***dgcampea-2 is now known as dgcampea
<sturm_>(because it fixes an anoying issue with Python REPL intergration) <singpolyma>sturm_: you can use the scheme equivalents. Or make a package variant with inherit <singpolyma>If you're used to using specifications->manifest you'll need to at least partly switch to packages->manifest <mange>sturm_: For the transformations, see "(guix) Defining Package Variants" in the manual. It has some examples of how to do it. <sturm_>awesome, thanks singpolyma and mange <cedb>is there a way to speed up the "computing derivation part" <cedb>even better, i have a better machine that runs arch, can I make it do the heavy lifting for the guix machine? <AwesomeAdam54321>cedb: Yeah, there's build offloading to build derivations on another machine <cedb>AwesomeAdam54321: may i have a crumb of tutorial? <AwesomeAdam54321>cedb: You can run `info guix "Daemon Offload Setup"` to jump straight to that part of the manual <SeerLite[m]>Would it make sense to patch a package so it can use a search path to load plugins? Like, patching Kakoune so that it looks at $GUIX_KAKOUNE_PATH to load plugins? <SeerLite[m]>Or Vim so it uses something like $GUIX_VIMRUNTIME, because I noticed that as of now its patched to look at a set of "hardcoded" paths and doesn't use Guix's search paths at all <cedb>`guix offload test` passes but i see nothing in top happening on the other machine when computing derivation <cedb>oh but /bin/gs is there when running `guix package -u`, interesting <sturm_>I'm duplicating the "emacs-next" definition and changing the commit and version number like this, but build seems to be getting stuck at git "'detached HEAD' state". Any ideas? <mange>Is it getting stuck, or just printing a message? I think git prints a message when you checkout a commit that isn't a tag/branch, but it's not actually a problem. When I tried to build that package definition it failed because of a hash mismatch. <sturm_>mange: arg, sorry I didn't read the error output properly. I was wondering why I hadn't seen a hash mismatch yet too hah <cedb>offload stopped working im getting : "the file does not exist or permission denied: "/root/.ssh/id_rsa"" <cedb>i dont understand since i changed nothing... <cedb>works after reboot, hmmm <karrq>hello everybody, is there anyone else getting permission denied for mkstemp using `guix refresh PACKAGE -u` ? ***jonsger1 is now known as jonsger
<apteryx>you need to use that with writable copy of guix sources, i.e. a git checkout <apteryx>after building it per the manual, you can use './pre-inst-env guix refresh PACKAGE -u', and it'll update the corresponding source <karrq>but how do I get the updated package in my system sources? I'm using guix on arch linux <karrq>because if I can only use if I specify to use the guix in the checkout I'm afraid it's not really usable ***robin_ is now known as robin
<karrq>oh btw, maybe you can also help me with this, when I run `guix info` it shows the manual in spanish and I can't get an english version... my locale has always been en_US.UTF-8 <xelxebar>karrq: It sounds like you are wanting build tranform options, not guix refresh? <sneek>xelxebar, you have 1 message! <sneek>xelxebar, lfam says: Regarding NTP and bare-bones.tmpl, Okay! I suppose that 68 years will be good enough for the time being <karrq>xelxebar: well from what I understoon guix refresh is to update a package to a newer version, in my example conan is 1.42 in the repo but 1.45 has been released already <cat[m]1>What are people here's thoughts on OpenIndiana? <xelxebar>karrq: guix refresh is mainly a helper for updating package definitions in the repo. If you want to install a newer release *right now*, then `guix build --with-commit=conan=1.45.0 conan` is probably your best bet. <xelxebar>FWIW, I just tried building it. Looks like tests are failing, so you might need to add --without-tests=conan (caveat emptor, of course). <jgart>HEYOOOOOOOOOOOOOOOOOOOOOOORG ***alMalsamo is now known as lumberjackimok
***lumberjackimok is now known as lumberjackok
<bdju>anyone know about video playback performance with intel iGPUs? does vaapi make a big difference? I might not have it. I'm dropping frames really bad on some videos but not others. wondering if there's any way to make things better <bdju>I have a haswell mobile i7 <bdju>is there a list anywhere of the best GPUs that work with linux-libre? <bdju>video playback, 3d perf, the usual stuff. not AI or whatever data scientists are doing if that's what you mean. otherwise I would think it's a pretty straight forward thing <bdju>gnoo: I am aware of h-node and it is sadly extremely unhelpful in nearly all cases... but thank you anyway <bdju>there is not much there and I'm not sure I can trust some of the few matches that are there. the filtering/sorting is also kind of messy <bdju>ideally someone would be able to tell me they had a specific one working on Guix System with linux-libre and they could say whether there were issues, how performance is, etc. ***anticomputer_ is now known as anticomputer
<EndOfAnEra>hi, how do you list the services in the %desktop-services variable? I'm using emacs-guix and geiser, but even with a REPL, can you list that variable somehow? <EndOfAnEra>in geiser, I see (guix-default-services) and (guix-all-services) interactive commands, but how would you list those in %desktop-services defined in my system.scm <elais[m]>Can probably do (display %desktop-services) <EndOfAnEra>;;; <unknown-location>: warning: possibly unbound variable `%desktop-services <EndOfAnEra>it's in the system configuration file, how would you find where this is defined ***dgcampea-2 is now known as dgcampea
<EndOfAnEra>seems impossible to develop guix without these basic logistics to even know what's going on <EndOfAnEra>(,m (emacs-guix)) (use-modules (gnu) (guix utils)) (use-service-modules desktop) (display %desktop-services) <EndOfAnEra>but one still had to guess their way with which modules to import, is it possible to "jump to definition" from the system config file? <elais[m]>When you’re a working in a file a good first step is to import all of the modules you’re using from the use-modules declaration. <elais[m]>You’d run into this problem in any other program where you are using imported objects and procedures <EndOfAnEra>elais[m]: ok. But after importing the modules, is the .scm file interactive in emacs-guix so that one can jump to definition? <elais[m]>You should be able to jump to the definition of anything imported into your repl environment <EndOfAnEra>elais[m]: you mean a command typed in the separate REPL window? (not the .scm file but the interpreter) <EndOfAnEra>normally in programming, you can jump to definition in source files, but I understand that you can run a REPL and run commands <elais[m]>When you evaluate something in the scm it should be loaded in the repl environment <elais[m]>Because the repl is what’s doing the evaluation <EndOfAnEra>it's more of a workflow question. Ok so I evaluate the systems .scm file in a separate repl window, then operate on it, if I understood correctly <elais[m]>I would just evaluate them in the buffer with `C-c C-c`. If the buffer is evaluated without problems the everything in the file should be loaded into the repl for further evaluation and use <EndOfAnEra>I can't evaluate my system.scm config file because it can't find third party modules, it requires (add-to-load-path) <EndOfAnEra>since it's from a /gnu/store/xyz-thirdparty/lib/guile/3.0/site-ccache/thirdparty.. <EndOfAnEra>is there a guix way of finding this path to add to my ~/.guile file? It doesn't seem right to add the path with the hash, or is it? <EndOfAnEra>I have a channel definition, that's the path in /gnu/store, that contains files I want to load in REPL, how would oyu add this to (add-to-load-path) with the hash <EndOfAnEra>I suppose I need to write code in ~/.guile that connects to the store to get it? Is that a bad idea? <SeerLite[m]>EndOfAnEra: You can just use `guix repl` instead of `guile`, should get all channels and modules in the load path automatically, at least for the system config <EndOfAnEra>SeerLite[m]: the thing is I'm trying to develop the config files in emacs <EndOfAnEra>and when evaluating the buffer, the use-modules isn't in the guile path <EndOfAnEra>I found it somewhere in /gnu/store/xyz-package from a channel added before <EndOfAnEra>ok it seems to run (geiser-eval-buffer) when evaluating, and it's not the REPL that is started with "(guix-switch-to-repl)" in emacs <EndOfAnEra>I guess, the procedure is in emacs to "(guix-switch-to-repl)" when editing system.scm, and then run "(connect-to-guile)" to connect to this REPL <EndOfAnEra>so now when evaluating the buffer, it is run in the appropriate guix repl <EndOfAnEra>thank you, but it feels like I'm doing something horribly wrong in the workflow that should be as simple as editing the system config <roptat>I'm trying to figure out how to query the graphql end point of mumi, but there's no doc for mumi nor for kolam, so it's not easy :/ <roptat>I keep getting errors that I don't have a valid graphql document <roptat>maybe that's not how it's supposed to work? <roptat>gah, I still get 500 errors... I don't understand what the query is supposed to look like <civodul>roptat: is that a followup to Arun's talk? :-) <roptat>kind of, I was trying to see if I could communicate with mumi to get some info and create some sort of review workflow <roptat>I'd be interested in usertags, but mumi doesn't have them, or at least it doesn't have an api to expose them <roptat>and the graphql api is a bit limited, so I might contribute <civodul>without going through graphql, you can already fetch patches <roptat>I think I found a bug already, I can't get the message-ids for bug messages <roptat>could you send me a "summary file"? I would like to see what it looks like <civodul>(i had to look it up in controller.scm) <ngz>In addition, you can get a specific version with /issue/NNN/patch-set/V where V is the version number, e.g., 2. <roptat>this is where I found the name "summary file" <roptat>should be in <data-dir>/spool/db-h/<id>/<id>.summary <roptat>civodul, do you have that for 41669? I'm wondering if it alread ycontains usertags or not <roptat>civodul, any ideas how these files end up there? <florhizome[m]>In general, would be great if those patch–sets would be linked in mumis UI so you can paste the link into your cli <civodul>roptat: no idea! this is all mumi magic i guess :-) <civodul>the code that does that is in mumi i suppose, or at worst in hydra/berlin.scm <roptat>that's what I thought but can't find it :/ <roptat>there's a worker, but it can only send email to the list (and it's disabled currently) <roptat>oh, that's index! in (mumi xapian) <SeerLite[m]>Hi efraim ! Were you the one who set up the default Vim config to search installed plugins? <SeerLite[m]>Any reason for not using Guix search-paths for this? I think it could be done by setting up a GUIX_VIMRUNTIME search path and then have Vim use that, instead of using the predefined list of paths like now <efraim>I'd have to look back to when I added it, I think there might have been a reason <efraim>off the top of my head I think it was because I figured we would need a wrapper for vim which would use the GUIX_VIMRUNTIME search path to construct the runtimepath and then pass that to vim <efraim>although I suppose THAT could have been a separate guix.vim to construct the runtimepath like guix.vim does now, but based on GUIX_VIMRUNTIME <SeerLite[m]>Heh, yeah I actually did exactly that, and it seems to work! <roptat>civodul, what does the associated .log file look like for 41669? <SeerLite[m]>I also thought we'd need a wrapper, but then I realized that since a guix.vim is used anyway then why not do it there directly <efraim>want to send in a patch? my hardcoded list was one of my pain points when I switched to using guix-home <SeerLite[m]>Yeah, does it look good to you? I also want to add the same search path and guix.vim phase to the neovim package, which currently doesn't use profile plugins like Vim <efraim>I'll test it out a bit first but it looks good to me <efraim>I might actually fix the TODO of removing duplicate entries also <SeerLite[m]>Alright nice! Not sure why there'd be duplicates in the first place though? I didn't look into that <efraim>at a minimum with your code we'd have a duplicate /gnu/store/...-vim-../share/vim/vimfiles <efraim>and at that point I'd rather deduplicate it in vimscript than in vim's source code <civodul>roptat: the .log file is a 1.2MiB thing <civodul>a three-line XML header followed by the raw messages <efraim>SeerLite[m]: do you have a copy of the vim sourcecode handy? take a look at src/os_unix.h, around line 379 for RUNTIME_GLOBAL <roptat>do you know how I can access the raw messages then? <SeerLite[m]>Nope I don't get it. Is that where you say the duplicates come from? <roptat>ah nevermind, usertags are not recorded with the bug itself <roptat>so I don't know how to access usertags, it would be useful <efraim>SeerLite[m]: that's where the default rtp is set from, if we don't add to it using --global-runtime <lifestronaut>I feel dumb for asking this, but how do I clear the terminal? "clear" doesn't work, and I cannot figure out for the life of me how to clear it <SeerLite[m]>efraim: Ok nevermind I get it now, for some reason I was thinking I was setting rtp instead of appending to it. <SeerLite[m]>lifestronaut: Most shells let you clear it with ctrl + L. If you want the clear command you can get it by installing the ncurses package <efraim>--with-global-runtime=\"$GUIX_VIMRUNTIME\" was too many quotes <SeerLite[m]>Would that work? Wouldn't the env var get evaluated at compile time? <SeerLite[m]>Sorry, I'm lost I don't understand C preprocessor lang haha <efraim>SeerLite[m]: it took $GUIX_VIMRUNTIME as a literal <efraim>it didn't work, it just added $GUIX_VIMRUNTIME as an actual path <SeerLite[m]>efraim: Did you actually get it to compile or did it error during the build? <efraim>it froze during the tests for me with --with-global-runtime so I turned off tests <efraim>when I tried it with just your code I only got the vim plugins inside of my GUIX_ENVIRONMENT <SeerLite[m]>efraim: Really? How are you testing it? I get everything from my profile or shell <efraim>after building it, './pre-inst-env guix shell vim vim-gnupg -- vim' and I see if it also picks up vim-airline <efraim>I have vim-airline in my .guix-home/profile/share/vim/vimfiles (and also in .guix-profile/share/vim/vimfiles for some reason) <SeerLite[m]>I got so confused for a second, and I still am I think <SeerLite[m]>But it's because you have to source $GUIX_PROFILE/etc/profile <SeerLite[m]>Yeah, if you install vim in your profile then the .guix-profile env gets updated with the search paths <civodul>SeerLite[m]: hi! hibernation is currently unsupported, though there have been patches floating around <dave[m]>Hello! I am attempting to install Guix System on my computer, but I am not having any luck with LUKS. After installing with `guix system init`, GRUB says something along the lines of "Device uuid not found" (I am not at the computer right this moment). When I try to use the rescue shell, commands like cryptodisk are not available. <dave[m]>I am connected through Matrix. If I upload a file will users on irc be able to open it? <civodul>dave[m]: hi! out of curiosity, why are you using "guix system init" instead of the installer? <civodul>normally "guix system" ensures all the file system UUIDs present in your OS config are valid <civodul>but then maybe GRUB is configured without LUKS support or something <dave[m]>civodul: I have my file system already set up and file on the home directory from a previous distribution. I did not see a way to this with the installer. <dave[m]>I feel like the GRUB without LUKS support is the problem, but I am not sure how to fix it. I am 100% sure the UUIDs are correct as they are pasted in from `blkid` and I double checked them. <dave[m]>Let me know if you are able to download that. Again, not sure about the matrix to irc bridge. <civodul>dave[m]: i think there's a bug: (swap-devices ...) falls out of the (operating-system ...) block <civodul>apart from that, assuming proper UUIDs, it LGTM <dave[m]>civodul: Good catch, thanks. I had that commented out when I installed the system because I could not figure it out. <dave[m]>No clue as to why grub would not have luks support with that? Should I be mounting the efi partition at /boot instead of /boot/efi? <phodina[m]>could you please suggest what is the correct way to write if statement in operating-system. I don't know how to use unqote operator corrrctly there <phodina[m]>```(if (string=? lts "lts") (kernel libre-linux) (kernel libre-kernel)) <civodul>phodina[m]: rather: (kernel (if (string=? ...) ...)) <singpolyma>Hey all, I'm having a strange problem. Trying to define a package using ruby-build-system with argument #:ruby ruby-3.0 and getting this error: has an invalid input: ("ruby" ruby-3.0) <phodina[m]>civodul: That works thanks but I need to add also some other stuff like kernel arguments and I'd like to have just single if statement <civodul>phodina[m]: oh i see; i think you'll need two 'if' statements <civodul>another option is to do (operating-system (inherit the-other-one) (kernel ...) (kernel-arguments ...)) <civodul>and then you have a single 'if' to switch pick between one and the other <phodina[m]><civodul> "and then you have a single 'if..." <- is possible to use if statements within the operating-system to enable or disable block of code? I'm asking as I want to share config between multiple machines. Inheriting configuration would be very tricky <phodina[m]>I haven't seen any such code in gnu/system/examples nor any gexps or unquotes <pinoaffe>phodina[m]: yes, conditionals are possible <pinoaffe>you could even write a function that gets a few options as its arguments and gives out a system config based on those arguments <phodina[m]>pinoaffe: ok, my struggle is that the conditional as is not evaluated ... what unquote or gexp should I use? <phf-1>Hello Guix! I was wondering: to what extent can Elixir/Phoenix be used with Guix? <phodina[m]>pinoaffe: Not sure what #~, #$ or , ` I should use <phodina[m]>(if (string=? lts "lts") (kernel libre-linux) (kernel libre-kernel)) <pinoaffe>phodina[m]: `(kernel (if (string=? lts "lts") libre-linux alt-libre-linux))` <SeerLite[m]>Is `search-input-file` recommended over `which` in the new gexp-like package definitions? Or are they unrelated things? When should I use `which`? <ngz>SeerLite[m]: IMO, search-input-file is superior to which as it will raise an exception if nothing is found. <phodina[m]><pinoaffe> "phodina: `(kernel (if (string..." <- This is what I currently do and civodul also suggested. However, my question is if I can do <phodina[m]> (operating-system (if (string=? feature "gnome" (... gnome stuff ...) (... xfce stuff ...))) <phodina[m]>I changed the var to feature as I want to configure multiple things e.g. kernel, arguments, modules. <phodina[m]>Is it possible just with single if statement or do I hav to use if for every option? <SeerLite[m]>ngz: Thanks! So apart from that they're the same? If I'm updating a package definition to the new style, would it be a good idea to replace all which calls with search-input-file? <phodina[m]> `(kernel (if (string=? lts "lts") libre-linux alt-libre-linux)) <phodina[m]> (kernel-arguments (if (string=? lts "lts") libre-linux alt-libre-linux))` <ngz>SeerLite[m]: They're not the same, as `which' searches PATH whereas search-input-file picks inputs (or native-inputs). I suggest replacing `which' calls. <phodina[m]>SeerLite[m]: Thanks so it's not possible to have conditional blocks only either inheritance or conditionals for each option - but not multiple options in one statement <phodina[m]>* Thanks so it's not possible to have conditional blocks only either inheritance or conditionals for each option - but not multiple options in one statement? <SeerLite[m]>phodina: Nope, it's not possible because the record syntax is quite strict in that it has to be `(<record> (<field> <value>))` <nckx>Ha, I *literally* just replied to it. <nckx>(Nothing pertinent, sorry.) <AIM[m]>I made a different startx script that calls xinit <AIM[m]>But the thing is that I can't type in xterm after I do startx.sh <AIM[m]>I also replaced desktop-services with base-services... Just to try debloating system... <efraim>singpolyma: not sure if someone mentioned it before, you're missing the comma, it should be ("ruby" ,ruby-3.0) <attila_lendvai>with some much appreciated help, i have managed to work out how to develop and test shepherd. i'm willing to write it up somewhere, but where? guix manual? or the cookbook? or the shepherd manual? (probably not the latter because the recepie is rather specific to guix) <ngz>attila_lendvai: I would ask this question on the devel mailing list, including your documentation. <attila_lendvai>if i write it up beforehand, then i'll just add it to the guix manual then, and then relocate it depending on feedback <elais[m]>I think it should go in the shepherd manual too <elais[m]>I’m not sure if this is still the case but I recall there being a disconnect between shepherd services as described in the guix manual and shepherd services as described in the shepherd manual <attila_lendvai>elais[m], i think it should live in contributing.texi, and the shepherd manual should point to it. ***dgcampea-2 is now known as dgcampea
<attila_lendvai>what i'll write up is how to have a reasonably short edit-compile-test cycle in a guix system vm <elais[m]>As long as it’s visible in some way you have my support <acrow>Hi nckx. Rain here but much better than some other places I am sure. <nckx>Assuming it's not a torrential downpour with deadly floods, er, yes, that's a pretty okay situation right now all things considered. :) Hope you're safe. <nckx>AIM[m]: What exactly happened to which site? <AIM[m]>Idk just guix failed to load back then... gnu website worked... <nckx>404 on the main page (/)? That's extremely strange. <nckx>Maybe there's a race condition when the Web site's rebuilt (hourly). Still, strange. <nckx>AIM[m]: I'm not finding anything in the logs… Which page was it, exactly? <civodul>nckx: the update is atomic (it's a symlink rename(2) in /srv) <nckx>I'd expect no less, civodul, but thanks for the confirmation. <AIM[m]>gnu.org worked fine tho.... at that moment <nckx>gnu.org is $somewhere in the US, guix.gnu.org is a box in Berlin. <nckx>There is absolutely nothing in the logs matching GET / returning 404. I guess we'll never know (and I've never heard of this happening before, ever, so it's not the end of the world). <nckx>AIM[m]: Thanks for reporting it, though! It does sometimes go down, and we sometimes don't notice until somebody mentions it here. <nckx>Just never with 404 AFAIK. <AIM[m]>Too thrilled to have it discussed with the official guix team! Love ya work! <ngz>Hmm. "guix import go whatever" reports an HTTP 410 ("Gone") error. <ngz>Is it me, or can anyone reproduce it? <podiki[m]>works for me with a known package I imported before, at least <SeerLite[m]>How should one update a package to the new input style if it contains an auxiliary file as input? Should the `search-auxiliary-file` call be done inside a phase directly instead? (I see Chromium does this) <SeerLite[m]>Looking into it, I see that 4b4ea166 converted Chromium's aux file input to an ungexp in the phase instead. So I'll do the same ***lukedashjr is now known as luke-jr
<elais[m]>attila_lendvai: will do when I get a minute <karrq>Hello guys. I'm trying to create a manifest.scm for a project and I need a newer version of a package than what is currently available in the guix repository. I'm kinda new but I know about `guix build --with-latest`, I don't know how to use it in the manifest... I'm using guix on arch linux. Also, on another note, I tried building the package but failed the checks <singpolyma>karrq: you can make the package inline in your manifest.scm, either by copying and editing, or by using inherit <karrq>singpolyma: I did try that in the past for another package but ultimately failed. Wish it would be easier <karrq>singpolyma: in the past I was trying to inline python-ledgerblue, now I'm trying to use conan but I need at least 1.43 <apteryx>civodul: hey! thanks for the review regarding the rootflags & cie! I've now sent a few replies. <civodul>apteryx: you're welcome, that's a much welcome improvement! <civodul>the key here is the 'options->transformation' procedure <civodul>which is a one-to-one mapping of the CLI options <civodul>i'm sorry i missed the live part of the Guix Days <civodul>but i'm glad we had a nice event with great talks <karrq>civodul: oh sweet, `options->transformation` would definitely be useful <apteryx>civodul: would you happen to know how to disable the guix-days mail alias? <apteryx>I have an account to fencepost, but don't know what to do with it <ss2>hello, I'm seeing to observe that whenever I reconfigure, Guix will loop over building certain modules. Today it had gotten to the point that it looped over it many times over before completing it: https://paste.debian.net/1232557/ <ss2>On my old laptop it took something like an hour today, and I'm having this now on another machine. <ss2>Is there maybe a reason for this behavior? <apteryx>I'm guessing there's something on your GUILE_LOAD_PATH/GUILE_LOAD_COMPILED_PATH from an older system that is incompatible with your current Guile <apteryx>it should get sorted after you reboot; in any case it is a warning rather than an error <ss2>I usually don't always pull a new checkout before reconfiguring. The one on my old laptop is maybe a week or two old (I know, but my laptop can't take the brunt of upgrading everyday), and I have done several reboots. <ss2>okay, will reboot this other machine and see if these messages go away. <AIM[m]>Is there an equivalent for build-essential in guix? <ss2>apteryx: yes, rebooting the host does make the difference. Thanks. <apteryx>civodul: I think I'm on the right track, reading /etc/aliases now <civodul>apteryx: yes, just look for "guix-days" in /etc/aliases there <apteryx>OK, I removed the alias, should be good ***EndOfAnEra is now known as DrunkAF
***lukedashjr is now known as luke-jr
<AIM[m]>Is there a way to use my own dwm on guix without creating a new channel? <apteryx>civodul: I'm now attempting to reword the comment that previously mentioned content-addressed; so far I have "The SYSTEM-KERNEL-ARGUMENTS? should only be used in necessity, as the '--load' and '--system' values are self-referential (they refer to the system which ends up with a parameters file containing these values)." <apteryx>that's still a bit vague; what exact problem does it cause? It is still valid, so it doesn't create an unresolvable cycle... I'd like to explicitly mention what effect it has. <apteryx>ah! the system doesn't end up with such a file, the boot configuration does. That's why it works. It *would* create a cycle. <civodul>apteryx: what you propose above sounds good to me <apteryx>v2: The SYSTEM-KERNEL-ARGUMENTS? should only be used in necessity, as the '--load' and '--system' values are self-referential (they refer to the system); thus susceptible to introduce a recursive dependency. <apteryx>civodul: are recursive and cyclic different when discussing dependencies? <apteryx>e.g., dependency cycle vs recursive dependency <acrow>AIM: Are you trying to install a local package of your own design? I can only guess what dwm is. <apteryx>sneek: later tell lfam I've removed the guix-days alias, you shouldn't receive mails anymore from it. <civodul>apteryx: "cyclic dependencies" sounds more natural to me, but it isn't a big deal <phodina[m]>Is there a way how to refer to packages and services in the operating-system definition when I inherit from another OS definition? The goal is to add/modify packages <apteryx>phodina[m]: (operating-system-services os) ? <apteryx>all the documented fields can be accessed, see <operating-system> in (gnu system). <apteryx>civodul: sent new series; sorry I forgot the -v3 option