, right? <nckx>the_tubular: I don't want to, but I need to head up to bed. <nckx>Good night, and good luck. <vagrantc>hrm. now guix fails to build on x86_64 :/ <KE0VVT>Can I SSH into systems that can't log in because elogind? <nckx>It's expected to fail but doesn't. <nckx>Something pulled in on the way to f98edfa, no doubt. <nckx>I thought it couldn't be related, but it obviously is. <vagrantc>why is the architecture x86_64-w64-mingw32.guix.x86_64-linux <vagrantc>oh, so something got fixed and therefor the tests fail? :) *vagrantc wishes for a XFAILBUTOKIFSUCCESS <nckx>Maybe! But that's not certain yet. *nckx sleepy, slow, but will look into it. *vagrantc very much wonders what w64_ming32w is doing in the job names *podiki[m] does a `guix pull --roll-back`, hooray for guix <KE0VVT>podiki[m]: roll-back does not work for elogind issue <jgart>apteryx, I'm on void linux so that gpg-agent service out of the box feature probably won't work for me? <raingloom>KE0VVT: do you have physical access to the machine? can you boot an earlier system config? <KE0VVT>raingloom: That's just it. I boot system config generation #1 and the problem persists. <vagrantc>KE0VVT: maybe some invalid/incompatible data is stored in the homedir or something? or is this failing before you're even logging in? <apteryx>jgart: if you installed gnupg from guix it should (it's a patch against the guix package) <apteryx>see the thread, I posted my workaround (ssh in, then restart elogind) <apteryx>KE0VVT: the ssh problem though is PAM related, I guess (you updated a remote machine?) <apteryx>I posted a workaround there too for the time being... <apteryx>but it requires an out-of-bound access to the machine <nckx>apteryx: You had to pwn your own machine to gain access? <nckx>I understand now. Never mind. *nckx can't wait for these tests to finish and turn off their brain for a good long while. <the_tubular>What I hate is that every time i retry install guix I get a different error <the_tubular>I don't even know which error I'm trying to fix anymore <apteryx>this is starting to sound like a hardware failure <vipstar>I'm interested in developing an Electron application with the help of Guix. Does anybody know of a Guix package that makes use of Electron? <apteryx>the_tubular: do you have another machine to try on? just to regain some sanity that things are known to work somewhere <the_tubular>And not really, but what would trying it on another machine do ? <apteryx>prove that your installer image works; and make you doubt your hardware instead of the software <the_tubular>I've tried that exact image before and it was working <apteryx>this is one more reason to doubt your hardware (software hasn't changed, known working) -- perhaps something started failing, e.g. your hard drive <apteryx>you had weird issues earlier where grub was not even able to make sense of your partitions anymore <apteryx>eh. it was after your first hard reset I think, after you couldn't login due to the elogind bug <apteryx>then GRUB would not detect your disks ore something similar <podiki[m]>vipstar: as far as I know there are no electron or other js applications like that <the_tubular>So that would makes sense it's the same issue apteryx <apteryx>so things were smelly already back then, hardware wise <the_tubular>Still the issue reside in the NVRAM and not in a disk ...? <apteryx>sorry, NVRAM? not sure how this came into the equation. <apteryx>NVRAM meaning cold storage, right? (not sure how I got video ram in my head) <apteryx>try updating it, or having it operate in the old BIOS-compatible legacy mode <apteryx>did you try the manual installation? <the_tubular>And where it crash just try and to the rest manually ? <apteryx>otherwise it'll leave you in an unknown state and probably fail <apteryx>I think you can grab a template from the installation media; you should touch it as little as possible <apteryx>just follow the instructions in the guix manual <apteryx>everything should be there (it used to be the sole means of installation) <the_tubular>It says Scanning for devices, please wait this may take serveral minutes : then there's a grub prompt <lfam>apteryx: I'm still going with the gnupg upgrade. Updating to 2.2.33 broke the build of python-gnupg, but it works with 2.2.32, which is currently the "next" variant that we already have packaged. So I'm going to wait for the big garbage collector lock and then I'll test the rebuild of all packages that depend directly on gnupg. Mainly because building qtwebkit is impractical here <apteryx>OK! I'm testing 2.3.3 although if 2.2.33 is giving you trouble that's probably not going to go well <lfam>Well, let's try both and pick the one that's best <lfam>The only new failure I found was python-gnupg <apteryx>sorry, it's unrelated to your problem <apteryx>sneek: later tell civodul found one typo in the latest blog post: "But the tools are our disposal " -> tools *at* our disposal <apteryx>also perhaps "reads an sexp" -> "reads *a* sexp" ? since written that way, sexp is probably pronunced the way it's spelled. <the_tubular> wanted to test something, unplugged the cmos, and got that error ... <the_tubular>Am I the only one having this much problem with the installer ? <the_tubular>apteryx I just did it manually, I get this after running guix init : initializing operating system under /mnt error chown read-only filesystem ... <vivien>the_tubular, I installed guix a few months ago, and to be honest, I didn’t use the installer. I installed debian in a small place on the hard drive, installed the guix package in debian, and ran guix system init from there. <the_tubular>But I can't believe no one here is able to help me with the official installer <sam_>you're giving very vague errors <sam_>a picture of the failure or something would help I guess <opalvaults[m]>the_tubular: the graphical installer is incredibly buggy unfortunately. I recommend doing the manual option if you can. <the_tubular>One telling me my VRAM is full, and the other one telling that /mnt is read only. <opalvaults[m]>It's worth doing manually at least once so you may be able to pinpoint more of these errors if they keep happening. <opalvaults[m]>The reference manual is actually very informative (even moreso than the Arch Wiki installation guide imo) and makes it quite easy. And if you run into a bug again, you'll know exactly which step is going wrong. <the_tubular>I mean, what do you mean "manually" wiping the disks and running guix init with your config.scm by hand ? <opalvaults[m]>You'd have to do that, without or without a graphical installer. ;) <the_tubular>Cause when I run guix init right now, this is the output I'm getting, initializing operating system under /mnt error chown read-only filesystem <opalvaults[m]>It means that it can't write to /mnt. Whether that be because of a filesystem error, or something otherwise I couldn't tell you. <opalvaults[m]>the_tubular: are you able to upload a picture of the stack-trace if there is one associated with that error? <opalvaults[m]>try and run fsck on the filesystem you suspect is acting oddly <opalvaults[m]>It will be either a block device name like /dev/sdX, or you could even run it on a mount point such as /mnt. Without knowing the particulars of your situation I couldn't tell you which one. <avp_>apteryx: Yes. I did 'make clean', then 'bootstrap'. <opalvaults[m]>It may not fix the problem but might give you a better picture of what's going wrong. In these instances when I run into installation errors I typically wipe and start again so if you're unable to get a better idea of the issue, I recommend doing it again. <apteryx>./bootstrap && ./configure --localstatedir=/var --sysconfdir=/etc && make <apteryx>(inside a 'guix environment guix' env) <opalvaults[m]>Ah, btrfs might have it's own way of doing filesystem consistency checks. <apteryx>with btrfs you don't usually don't want to try to repair the file system; it doesn't do much and if you force it it can cause more harm <avp_>apteryx: Alas, the same error appears. <apteryx>beware to not loose your uncommited work <apteryx>since you said you don't care about the data <apteryx>I'd clear the file system just to be safe, as suggested by nckx the other day <apteryx>they suggested wipefs; I don't know that tool but a good way is to zero the first megabyte or so of the disk with du for example <apteryx>this may avoid confusing the installer <avp_>apteryx: This worked, thanks! <zimoun>apteryx: A question version-1.4.0 branch. Do you consider it is in “frozen” state and now let only fix important and blocking bugs? Because if no, I feel duplicate and/or extra-work. And what is the schedule? <vivien>I joined, now waiting for jgart to start the meeting :) <zimoun>florhizome[m]: maybe I am wrong with timezone. It is 10:30 CET, not ET. It is 4:30 in Eastern Time, I guess. <jgart>funny I'm still awake, I should get some rest before the meetup <jgart>trying to get Guix running on alpine linux <vivien>Could you tell us in how many hours that is from now? <jgart>vivien, what timezone are you in? <vivien>So that will be at a time when I won’t be there <jgart>we'll have another packaging and separate documentation meetup next month <jgart>or I should say next year ;) <florhizome[m]><jgart> "It is 4:44 AM here (ET)" <- So 16:30 for the european crowd (i think that's quite many ppl) <avp_>jgart: Where are the meetings take place? <florhizome[m]><vivien> "I joined, now waiting for..." <- Oh i was close before joining, too. <jgart>Was my UTC conversion on the mailing list thread correct? I though it was... :) <avp_>So when will the meeting be in Moskow time? <jgart>anyways, what would people like to work on? <jgart>I use it all the time. It's pretty cool and helpful <jgart>emacs has a built in time zone helper also <jgart>anyways, I'm going to get some rest before the meet <avp_>Okay, thanks! So it will be 18:30 in Moskow time. I think I'll be able to join. *rekado pushed wip-java-bootstrap-simplify and asked ci.guix.gnu.org to build it <unmatched-paren>florhizome[m]: did you ever manage to do those changes to the channel repo? i just realized that the last commit i made stopped it building <unmatched-paren>sneek: later tell florhizome[m]: did you ever manage to do those changes to the channel repo? i just realized that the last commit i made stopped it building <lilyp>Polari 40 feeling fast today <florhizome[m]><unmatched-paren> "sneek: later tell florhizome..." <- What do you mean? No i never managed guix pulling ^^ it broke indeed. <florhizome[m]>Though i still have the packages in my local GUIX PACKAGE PATH and i don't know how guix deals with that. <sneek>florhizome[m], you have 1 message! <sneek>florhizome[m], unmatched-paren says: did you ever manage to do those changes to the channel repo? i just realized that the last commit i made stopped it building <unmatched-paren>florhizome[m]: yeah, i renamed nvim.scm to vim.scm and forgot to change the module declaration from (paren packages nvim) to (paren packages vim) <florhizome[m]>I started to do some stuff on the local git checkout though yesterday night, i think i will push some stuff to you as the day progresses :) <unmatched-paren>i never use guix shell, i just guix package -f the file, i should probably learn how to use guix shell <unmatched-paren>wdym? just add a line to the bottom of the file to return the package; guix package -f evaluates the file and installs the package that it returns <florhizome[m]>There is just some more stuff i would like to clean up (modules, GCC inputs not needed (anymore)) <unmatched-paren>yes, it probably is best to guix shell, i've just never tried it -.o.- <florhizome[m]>Maybe we'll have a " Basic workflow for collaborative work on a guix channel" at the end. <unmatched-paren>i'll have a look at the guix shell blog post, maybe i'll learn something <florhizome[m]>There seem to be mostly individually Drive channels but Like this it's not so much fun. <unmatched-paren>`info guix-shell` will probably tell me more, but browsing info pages is... not fun <florhizome[m]>I'm not that well versed on the theoretic side of things, so all the virtualenv vs container stuff ist a bit confusing to me. <unmatched-paren>containers are sort of like lightweight virtual machines, and environments are basically just subshells with sanitized environment variables and a different package profile; that's my understanding of it at least <lilyp>Can someone grab me a link on the Guix vs. pip comparison? <unmatched-paren>containers can emulate different operating systems and architectures using the kernel's container system, environments run only on the host os unless you --container them <jpoiret>Containers can't emulate architectures <jpoiret>I think there's too much mysticism around containers, it's just linux namespaces in the end. <unmatched-paren>jpoiret: ah, so that's the main difference in capabilities between containers and vms? <unmatched-paren>so does a container emulate an operating system, while a vm emulates a processor? <jpoiret>Programs in namespaces run like normal programs, but the kernel reports things to them differently <the_tubular>Hey jpoiret, I made some progress but still haven't been able to install ...! <jpoiret>unmatched-paren i'd suggest reading a bit about namespaces, it'll be eye-opening. It's a great feature <jpoiret>the_tubular: i'm sorry the install isn't as smooth as we'd like <the_tubular>Yeah, I'm getting a NVRAM error I think and I don't know how to fix it <unmatched-paren>i've only used a vm once (to test gnu/linux while i was considering switching to it from windows) and never used containers, so i was basically just saying what i'd heard and deduced (incorrectly, it seems), i should have made that clearer <the_tubular>I've check with efibootmgr -v, but everything I see there looks correct <jpoiret>Unfortunately i won't be available until later this evening (CET here). <the_tubular>I removed CMOS battery, see if it was gonna fix the NVRAM issue <kwjc>why is bzcat not working?? <kwjc>what am I doing wrong here: 'bzcat /path-to-file' <kwjc>thanks. so it isn't a standard '.txt' and doesn't know what to do I suppose <unmatched-paren>`echo "hello" > file; and bzip2 file; and bzcat file.bz2` prints 'hello' <unmatched-paren>this is in fish, not sure if other shells support `; and`, you might need to replace that with && <lilyp>btw I found it, y'all no longer have to search on my behalf <kwjc>unmatched-paren: thanks for the advice, i'll make sure to do that in the future. yes, it does work perfectly fine. guix tells me to "View build log"; I automatically (wrongly) assumed that it would equivalent to a basic .txt and work fine <kwjc>does compiling something take a significant amount of memory...? guix pull gets to here: '[459/650] compiling... 41.2% of 325' and then stops. I only have 4gb of ram, but I have 4gb of swap (iirc) <AIM[m]>Hi all! I'm new to guix! I am currently an Arch Linux user... I use it vanilla, I installed it manually... I came to know about guix being fully opensource and it caught my attention... Currently planning to leap to it... And maybe use it as my daily driver... <kwjc>okay, this might be the root of all my issues. I don't think guix is using/recognizing my swap /dev/sda1 <AIM[m]>Can someone help me with btrfs on guix? Is there any example config I can refer to? I did check the official manual... But I'd like to see more examples to test out... If you have your own public config in scheme, that has btrfs config, that you'd like to share, that would be nice... <AIM[m]>kwjc: You using manual mode or graphical install? <kwjc>AIM[m]: I used the graphical install ages ago now. yes, I am post install on bare metal <AIM[m]>You're installing packages but swap ain't on? <kwjc>nothing even recognizes that /dev/sda1 is a swap except for: 'sudo fdisk -l' <kwjc>lsblk says type is "part"?? isn't it supposed to be TYPE SWAP <kwjc>AIM[m]: had to do 'sudo', but yes it worked! thanks a lot <AIM[m]>kwjc: You may need to format it using mkswap <bost>Hi. Can somebody explain me why an installed package is not available? <bost>$ guix package --list-installed=postgresql | grep "^postgresql.*13\." <bost>postgresql 13.3 out /gnu/store/1nlzmg4hw4gga56g58dsqf9nx90z9kkn-postgresql-13.3 <bost>$ guix package --list-available=postgresql | grep "^postgresql.*13\." <bost>postgresql 13.4 out gnu/packages/databases.scm:1077:2 <kwjc>AIM[m]: I know that apteryx has a working btrfs+raid setup <rekado>bost: you installed 13.3 but 13.4 is available. Have you run “guix upgrade”? <AIM[m]>raghavgururajan: Yours doesn't have the subvolume right? <raghavgururajan>singpolyma: I could use subvolumes instead of LVM, but its not yet supported in Libreboot BIOS. <bost>rekado: the thing is that I DON'T want to upgrade postgresql 13.3 -> 13.4. The 13.4 breaks something with `locale` so I want to stick with 13.3. <bost>rekado: I'd like to be able to run `guix shell postgresql@13.3` <stikonas>rekado: you can also simplify that delete file snipped for jamvm 1.5 <KE0VVT>Greetings from the TTY of a Guix live image. <AIM[m]>Oh man, you guys are making me jealous... I haven't completed reading docs and researching to install guix... <AIM[m]>Idk if I should take a blind leap and then figure out everything on the way.... <raghavgururajan>AIM[m]: I feel ya. I felt the same about 2 years ago. But don't worry, you and guix will grow into each other. :) <AIM[m]>Like I'm guessing the first time install will be totally messed up with stuff just not clean <KE0VVT>raghavgururajan: I am using WeeChat inside tmux. <raghavgururajan>AIM[m]: I'd say start with a typical configuration and move on step-by-step. Even if you screw up, you can roll-back your system to previous generation, using `guix system roll-back`. <kwjc>AIM[m]: take a blind leap! leave the proprietary blobs behind!* [*if you have multiple pc's] <AIM[m]>We should have a user wiki for guix no? <AIM[m]>Like I'd like to see what other users have been tinkering with in a single place <tex_milan>Hello Guix. When overriding a package, one must repeat the whole (source (origin... Isn't there any other way how to just update version number and hash? <AIM[m]>raghavgururajan: Does it work like apt autoremove? <KE0VVT>AIM[m]: I am stuck in a live system, because I did a "guix system reconfigure" when they pushed core-updates-frozen, and now I cannot log in to the system, and GDM does not start. <AIM[m]>KE0VVT: Don't scare me like that.... <singpolyma>AIM[m]: there is a guix area on the libreplanet wiki *raghavgururajan looks at KE0VVT 's error <KE0VVT>AIM[m]: core-updates-frozen was a HUGE change. It normally goes just fine, and only a minority get the error that I do. (Namely, HDD users.) <KE0VVT>raghavgururajan: Sorry it's not in plain text, but I can't just copy and paste from a TTY, and I can't be running Sway inside tmux where I can paste. <KE0VVT>I like how I don't have to deal with all these different package managers, how I can just "guix install emacs-poly-noweb" etc. <AIM[m]>I'll have to find Guix compatible hardware tho, my wifi and bluetooth will be gone after leaping... I don't know if that's swappable in Thinkpad X1.... Or else I'll have to get dongle... For now I think I'll use USB Tethering.... <KE0VVT>AIM[m]: I'm guessing you have ethernet to do the installation with. <KE0VVT>AIM[m]: My uncle uses a dongle all the time. <AIM[m]>I do have Ethernet port (dongle coz Thinkpad has it as a dongle) <KE0VVT>I wish I could check my email from here. <raghavgururajan>KE0VVT: Some setuid programs are not available in live environment. That could cause some of your errors. <KE0VVT>raghavgururajan: Is there a way to run i3? I am not familiar with X. I'm just used to typing 'sway' and getting a desktop. <raghavgururajan>KE0VVT: Also, graphical application require X session which in-turn required login-manager service-type which are not used/active in live sessions. <mbakke>apteryx: would it make sense to take "stable" uipdates such as Python 3.9.9 and Meson 0.60.2 into version-1.4.0? <mbakke>possibly also 'file' (see 1f4319886a9) <AIM[m]>Of I want a rolling relwase like Arch should I go for latest? <KE0VVT>AIM[m]: Guix is all rolling, IIRC. <AIM[m]>What's the diff between stable and latest iso then? <unmatched-paren>AIM[m]: guix is about as unstable as arch (although it may be slightly behind in some places simply because we're not that big); what drove me to use guix is the rollback feature :) <mbakke>Guix is "rolling", but has lots of weird edges and rolls somewhat erratically :) <raghavgururajan>AIM[m]: I'd also advice to read the 'Managing Software the Guix Way' section of manual *before* installing guix, to get a hold of it. :) <AIM[m]>So I'll flash the stable image then <unmatched-paren>also, when you get to 'Disk Partitioning' (if you do graphical guided install) be sure to select the new boot partition and change 'Format' to 'true' <KE0VVT>unmatched-paren: Except the rollback feature doesn't work for this bug that I'm experiencing. <AIM[m]>I hope I don't get sleepless night configuring stuff.... <singpolyma>AIM[m]: the moment you run guix pull after installing you will be on latest. The only difference is the installer version <AIM[m]>unmatched-paren: It's false by default? <KE0VVT>AIM[m]: I installed with the stable image. It worked until they merged this rare, giant change called core-updates-frozen. <KE0VVT>AIM[m]: I don't know, but it incl. a lot of changes, incl. GNOME 40. <KE0VVT>I use GNOME and Sway. I don't use X anymore. <unmatched-paren>if something goes wrong with your packages, services or system configuration, usually you can just select the 'previous configurations' option in grub and everything will work again <AIM[m]>Since "make install" for DWM won't work, I'll go for Xmonad <AIM[m]>unmatched-paren: Roll back to prev generation, right? <unmatched-paren>of course, something like grub-install is not transactional, it's still dangerous <KE0VVT>AIM[m]: unmatched-paren is right. This has never happened, where we can't just roll-back. This rare change was HUGE, and somehow we can't roll-back. It's very strange. <unmatched-paren>it won't undo a filesystem disaster, but it will undo almost anything else <KE0VVT>unmatched-paren: The other killer feature is not using a giant blacklist, and banning stuff like Chromium and every emulator under the sun. <AIM[m]>I thought of leaping to guix, when one of my friend told that his professor uses trisquel.... Which is based on Ubuntu.... I personally am not a fan of apt... So I went for guix.... <unmatched-paren>services, packages, system, etc. are all managed in the /etc/config.scm file, which can be rolled back at will <singpolyma>Since it's trivial to package your fork with your config <KE0VVT>I installed with the graphical installer, installed GNOME, and everything was going great for a while, until they did this rare change. <singpolyma>Then you can get rollback on dwm config and everything <KE0VVT>AIM[m]: You extend the DWM package definition. <AIM[m]>unmatched-paren: Yeah... I do know lisp... And I did checkout some of the guile scheme examples... I'm still a beginner tho... <KE0VVT>singpolyma: I'd like to see those. Normally, Guix puts a hamper on those types of apps. <ngz>Duh. It seems I need Rust 1.56 for a package update. <ngz>Is anyone working on a Rust bump? <KE0VVT>AIM[m]: If I am not mistaken, one can just copy and paste their patch into the package definition, basically. <ngz>unmatched-paren: what was the bug number? <AIM[m]>But, I am checking out Haskell and therefor going for Xmonad <KE0VVT>AIM[m]: Probably simpler to use Xmonad, anyway. <AIM[m]>Hey if I install the Xfce version and then edit system config.scm to just have Xmonad and reconfigure the system... Will all the Xfce packages go * poof *? <KE0VVT>AIM[m]: I installed GNOME, then replaced it with Sway. <AIM[m]>KE0VVT: That gives me confidence... Atleast I can edit the config with browser open... <KE0VVT>It's very nice how you can cleanly uninstall DEs like that. <KE0VVT>I mean, I hate KDE, but people should be able to use it if they want to torture themselves. <AIM[m]>Here I go guys, Imma install guix now... I hope all goes well! 🤞 <KE0VVT>singpolyma: Is that plain text or HTML? (Stuck in a TTY here.) <ngz>hmm it is not in master branch <ngz>It was deleted. There is a core-updates branch, tho. <ngz>Do you have a commit hash ? <KE0VVT>Ugh. -bash: file: command not found <singpolyma>KE0VVT: HTML I guess but lynx/w3m is your friend? *KE0VVT opens file in links. <ngz>It's a debbugs command <KE0VVT>singpolyma: Wow, that's really cool. The patches are pulled in just like tarballs for the packages themselves! <ngz>unmatched-paren: No one reviewed it yet =/ <unmatched-paren>one was the rustc one, one was the cargo one, and one was a follow-up cargo one <unmatched-paren>unfortunately, i didn't know how to post to an existing thread, so i sent the cargo one in a seperate issue <unmatched-paren>once i was informed of how to reply, i sent the follow up to the first thread <ngz>Could you close the merged bugs, and create a new one with all the patches? <ngz>In a Tetris fashion. <unmatched-paren>aha! i can still retrieve the follow up by pasting the url, even if it doesn't appear in the search <KE0VVT>I cannot search issues.guix.gnu.org in a text browser. <KE0VVT>I want to read the bug about elogind not letting people log in. <florhizome[m]><AIM[m]> "Hey if I install the Xfce..." <- Only Thing that might stay are config Files in home. Maybe one day we'll manage that too :) <KE0VVT>unmatched-paren: I just cannot find the URL for the bug. I can read it just fine, if I had the URL. <ngz>unmatched-paren: There is no commit message, you just dumped a patch file, didn't you? <ngz>hmmm I can't see it. <florhizome[m]><KE0VVT> "I mean, I hate KDE, but people..." <- I built kwinft the other day, maybe you could build it yourself now <ngz>unmatched-paren: I see. You need to expound it a bit. For example, it should list all variables being updated or created. <unmatched-paren>florhizome[m]: nope, but git send-email does use format-patch i think <unmatched-paren>i think it's basically equivalent to running format-patch then firing off an email containing the patch <KE0VVT>I wish I could compose an email about that bug. <singpolyma>unmatched-paren: piping the patch (which is formatted as an email) to sendmail, mostly <AIM[m]>raghavgururajan: Thank you, but I am leaving suckless... <raghavgururajan>AIM[m]: You could also create definitions locally and use `--load-path` argument to install it. <KE0VVT>singpolyma: Do you use Suckless? <AIM[m]>Been thinking of going Xmonad for a long time.... <singpolyma>KE0VVT: in many places yes. I'm currently experimenting with some other stuff for various reasons, but suckless always feels like home base <KE0VVT>Are there such things as USB ROMs? <raghavgururajan>AIM[m]: I see. I am too. My channel use to have customization for dwm, dmenu and slstatus too. Now just st. *KE0VVT imagines singpolyma using Gajim in dwm. <ft>what does "guix upgrade" try to tell me, when it fails with "guix upgrade: error: integer expected from stream"? (Happens after it figured out which derivations will be built and how much to download) *raghavgururajan use to use gajim in dwm <raghavgururajan>AIM[m]: If you'd like something lispy/hackable as guix, i'd recommend StumpWM. <KE0VVT>I dream of a Wayland compositor running in a Guile REPL. <KE0VVT>singpolyma: What IRC client do you use? I cannot type emoji in Terminal. <KE0VVT>singpolyma: Conversations is for XMPP. <AIM[m]>raghavgururajan: Is it mature and documented really well? <KE0VVT>Good morning. :-( Still stricken by the bug. <KE0VVT>singpolyma: Whoa. I have to check this out. <raghavgururajan>singpolyma: I recently shifted from irc.cheogram.com to chat.sr.ht, to try the latter out. <singpolyma>raghavgururajan: we'll miss you. Come back any time 😉 <nckx>Sorry to hear that, KE0VVT. <USB ROMs?> I have a few old ones with a hardware write-protect switch but that seems to've gone out of style. <nckx>unmatched-paren: …merging bugs should never destroy data? :-/ <KE0VVT>nckx: I want to write live images to a ROM of some sort, so I can't say "I'll just download the ISO again and rewrite it." <KE0VVT>Oh! I have my FSF card with Trisquel! <nckx>Three cheers for the FSF. <raghavgururajan>singpolyma: :D I mainly had issues with IRC channels loading very slow on clients like Gajim. And if the WiFi was laggy, oh boy. <AIM[m]>I'm jumping onto FSF train with my Guix! <singpolyma>raghavgururajan: interesting. I haven't had similar, but this is easily my biggest regular use IRC channel <KE0VVT>singpolyma: Cheogram might be easier for IRC than Matrix, ZNC, a native mobile IRC client, or a WeeChat relay. <KE0VVT>nckx: Anybody else here with the same bug as me? *KE0VVT cannot see images <AIM[m]>I want browser to edit my config <KE0VVT>nckx: I haven't been able to read the thread. It renders poorly in both links and mobile Chromium. <AIM[m]>kwjc: I'm an Xmonad fan... Xfce stays till I learn how to modify config.scm for Xmonad.... <KE0VVT>AIM[m]: Why do you still need to edit the .scm for Xmonad? I thought it takes a Haskell file. <AIM[m]>I mean add Xmonda to config.scm and reconfigure system <AIM[m]>Remove all unnecessary bloat as well <kwjc>not sure if you know about 'info guix' but there ya go. full manual in terminal afaik <patched[m]><kwjc> "not sure if you know about 'info..." <- guix got me woke on infopages <patched[m]>Will learn texinfo someday to use it for my own documentation <KE0VVT>I wish TexiWEB Jr. had good Info output, but it only looks good in PDF. <KE0VVT>I'd like to write literate programs and read them in Info. <ngz>KE0VVT: see Emacs Org mode. <KE0VVT>ngz: Org's noweb features do not implement all the features of literate programming. Macros get expanded everywhere they appear, so readers will just see giant chunks of code. <KE0VVT>Installing Trisquel and putting Guix on top of it, until this bug goes away. <KE0VVT>Can I use Guix's GNOME on Trisquel? <KE0VVT>I can no longer find the article about GNOME, Guix, and RHEL 7. <sneek>Welcome back civodul, you have 1 message! <sneek>civodul, apteryx says: found one typo in the latest blog post: "But the tools are our disposal " -> tools *at* our disposal <AIM[m]>Ice cat is a maintained project, right? <AIM[m]>GNU hasn't come up with a replacement, right? <kwjc>AIM[m]: in fact, the version we get here in Guix is the more... bleeding edge version of ice cat <raghavgururajan>nckx: For every new commit on master, new jobset is created in CI right? <nckx>Not necessarily. It polls every N minutes. <nckx>So if you or multiple people push 5 commits at once or within N minutes after the last evaluation, Cuirass will evaluate them all together. <KE0VVT>OK, got Guix installed on Trisquel. Running WeeChat and tmux from it. <Guest2041>im trying to run guix system image and i get the error no space left on device <Guest2041>In procedure copy-file: No space left on device <Guest2041>`/gnu/store/0zs49yx0nj9yrb9xy9l2aadh3lkgrm7v-e2fsprogs-1.46.4/bin:/gnu/store/0zs49yx0nj9yrb9xy9l2aadh3lkgrm7v-e2fsprogs-1.46.4/sbin:/gnu/store/m4y88rzsfmg9cxz41hykz08dk31h0wyn-fakeroot-1.26/bin:/gnu/store/57gsi71wf49n5yg6m74dfnkkcb62q3gq-dosfstools-4.2/sbin:/gnu/store/dyagl2vls2gn52md3wpmj6i8jk33xanl-mtools-4.0.35/bin' <Guest2041>its frustrating because 50% of my drive isnt even used lol <Guest2041>now it works that i ran guix gc and there is less stuff in /gnu/store <Guest2041>but i dont understand why even tho /gnu/store is mounted on the same drive as / and that drive has 50% free space <nckx>‘Guest2041’: …was your /tmp mounted as tmpfs? That's where Guix builds things (including ISO images) before copying them to /gnu/store. <unmatched-paren>argh, frustrating! gnome-shell-extensions-app crashes, there's an issue about it on its bug tracker but it hasn't been touched in a year... i can't find any other instance of the issue online <unmatched-paren>i would just use gnome-extensions CLI, but it doesn't seem to be able to manipulate settings <nckx>You found an orphan? Pretty sure you're their mummy now. <nckx>You get to debug it and perhaps fix it or at least provide more info. <KE0VVT>Is there a guide on installing Guix GNOME on Trisquel? <AIM[m]>Does guix install add the packages to system config.scm? <lilyp>unmatched-paren: I have a patch for that already <unmatched-paren>if you wish, you can create a declarative config for your user, instead of using guix install, which helps if you want to move your system around <unmatched-paren>it's probably best for reproducibilty and system portablity to use a user manifest, since afaik there is no real way to move your `guix install` profile around <unmatched-paren>although i'm fairly sure there's a way to export all your `guix install` packages into a manifest <unmatched-paren>lilyp: can i put that patch in my channel so i can use it until it's fixed, please? <ytc>hello. how can i query the variables like `%base-services'? from REPL? <lilyp>It's free software according to the GPL3+ <unmatched-paren>it feels impolite to just take it tho... maybe i'm too used to nonfree software <KE0VVT>I'm glad Sefaria is a free software web program. <lilyp>Well, according to the GPL you can't just "take it", you have to properly attribute, [other clauses] and make it free according to the GPL3+ for everyone else :) <unmatched-paren>hum, didn't realize that the patch changes gnome-shell, not gnome-shell-extensions... i guess i'll have to rebuild everything that depends on gnome-shell? <KE0VVT>I update with "guix pull && guix package -u", right? *unmatched-paren is not surprised <unmatched-paren>ok wow i feel very stupid, i forgot to import any modules in my gnome.scm :P <unmatched-paren>i thought it was optional too, maybe that was changed at some point? <nckx>The bogus warning was only recently removed, but it was optional for a very long time before that. <unmatched-paren>weird, i pulled and reconfigured but my gnome-shell patch (reversioned as 41.0-fixed) wasn't applied... <nckx>How are you applying it? <AIM[m]>What's the difference between "guix upgrade" and "guix system reconfigure"? <unmatched-paren>guix upgrade updates your user packages, guix system reconfigure swops out your system configuration for the new one you gave it <unmatched-paren>so if you add a new service to your config, you'd need to reconfigure to add it to the running/default services <nckx>‘warning: diffoscope: complex expression, bailing out’ — guix style… tried. <unmatched-paren>nckx: i've just pulled and reconfigured after pushing it to my channel, that's all <nckx>Does guix show gnome-shell show the right version field? And does sudo guix show do the same? <AIM[m]>unmatched-paren: Thanks unmatched-paren <nckx>unmatched-paren: Oh… By the way, looking up by name != looking by variable. If you use ‘gnome-shell’ as a variable in e.g. your system.scm, it might be the (gnu packages …) one. <nckx>Mystery (probably) solved. <unmatched-paren>but i'm using the gnome-desktop-service-type to add gnome-desktop... i guess i'll have to make my own version of that, too? <AIM[m]>If I remove a package from config.scm and run the reconfigure, will the entire package and its unused dependencies go away? <AIM[m]>The dependencies being stressed in the above question <nckx>No, but if it's truly unreferenced by your current and any live previous generations, and you run ‘guix gc’, it should vanish from your store. <unmatched-paren>AIM[m]: the package will get removed from your system profile (so you won't be able to run it), but it will stay in the store until your next `guix gc` <lilyp>unmatched-paren: Change the gnome package used by it to propagate your own gnome-shell <nckx>Just reconfiguring creates a new generation, the older one is kept until you delete it. <kwjc>burn it with hellfire I say! <AIM[m]>unmatched-paren: Like stuff that gets installed along with xfce and all... <unmatched-paren>AIM[m]: there are seperate base-desktop and <de>-desktop services for exactly that reason, i think <f1refly>when specifying my own guix channel, can I tell it to depend on the %default-channels? <unmatched-paren>are you worried about stuff liku gdm, dbus and network-manager that are usually only needed for desktops? <AIM[m]>Btw, I'm not sure how libera chat works, I hope you are seeing the "In reply to Username.... Message Content" on your IRC... I'm using Matrix Client Element to chat with you guys on Libera Chat IRC.... <unmatched-paren>they're in their own base desktop service, while xfce is in its own xfce desktop service, so there's nothing to worry about i'd think <AIM[m]>unmatched-paren: Yes, all gui stuff like gdm, file manager, terminal emulator, utils and all <unmatched-paren>AIM[m]: you should be able to install a terminal emulator or file manager in your user profile using `guix install`, then reconfigure won't make it disappear when you remove xfce <coventry>If I run the guix qemu image from the downloads page, and immediately invoke `guix pull`, I get an error. Is this expected behavior? It doesn't happen if I first `guix install emacs`, as in the manual. <unmatched-paren>i'm pretty sure almost all x wms depend on xterm tho, so you need not worry about that <lilyp>I think we should just rename java-log4j to java-log4shell :) <unmatched-paren>AIM[m]: do you want to keep the xfce terminal and file explorer, then? <AIM[m]>I'm thinking of something like alacritty or urxvt <AIM[m]>I wanna like make a good config.scm to upload to gitlab also, that's why I'm currently focusing to making my config.scm perfect <unmatched-paren>AIM[m]: that's a good idea, in case something goes wrong and you have to reinstall all you'll have to do to get your system back is reinstall with the image, go to gitlab, download the config, and replace your new conig.scm with the one from gitlab <unmatched-paren>guix, i'd say, is actually safer than debian, while being as unstable as arch is; the best of both worlds \o/ <unmatched-paren>in debian, updates can go really wrong and leave your system broken, but in this scenario in guix, you just have to rollback in grub <KE0VVT>unmatched-paren: Huh? I did roll back in GRUB. <KE0VVT>unmatched-paren: I did try the rollback before. It was sweet. <KE0VVT>unmatched-paren: I accidentally uninstalled nss-certs, and I rolled back to get it back. <unmatched-paren>KE0VTT: yeah, it's really freeing knowing that you can do all sorts of terrible things to your os and most of the time you can just revert it <KE0VVT>Does "guix pull" seem slower than usu.? <unmatched-paren>(it's also freeing to know you have no proprietary software on your machine (except for in the ROM, if you're one of those unlucky people whose laptop is not eligible to be upgraded to libreboot :( )) <AIM[m]><unmatched-paren> "AIM: that's a good idea, in case..." <- Yep, I have all my configs on gitlab... I just clone it and copy them over.... I used to take time installing stuff and setting it up... Now it's all automated with bash scripts and config files that I wrote on my own... Configuring a newly installed system is just a matter of minutes now, while it took hours before that... I wish to do the same for guix... Using Guile Scheme, I expect to <AIM[m]>simplify my automation procedure... <unmatched-paren>i think invoke's the guile function for calling an external program like in a shell <unmatched-paren>for when you don't care about the name, you just want to illustrate a concept <unmatched-paren>btw, about your earlier question: no, irc has no reply functionality, it's kinda primitive but it makes it nice and simple <AIM[m]>unmatched-paren: Yeah Yeah, I'm a computer science student, I do know what foo, bar and foobar is... But wanna know what foo calls in guile scheme is... <nckx>They teach foo and bar in computer science now? <coventry>I think my qemu failure was probably just due to the small memory allocation I'd given the VM. <AIM[m]>nckx: Not in uni but in reference books <unmatched-paren>huh, weird, (invoke "ls") doesn't work, i saw it in a guix recipe at one point <nckx>It is not built-in Guile; that's system and system*, but we don't use them. <nckx>Because they don't throw exceptions, so you'd have to check each return value. <nckx>And we know how good humans are at that. <nckx>*throw exceptions on failure. <unmatched-paren>one of the good decisions rust made: you have to unwrap values explicitly from a Result enum if they have a chance of failure :) <vivien>Plus, you can use a generic logging system to take care of the exceptions. We know how good humans are at logging things securely :) <lilyp>Though in case you're coming from an OO background like Java or C++, Maybe/Either is just try/catch with extra steps <unmatched-paren>(ftr, for anyone who doesn't know, 'enums' in rust are actually tagged unions) *nckx hasn't touched Rust with a ten foot rust-proof pole. Yet. <lilyp>also, of course there's an SRFI for maybes ;) <coventry>Maybe/Either is completely different from an exceptions-based system. It's based on a return value, not flow control. <lilyp>okay, but what do they do if you ... say dereference a maybe and it has no value after all? <vivien>I’m not a fan of the "either" workflow, because if multiple things can fail at unrelated places but you treat these failures in the same way, duplicating the error handling code is useless <unmatched-paren>lilyp: wdym? do you mean 'dereference' as in 'get the value inside'? <AIM[m]>We have Scheme in one of our Uni courses which was an optional (elective) course... I haven't picked it yet... Maybe I'll try it next semester or something... I didn't know what it was when I heard of it for the first time.... Now I wanna pick that course... Lol... <coventry>vivien: It's the old tension between explicitness and concision. I generally like my control flows to be explicit and local, but it can result in more code. <unmatched-paren>you can't access the value inside a Maybe-style thing without accounting for errors, matches in Haskell and Rust must cover all possibilities <unmatched-paren>you _can_ do an `if let` thingy in rust to get the value without accounting for errors, but it's really, *really* bad proctise for obvious reasons <unmatched-paren>`if let Ok(x) = Ok("hello guix!") { println!(x) }` prints 'hello guix!' <unmatched-paren>generally, with Result values, you stay well away from if let, and use possible_error.unwrap_or_else() or a match <unmatched-paren>really, you're supposed to use 'else' with 'if let'; i wish they enforced it <unmatched-paren>there looks to be a call to (gnome) in gnu/services/desktop.scm, i guess that's what populates the packages? <rekado>katco: could be. This is due to -Werror. Don’t know if that’s indicative of a real problem. <lfam>apteryx: Are you waiting to commit anything to version-1.4.0? Or is it okay fix things like the glibmm / libsigc++ issue? <opalvaults[m]>Is it common for herd service term-auto to be unable to be started? <opalvaults[m]>When I guix system reconfigure, it's never able to be restarted even manually <opalvaults[m]>The warning message kind of implies that the upgrade isn't complete until it is restarted however <opalvaults[m]>To complete the upgrade, run 'herd restart SERVICE' to stop, <opalvaults[m]>upgrade, and restart each service that was not automatically restarted. <opalvaults[m]>That's how I interpreted it, anyways. It doesn't give any feedback that it's not necessary. <katco>rekado: i'm unsure of what to do here. do we know when this broke? how is this not a more visible issue? (lots of things use autogen, right?) <lfam>Yes, it's a "user experience" bug <lfam>opalvaults[m]: Some advice: to fully complete a "reconfigure" upgrade, you should reboot. It's definitely intended that the system can be upgrade while running, but in reality it's not really possible to do 100% without rebooting <lfam>You might have issues with stateful services, things like that <lfam>If you are concerned about completed the upgrade, then reboot <lfam>I mean, about completing the upgrdae <opalvaults[m]>lfam: will do. I've been running into an issue where Gnome crashes on login so I'm probably not going to do it just yet as I keep having to roll-back until I figure out what the issue is <lfam>I think that has been reported <opalvaults[m]>Actually become a huge issue before I switched over here, thinking it was strictly a Fedora issue :P <lfam>Well, that makes me feel a little better <lfam>If they can't figure it out, I guess that we are in good company <lfam>It's very challenging to make a GNOME system <lfam>It's extremely complicated <lfam>Ugh, I am having trouble upgrading xorg-server without changing the derivation of xorg-server-for-tests <lfam>It seems like the ordering of inputs changes in the derivation <opalvaults[m]>Well if you're a contributer/maintainer of the Gnome package for Guix,thank you very much for all of your hard work :) <lfam>But they appreciate your thanks, I am sure! *opalvaults[m] yells into the void to the Gnome package maintainers, thanking them <KE0VVT>I installed WeeChat with Guix on a foreign distro. Connecting to Libera Chat is failing with an SSL error. What can I do? I tried installing nss-certs. <lfam>You need to figure out exactly what the error message says <opalvaults[m]>Is there a setting for pointing weechat to whereever your foreign distro keeps SSL certs? <lfam>Then, you need to interpret it, and then figure out how your environment differs from what is expected <singpolyma>KE0VVT: on my system I have to set SSL_CERT_DIR to /etc/ssl/certs iirc to make ssl work in guix stuff <opalvaults[m]>otherwise it might be the environmental variable SSL_CERT_DIR <KE0VVT>"SSL_CERT_DIR=/etc/ssl/certs/ weechat" produces the same result. <lfam>Did you figure out what weechat needs yet? <lfam>Is there anything in /etc/ssl/certs? <opalvaults[m]>export SSL_CERT_FILE="$HOME/.guix-profile/etc/ssl/certs/ca-certificates.crt" <opalvaults[m]>You may need to log out and log back in, you could also try rebooting. Weechat is probably a program that looks in a very specific place for its certs and as I've been made aware recently that location is not very uniform or standardized. <KE0VVT>opalvaults : DId those in a terminal and started WeeChat. Same result. Will restart desktop session and try again. <ngz>It's better to put them in .profile than .bashrc <opalvaults[m]>.bashrc will only make them available to your shell, whereas profile related files will be available to your user profile, and be extension the applications. (as far as I understand it anyways) <nckx>There should be a no-bot that just triggers on mentions of ‘.bashrc’. <nckx>opalvaults[m]: It will forcefully overwrite them in *all* shells. <lfam>Erm, the xorg-server-for-tests derivation changed since commit 94836b215630ad0f4a7c06b8e4284d923ac1197f <lfam>Seems surprising, right? <lfam>That's after the core-updates merge <lfam>Did we accidentally make another big change since then? Or am I missing something? <nckx>You didn't have to do anything. You made your own bed. <KE0VVT>nckx: This is the only way I could get tmux to run! <lfam>It's not a bid deal VE0VVT. You can move them into ~/.bash_profile or ~/.profile, do `bash --login`, and you'll get the same result <lfam>I mean, "it's not a big deal" <KE0VVT>I'm just trying to get things to work right now. I'll pretty things up later. <KE0VVT>Really, I'm just using Fedora until that bug is fixed. <opalvaults[m]>anyone have any luck getting extensions-gnome.org extensions to work? it's complaining that it needs the connector. Looks like the requirements aren't anything that Guix doesn't have..? <ngz>opalvaults[m]: Could you describe how gnome crashes? <nckx>lfam: I thought it might be imake but it doesn't seem to be… <lfam>I don't have much time today until later <lfam>I'm not missing something? <lfam>I don't see any massive rebuilds on the CI <nckx>I dunno, I'm taking your ‘it shouldn't have changed’ at face value. <lfam>Except for the merge of course, which still has ~30k pending jobs <lfam>Like `guix build -e '(@@ (gnu packages xorg) xorg-server-for-tests)' -d` *nckx also checked the CI with no suspicous results. <opalvaults[m]>ngz: It crashed on login via GDM into the "Oh no, Something has gone wrong and Gnome can't recover" with only the "Log Out" button available. I didn't see much in the ways of errors in dmesg or anything. I rolled back and it was fine. I just did a system reconfigure about an hour ago and that appears to have fixed the issue? <opalvaults[m]>raghavgururajan I'm not loading any extensions at the moment. I was trying to get them to load in via Icecat (the extensions.gnome.org site has a toggle switch). I'm going to attempt to build the connector. <opalvaults[m]>If it's a package that I can build I'll try and figure out how to contribute the package upstream. <lfam>Maybe I am overestimating the value of the derivation store path <lfam>I get a different result on berlin, too, with a commit from yesterday <nckx>Is it /gnu/store/cx3n9njki6w4qdd806rs1r982npa1b6n-xorg-server-21.1.1.drv ? <opalvaults[m]>if Cmake needs /usr, would ~/.guix-profile/usr be functionally equivalent for my purposes? <lfam>nckx: That's current value, on my laptop <nckx>opalvaults[m]: /usr/foo → ~/.guix-profile/foo <nckx>I'm travelling backwards in time. <lfam>Oh, on berlin, the difference is caused by the Guix fallback variable thing <lfam>It changes derivations but not output <lfam>nckx: Are you bisecting? <lfam>I was going to bisect using berlin <nckx>Berlin would have been a faster choice ☺ <lfam>Alright, I'll let you do it. Like I said, I don't have much time today <lfam>It shouldn't take many steps to bisectr <nckx>I'm slightly more hesintant to ssh into berlin for random things lately. <lfam>It's not so suprising that by exercising 96 cores fully for years at a time, we'd hit a CPU bug eventually <lfam>I'll wait for your results "off the air", as we say in the US <nckx>I don't actually know what that means here… AFK? <jgart>Hi Guixers, here are some of the patches from today's packaging meetup: <lfam>Yes nckx. It's from people who call in to radio stations and ask a question, and then get off the phone. They wait for the answer "off the air" <lfam>I'll read the log for your results <rekado>(re CPU bug: we have 96 CPU cores there.) <lfam>Yeah. We'd find the bugs eventually rekado <jgart>We'll send the java patches from the room roptat moderated soon :) <lfam>I'm grateful for your efforts at diagnosing the incident rekado *nckx suspects proprietary poopware more, but we can all share, because there are plenty of hardware bugs to go around. <raghavgururajan>Anyone know how to import modules from guix-channel via `(use-package-modules foo)` instead of `(use-modules (channel packages foo))`, inside config.scm? <opalvaults[m]>nckx: so if a package needs /etc, am I safe in assuming that I just need to change the makefile to be pointing to ~/.guix-profile/etc, and so on and so forth for each root level directory? <singpolyma>opalvaults[m]: you can't assume ~/.guix-profile exists, in general <nckx>opalvaults[m]: That means it will only work when the user installs it to their default user profile (not system, or an environment, or…), it's really not great. That said, there are a small number of packages in Guix for which we do that because presumably it was too hard to do better. <opalvaults[m]>singpolyma: interesting point. is there a good resource I can read to learn what the best practices are for building packages? <nckx>It's a very suboptimal non-solution. Investigate all other ones first :) <opalvaults[m]>the reason I ask is Cmake is throwing CMake Error at cmake_install.cmake:63 (file): <opalvaults[m]> file cannot create directory: /etc/chromium/native-messaging-hosts. Maybe <raghavgururajan>opalvaults[m]: Is your package's makefile looking for a file in /etc, that is provided by another package? <singpolyma>opalvaults[m]: in general the rule is that if you need a file you should have the exact path to the version you need in the store <KE0VVT>I guess I have to restart my desktop session to see newly-installed apps. :-( <nckx>opalvaults[m]: Why does it [claim to] ‘need /etc’ in the first place? <singpolyma>You need to set it to install to your output directory <KE0VVT>Unless I also need to put something for ~/.local/share/applications/ in .profile too. <opalvaults[m]>maybe I'm making it more complicated than it needs to be then <KE0VVT>Surely, singpolyma. I believe Mumble uses it. <opalvaults[m]>cmake -DCMAKE_INSTALL_PREFIX=~/.guix-profile -DBUILD_EXTENSION=OFF ../ <nckx>It's probably the software making things complicated. Upstreams like to challenge us packagers. <opalvaults[m]>and then I did 'make install', assuming that sudo would be the wrong route to take for that <raghavgururajan>opalvaults[m]: We usually patch the /etc or /bin references with absolute references to another packages' [output]/{etc|bin). <singpolyma>Rule of thumb I go by is that a package should work in a world where profiles don't exist at all <singpolyma>And the profiles are just for convenience/gc roots <nckx>raghavgururajan: Or its own if it's an installation error. <nckx>That's why ‘why’ matters. <bdju>barcode package failed to build <raghavgururajan>opalvaults[m]: For examples, xfe and spacefm. `guix edit spacefm` or `guix edit xfe` <nckx>You were right, it's just that there are multiple scenarios. <bdju> http://ix.io/3INd log here for barcode build failure. leaving the house so can't provide more info for a while <opalvaults[m]>ah, okay raghavgururajan . I've never packaged anything before so this is a bit of a cliff. I'll need to do a bit more hacking. <singpolyma>opalvaults[m]: are you packaging something that builds with cmake? <singpolyma>opalvaults[m]: are you using cmake-build-system ? <raghavgururajan>Yeah, we test the package inside pure environment. `guix environment --pure --ad-hoc package` <singpolyma>opalvaults[m]: it's for your package config, not a binary <raghavgururajan>opalvaults[m]: Could you share the URL for that connector application? <singpolyma>opalvaults[m]: in your package definition what build system did you put *raghavgururajan is triggered by yet another gnome-related package <opalvaults[m]>I don't have a definition yet singpolyma , I've never done this before <opalvaults[m]>ah, that is the missing piece of my puzzle. okay well then it's obvious i'll need to rtfm. <singpolyma>Yeah. And many many examples in the guix source tree :) <opalvaults[m]>Excellent, I'll take a look, thank you everyone for helping me :) <raghavgururajan>opalvaults[m]: Caution! It seems that application is used to install stuff via browser. It may install stuff in FHS locations, which won't be picked up by GNOME installed via Guix. <lfam>sneek: later tell bdju: I fixed barcode <lfam>so-called Filesystem Hierarchy Standard <opalvaults[m]>Can that problem be fixed easily? I think that the "need" for the package is pretty low for me. :P *raghavgururajan also suspects FSDG incompatibility <opalvaults[m]>but it might be useful for the Gnome Extensions side of things as far as Guix goes as it's kind of the standard for installing extensions these days right? <singpolyma>Ideal way to get a gnome extension bin guix is to package it for Guix <singpolyma>Other ways can be made to work I'm sure, tradeoffs and all that <opalvaults[m]>Ahh, yeah. If it tries to install to root level directories though it's kind of fruitless as that's a change that'll need to be wrangled upstream <opalvaults[m]>that's IF, if that is indeed where it tries to installs extensions via the browser <singpolyma>I have to run it with env PYTHONPATH= gajim or my guix profile makes it crash, haha <raghavgururajan>Ah. That might be the reason why you didn't have the issue I had. Older versions of gajim were faster that newer ones. <singpolyma>Could be. I'm in Debian stable version, so new to me as if August <KE0VVT>OK, time to restart graphical session. Hope internet works in ||WeeChat. <apteryx>lfam: hi! I have a good set of changes for the branch, including the glibmm thing <apteryx>I should have more time to build more of it tonight and hopefully push it <KE0VVT>I restarted my graphical session. WeeChat still has the SSL issue. <nckx>lfam: ‘Oh’, it's ‘just’ d080368cb5 ‘duh’. <nckx>lfam: From a quick non-exhaustive look at both .drvs, it modifies the order of inputs? <nckx>Be back later if more scrutiny is needed. <singpolyma>I'm just being a dick. Weechat is probably a fine piece of software <KE0VVT>singpolyma: I'm still getting SSL errors in WeeChat, even after installing nss-certs and setting env. vars. <KE0VVT>...and restarting my GNOME session. <singpolyma>If you look at the weechat pid in /proc do you see the env var as you expect? <KE0VVT>singpolyma, no `weechat` in `/proc/`. <KE0VVT>singpolyma, I cannot find WeeChat with ps. <KE0VVT>So, Telegram Desktop gets internet, but WeeChat gets SSL errors. <unmatched-paren>i think we should split `gnome` into `gnome-core` and `gnome-apps`, then make `gnome` depend on both. that way it's easier to choose which apps you install; for example, i have no need for gnome-weather, but `gnome` depends on it <unmatched-paren>then we could have an `apps` extensions for `gnome-desktop-service-type` that adds the apps <lilyp>You can delete the respective propagated input, so that'd be a step backwards actually <KE0VVT>singpolyma, OK, what dir in /proc/<<PID>>/? <unmatched-paren>lilyp: how would i do that? i'd like to add a `gnome-minimal` to my channel along with your patch <lilyp>(package (inherit gnome) (propagated-inputs (modify-inputs (package-propagated-inputs gnome) <your code here>))) <nckx>KE0VVT: It's nul-terminated, so | tr '\0' '\n' for sanity. <nckx>But looks like the variables are set <jacereda>Hi, I have (specification->package+output "git" "send-email") in my package list, but it doesn't seem to work... what am I doing wrong? git shell git:send-email works fine <nckx>I get that same error in weechat by the way <nckx>23:11:29 libera =!= | gnutls: peer's certificate is NOT trusted <nckx>I doubt it has anything to do with bad configuration. <nckx>jacereda: Overthinking? :-) "git:send-mail". <jacereda>nckx: using specification->package diretly <nckx>The bug is you're passing 2 package names (git, and send-email) instead of one package:output spec. <jacereda>guix system: error: git:send-email: unknown package <jacereda>using (specification->package "git:send-email") <nckx>Why did you drop +output? <KE0VVT>SSL_CERT_FILE=/home/caleb/.guix-profile/etc/ssl/certs/ca-certificates.crt <KE0VVT>Not using WeeChat is kinda a deal breakrer. <nckx>weechat.network.gnutls_ca_file ? <jacereda>nckx: I tried with and without +output, none works <nckx><Not using WeeChat is kinda a deal breakrer.> As is not using SSL, apparently. <KE0VVT>Agh! 5.fset │ weechat.network.gnutls_ca_file string "/etc/ssl/certs/ca-certificates.crt" <nckx>jacereda: You can't get that error if you'd used +output… <jacereda>(define* (specification->package+output spec #:optional (output "out")) <nckx>KE0VVT: I'd like to thank ‘random search engine at home’ and my many fans. <jacereda>nckx: if I read that correctly, it receives the package and the output separately <jacereda>ok, I read the docs and the way you say should work <jacereda>but somehow, after system reconfigure, git send-email isn't working <nckx>jacereda: s->p+o returns two outputs, not one and/or a list, which is kind of ‘unusual’. ***KE0VVT_ is now known as KE0VVT
<nckx>Chalk this up too ‘weechat being a little non-environment-respecting ███. <nckx>So depending on the context you might need something like (compose list package->specification+output). <nckx>It gets Schemey pretty fast. <f1refly>when I have multiple versions of my guix I can choose from at boot, and I broke a thing some revisions ago, is there a way I can make guix print out the configuration.scm of an older version? So I know how I did it before? <taterbase>I've successfully compiled libreboot on my x200s with Guix as my OS! <unmatched-paren>f1refly: i think you can find old profiles in /var/guix/profiles/*/, but it sounds like you've already found it <f1refly>I think the currently booted profile file should be /run/booted-system/configuration.scm, right? <f1refly>so when I'm in my still-working previous revision I can just check what's in that file <unmatched-paren>f1refly: your old ones are in /var/guix/profiles/system-*-link/configuration.scm <f1refly>most curious, i think the only change was that I added the ibm plex font to my global package list <f1refly>Can this happen because I added a repo in my filesystem and removed it again? <jacereda>in case anyone is struggling with git send-email in config.scm, the easy answer seems to be `(package (cons* git (list git "send-email") rest...` <florhizome[m]>unmatched-paren: jgart has opened a channel over on sr.ht that’s supposed to be a general community channel for alpha/beta quality and odd packages. I find this idea quite good and would rather push wayfire and my other stuff up there, if you don’t mind. :) 
