IRC channel logs
2024-11-15.log
back to list of logs
<PotentialUser-18>PotentialUser-18 i did a `find` on the root fs and it looks like the `guix pull` and `guix package -u` didn't install an updated guix, but it did install `.guix-profile` in the users home directory (pointing to the wrong on for root) <unmush>current-guix is your 'guix pull' profile; it holds the guix you want to actually run. 'guix-profile' is the default profile that package management stuff (guix install, etc) uses. <unmush>example: 'guix install emacs' --> emacs is now in ~/.guix-profile/bin, 'guix pull' --> updated guix is now in ~/.config/guix/current <unmush>note that technically you can 'guix install guix' last I checked but it doesn't work well at all ('guix pull' exists for a good reason) <PotentialUser-18>also, the permissions are broken now because when i su'ed to root it installed everything in the $USER directory with root permissions <PotentialUser-18>if i su to root, it doesn't run /etc/profile.d/guix.sh, so it points to the system guix, so i have to manually source it to point to the guix installed with guix, which it sounds like im not supposed to do? (even tho its an option) <unmush>to be clear, it's just that you're not supposed to 'guix install guix', you're supposed to install newer guixes with 'guix pull' <unmush>the reason there is a publicly-visible guix package is so that you can do stuff like 'guix environment guix' <unmush>(or whatever the 'guix shell' equivalent is these days) <unmush>which will look at that publicly-visible guix package for purposes of figuring out what packages you need to hack on guix <PotentialUser-18>the problem on debian is that `guix pull` doesn't update guix, because its managed thru apt <PotentialUser-18>and im guessing each guix pull is independent? because running it as $USER pulls the entire repo again <unmush>'guix pull' updates the invoking user's 'guix pull' profile (~/.config/guix/current, which is itself a symlink into /gnu/store/...) <unmush>so it should be fine to run with the system-provided guix <unmush>in fact on guix system, this is how users get their first 'guix pull' profiles: they run 'guix pull' using a guix from /run/current-system/profile/bin/guix <unmush>probably, I haven't used guix on a foreign distro in many years but I don't see why not <unmush>as long as it doesn't nuke /var/guix and /gnu/store also <PotentialUser-18>im wondering if i should just use the shell script and skip apt's install <unmush>the shell script does sound like it might be a rather lower-pain solution at this time <unmush>no trouble, thanks for being persistent <PotentialUser-18>i got everything working previously and then gnome broke and it meant i couldnt do anything i actually wanted to work on <unmush>how does one distinguish between two input packages with the same name in a phase? <mange>I think the usual advice would be "don't have multiple inputs with the same name". What are you trying to do? <unmush>build a package that depends on a slightly older version of itself and also a much older version of itself <unmush>its build system insists on copying some files from the much older version of itself <mange>Can you pass the older version as an input with a different name? Either using a list to give the input a name, or by creating a new package that inherits and renames? <unmush>if I create a new package then there's a redundant build, and I'm not sure how you use a list to give an input a name <mange>If you *really* need it to have the same name, then I expect you could either reference the older version directly in a gexp (which has other downsides), or you could iterate over all the build-inputs to find the one you want yourself. <unmush>I know how it was done before gexps, but not sure if it still works that way <mange>It does still work, but it's not used as much any more, and people don't usually like to mix the two styles in the same inputs list. You can look at extempore as an example package that uses the old style (just the first I found, I know nothing else about this package). <unmush>says the package has an invalid input <unmush>I think it probably doesn't work to mix the two styles due to how the conversion from new style to old style internally works <mange>Ah, yeah, looking at add-input-labels in (guix packages) it seems to decide what to do based on the first one in the list. Alas. <apteryx>is there no updater for go packages? <carloratm>Hi all, I am using a wrapper on top of nix as a tool to manage all of my dev projects at work, I am now considering switching to nixos or guix as my main OS at work. I am now using Debian stable. I have a few questions regarding guix <carloratm>What are my options if a package is not present? In particular, solr. I use that a lot at work. <Franciman>you can check other channels or package it yourself <dariqq>hi apteryx, any update on the lightdm situation? <unmush>does anybody have any experience with bootstrapping microsoft's .NET stuff? I've got mono bootstrapped up to 4.9.0 so far and am wondering about how to get started on the microsoft stuff as the mono version gets higher <futurile>Survey - we've had 509 full responses to the survey, so that's great. It's slowing down now - so if you haven't done it yet please take it! <futurile>We have had a lot of incomplete ones - some hit the first page and drop out; but there's also quite a few who get about half way through and probably think "oh this is too long!" - which I have some sympathy with - it's long <futurile>but putting the notes on the last page that you have to hit the Submit button has helped a lot! <civodul>i realize there’s no deadline mentioned in the blog post <civodul>maybe we should re-advertise it and set a deadline? <sepeth>Hi Guix! On my VM, I don't see the grub menu after I did guix system reconfigure. This has happened a few times before but I got the bootloader back after booting via UEFI shell and running guix system reconfigure again. This time, no luck. Does anyone know what could be wrong? Here's what I see in the output: <sepeth> (bootloader (bootloader-configuration <sepeth> (bootloader grub-efi-bootloader) <sepeth> ;(keyboard-layout keyboard-layout) <sepeth>building /gnu/store/lgwbkp8ha3dw271ispxx2g7lqg5lrx7p-install-bootloader.scm.drv... <sepeth>guix system: bootloader successfully installed on '(/boot/efi)' <Rutherther>sepeth: do you see grub at all, like the rescue line? <sepeth>But once I am in UEFI shell and do fs0: then EFI\Guix\grubaa64.efi, everything works as expected. <Rutherther>sepeth: have you tried adding the boot entry with bcfg boot add in the uefi shell? I am not completely sure when guix tries to add the efi to the boot order, maybe it just doesn't run on system reconfigure, but only on system init or something like that <sepeth>Rutherther: I have not! Thank you ^-^ Trying that now. <sepeth>Rutherther: it now works ^-^ but didn't got a chance to try that yet. It must have been fixed when I ran reconfigure 3rd time round to paste the message here. <apteryx>dariqq: I haven't taken the time to try it yet, it's on my TODO though! sorry for the delay. I'm multitasking with a single thread. <janneke>what package provides the "docker" command? (for docker load, as per manual)? <Rutherther>janneke: docker-cli package, but if you want to use docker load that seems like you should be running the docker service, and at that point you should have docker-cli already installed in system profile <janneke>Rutherther: i don't want to run docker, i want to provide a docker image <janneke>thanks, so possibly i need to run a docker service in order to actually provide/create an image? <Rutherther>janneke: the thing is that "docker load" you asked for specifically is for loading an image into your docker registry. It's not for creating images <janneke>i would have thought the manual would describe creating an image there <Rutherther>janneke: docker build is usually used for building images, not sure if any other command supports it. I am not sure if that one works without docker service running, my guess would be it works by offloading the work to the docker instance, so I would guess you need docker service. But I might be wrong <Rutherther>janneke: I see. guix pack is the command that builds the image, you do not need anything else to build it. docker load is to load it into a docker instance, so that you would run a container from it <dariqq>apteryx: dont worry, glad to hear it made it onto yo just dont want it to get lost (and the status quo is not very good)ur todo list, i <dariqq>whoops something went wrong here <Kabouik>Does anyone know of an alternative to https://github.com/babariviere/guix-emacs? The channel is now unmaintained and there is no way to post issues or contact the former maintainer to know learn how to revive it. I had many emacs packages depending on it because they were not in main Guix channel, and now guix pulling is broken unless I remove all my Emacs packages and break my Emacs config. <Kabouik>The thing is I don't really understand Scheme and am afraid I'm not skilled to make the update script work, but you're right I should definitely spend more time on the doc, see if I could do it myself <janneke>scheme@(guile-user)> (lzlib-procedure '* "LZ_decompress_open" '()) <Franciman>hi, i'm updating my guix system. And i noticed that it is installing telegram-desktop-5.5.5-checkout <ieure>Franciman, You usually see that when something is built locally, either because substitutes aren't available or because the package isn't substitutable. <ieure>Franciman, Sometimes when that happens I just C-c and wait a bit. <ieure>Franciman, `guix weather telegram-desktop' will let you know if substitutes are available. <ieure>I also set up a remote build machine. My laptop is able to build most stuff reasonably, it's just a lot nicer for the machine not on my lap to get warm. <Franciman>eheh true. What specs do you have on your lappy? <ieure>Slowest I use regularly is an 8th gen Core i7 with 16gb. <Rutherther>Franciman: I am quite surprised you would see that the checkout is being installed. It should just be downloaded(built), it's the source of the package <Franciman>but is there a way to know why a substitute is not available? <Franciman>do we have public logs on the various substitute servers? <ieure>Franciman, Have used one 10th gen, it's okay. My personal opinion is that 8th is the first truly good Core CPU for laptops, 9th is meh, 10th is okay, 11th meh, 12th bad, 13th-14th terrible. <ieure>12th get too hot and are battery hogs (in a laptop), 13th and 14th shipped with faulty microcode that permanently damage the CPU. <ieure>Intel is not doing great these days. <ieure>Franciman, My daily driver is a Ryzen 7 5850U, it's very good. <ieure>Franciman, ThinkPad X13 Gen 2 AMD. <ieure>Franciman, They're out there, still a bit expensive though. <ieure>Franciman, Sadly, RAM is soldered on those. So I got the 32gb model. <Franciman>i am currently on a t440, with 4th gen cpu lol, the battery lasts 4 hours, but the lappy is a rock <ieure>Yes, recent models haven't impressed me. I like and use the trackpoint though. If Framework ever gets it together to make a trackpoint keyboard, I'll probably switch. <ieure>I've built a couple T440p machines just for fun. <cehteh>many years ago i made a trackpoint emulator for touchpads, but once i got a laptop with real trackpoint again i didnt maintain it and eventually it became removed from xorg <cehteh>wasnt terribly ergonomic either because trackpad position under the keyboard <cehteh>if there is no trackpoint i could prolly compromise with a touchscreen but trackpack feels somewhat unnatural to me, i never use it, either keyboard nav or external mouse then <ieure>I dislike touchscreens on laptops, just feels very awkward. <cehteh>yes theyre not the premier solution, but for some things useful, i notice when using a tablet with a keyboard flipcase that its not *that* bad <cehteh>i rather point up to the screen to manupulate something than moving the hand down to the touchpad <Guest60>hello, I'd like to use in Emacs and connect my Scheme buffer to it, but if I do e.g. and then use in Emacs, the window just becomes unresponsive after bringing up a blank REPL buffer? <Guest60>that should say "guix repl --listen=tcp:8080" <Guest60>I am running geiser-connect after starting the REPL process <ekaitz>efraim: could you please do me a favor? <ekaitz>efraim: could you please run `make check` on guile's codebase in RISC-V? <meaty>whenever I try to "guix pull" I get a bunch of messages about postgresql-11 being deprecated and it says "nothing to be done" even though I haven't pulled in 10 days, what's going on? <meaty>*I haven't pulled since I installed the system <Rutherther>meaty: everyone gets that now. You can just ignore it <meaty>Rutherther: has there really been no new patches in 10 days? <Rutherther>meaty: what do you mean by that here? new patches have been open even today. Newest commit was just a few hours ago. <meaty>Rutherther: I mean is it normal to get "nothing to be done" when I last pulled on nov 5? <yelninei>meaty: what commit is in 'guix describe'? Maybe you pin to a commit in your channels.scm ? <meaty>yelninei: guix descrive shows guix commit 33445c5 <Rutherther>meaty: I don't think that commit exists in guix upstream repo. What is your channels.scm? <Rutherther>so did you men to write 33665c5? sixes instead of fours? <Rutherther>apart from that... that is the channels.scm that is used for pull? because if so, then it makes sense there is nothing to do, you have guix pinned to that commit <meaty>but the file cannot be written to, I presume it is derived from the "channels" field I wrote in my configuration.scm <meaty>let my try disabling channels.scm and reconfiguring from my configuration.scm <meaty>no dice, gives "Git error: could not find appropriate mechanism for credentials" <meaty>channels.scm was owned by me so I just disabled it and piped its contents to a new file and removed the pins <wph>Hey, is anyone running sway@1.10? I think xwayland is broken, but I'm not sure if it's only on my machine <nckx>ACTION no, but makes mental note not to upgrade.