IRC channel logs
2023-03-09.log
back to list of logs
<Guest40>How to I create globally available binary? I tried (extra-special-file), and that does put the symlink as (in my case) /bin/sendmail, however, I now see that /bin is not in the $PATH. How can I create symlink in some directory that is on the $PATH by default? <Guest593>if you install packages system wide they are available for everyone <lechner>mirai / Hi, i just responded. Aside from any final items, please just let me know when it's okay to amend #41180, i.e. when you sent the email about required fields <Guest40>Right, so the right way to do this is to create a (very simple) package that would basically provide a wrapper script? <lechner>no, the other user just has to request the same file, either via guix package -i or via guix shell <Guest593>isnt your package sendmail? if you install that in config.scm everyone has it (whole system) <Guest40>The is no package (yet). I've installed msmtp, created /etc/msmtprc and got thinking "now how do I get working sendmail command" <lechner>most folks here will recommend opensmtpd <Guest593>opensmtpd on the other hand could be used natively on guix as of a service <lechner>i replaced all my instances of msmtpd in Debian with opensmtpd when i switched to Guix <Guest593>but as I mentioned, if for example you install emacs in config.scm everyone has the cmd emacs available, if you use guix install emacs only that one user has emacs <Guest593>there is no such things as /bin or local/bin in Guix (we do not use FHS here) <Guest40>The is /bin/sh and /usr/bin/env, so I thought (originally) that I would stash sendmail symlink in there as well <Guest593>ah, yea you need to stop thinking of the traditional way in distros <Guest40>lechner: out of curiosity, why did you switch from msmtp? It looks lighter, simplier to configure and does not seem to require a running daemon. <Guest40>Did you need any features of opensmtpd or it was just easier to use under guix compared to msmtp? <Guest74>what was the cmd to lett someone later with seek? <Guest74>sneek: later tell awb99 I have those errors as well, must be something else though the cmd "pacmd list-cards" prints out my actual cards <Guest40>sneek: later tell Guest74 I'm curious what happens <sneek>Guest40, you have 1 message! <sneek>Guest40, Guest40 says: I'm curious what happens <Guest74>as you can see, you said 74 but he wrote to you (40) <Guest74>yea, happened to me as well, it is useful but using it with someone that is named guest is basically useless. just letting you know <Guest74>also the user needs to write something first or he won't tell <Guest40>Ah, that explains why it answered right after "So on next join I guess" <Guest40>Yeah bit stupid. Is the source for that thing available? <AwesomeAdam54321>There's a Python project called build, so it's package name conflicts with the procedure python-build in Guix <wdkrnls>Dear Guix, what am I missing about channel configuration? <wdkrnls>I have a working guix profile for my user account where the channels were read from ~/.config/guix/channels.scm <wdkrnls>Similarly, I have a /root/.config/guix/channels.scm where I hope the same thing occurs. <wdkrnls>Unfortunately, my decade old laptop is dying so I'm trying to replace it. <wdkrnls>But on the new system it doesn't see the libraries from the new channels.scm when its in that location. <wdkrnls>How do I figure out what is happening? <wdkrnls>guix system reconfigure doesn't seem to have an option for specifying channels. <wdkrnls>I will try pulling first with -C option to create a profile for the root user with the new channel. <vagrantc>i think channels are only relevent to guix pull <vagrantc>(or guix time-machine ... which is basically just a special one-off guix pull as i understand it) <vagrantc>so you would have to guix pull with your channels, and guix system reconfigure with the version that was generated from guix pull ... <vagrantc>although why have a separate guix pull generation for root? <vagrantc>that tends to get people confused ... but maybe there is a use-case for it? <wdkrnls>My use case is updating my system configuration. <vagrantc>normally, "sudo guix ..." using the guix from the user's profile works fine. <wdkrnls>You are probably right, but I find it confusing. <wdkrnls>I feel like the root user should be in charge of breaking the computer, not the logged in user. <vagrantc>long as you can keep track of the extra complexity and not get confused :) <wdkrnls>I'm completely confused. My hope that using Guix would teach me to not be confused :) <wdkrnls>I just don't know how to search for things yet. <wdkrnls>I ran guix pull with the new channel. Now that is listed under `guix system describe`. <wdkrnls>However, when I call guix system reconfigure /etc/config.scm it still fails to find the channel. <wdkrnls>more precisely, it fails to find the modules provided by the channel. <wdkrnls>I don't see it as extra complexity though. I see it as keeping unlike things apart. <wdkrnls>But either way the error is the same. <lechner>Hi, is there a way to use my regular channel configuration with a Git checkout of Guix? <apteryx>./pre-inst-env guix build x -L your-channel-dir <lechner>apteryx / thanks! i hope to use it with deploy, but do i have to maintain a Git checkout of the channel? <unmatched-paren>In procedure struct-vtable: Wrong type argument in position 1 (expecting struct): #f <ellysone[m]>anyone knows how to deal with a package that depens on its build output to mutate it's own test data file? <ellysone[m]>you're supposed to go build -> use the output binary to rewrite the data_test.go but at this point the timestamps have already been set to 1970 <civodul>ellysone[m]: hi! maybe you can add a post-build phase that makes the relevant file writable? <attila_lendvai>so, the locale of shepherd running as pid 1 is "ANSI_X3.4-1968". shouldn't that default to UTF-8? <PurpleSym>Hm, I’m seeing a hash mismatch for openjdk-9.181.tar.bz2. Just me? <civodul>attila_lendvai: hey, did you hit encoding issues in PID 1? <attila_lendvai>civodul, yep. the logging of my service code had trouble, and (port-encoding ...) returns ANSI_X3.4-1968. i'm just testing the code that sets it to UTF-8 explicitly when opening the log file. <civodul>attila_lendvai: service logging is explicitly utf-8 and explicitly safe against encoding failures though <civodul>ACTION runs: guix build /gnu/store/gpik63l8lrasiz0mmc3yyy0rphy56w01-openjdk-10.46.tar.bz2.drv --check <civodul>PurpleSym: apparently it's fetching an hg-generated tarball, so no wonder :-/ <PurpleSym>Yeah, that’s weird, because we do have a hg-download. <attila_lendvai>civodul, it's my own logging code, isolated from the default "logging" (aka redirecting stdout into a file -- was confusing for me as a newcomer) <civodul>PurpleSym: so yes, hash mismatch as well; could you report it X-Debbugs-Cc: Java team? <civodul>attila_lendvai: oh ok, you can check how service.scm deals with that <civodul>PurpleSym: that it's bz2 makes it even more problematic because Disarchive can't handle that (yet) <attila_lendvai>my second order issue is that the root user of the server doesn't have the locale binary, and i cannot find out why. <devmsv>¿it is possible to pin a package and dependecies to a specific commit on system .scm file? <rekado>devmsv: you can use inferiors for that. See the manual for details. <devmsv>florhizome[m]: what kind of problems? <civodul>should we have a hackathon to get motivated? :-) <florhizome[m]>devmsv actually it just doesnt start from the login manager. i already disabled all extensions and installed a different lm which didnt change anything <devmsv>florhizome[m]: what commit are you in? do you use Wayland? <florhizome[m]>civodul i need a libgudev upgrade (: i believe it belongs into c-u <ellysone[m]>civodul: Ok I did the changes, added a new phase to regenerate the file but do I need to explicitly reset the timestamp of the generated file? Or can I just let guix handle this? <devmsv>florhizome[m]: have you try to update to the last commit just in case? <devmsv>I may have time to test that commit later but right now everything works (I did guix pull && reconfigure yesterday) <florhizome[m]>I updated several times the last few days so I don’t have high hopes but yeah I will… <florhizome[m]>I don’t even know where to look for specific logs yet, that might be helpful <civodul>ellysone[m]: Guix resets timestamps once the build has completed <devmsv>florhizome[m]: if xorg isn't starting maybe try slim manager just to see... there is an example in guix manual. gdm logs should be on /var/log/gdm/greeter.log and Xorg on /var/lib/gdm/.local/share/xorg/Xorg.0.log <devmsv>rekado: so trying the inferior thing. it woks for manifest but doing it on my private channel I get 'store-connection-error' to /var/guix/daemon-socket/socket upon guix pull <cel7t>civodul: you mentioned learning more about package transformations in reply to my e-mail about parametric packages, where would be a good place to start? <cel7t>I did examine patches that mention transformation in the issue tracker and am in the process of committing a trivial patch and a simple package to Guix. however it does not really show my strength with Guile or my current understanding of package transformation. <civodul>cel7t: hi! one idea you could work on is a package transformation that lets you pass extra #:configure-flags from the command line <civodul>as in "guix build hello --configure-flags=hello=--without-nls" <civodul>you would need to subdivide that into smaller tasks <civodul>first task would be: define a procedure that takes a package and returns a package with addition configure flags <cel7t>Sounds good, I will begin work on it in about ten days as I have midterms the coming week <civodul>if you're reading this and would like to help, you're very welcome :-) <cel7t>Once I am done with that task, I guess I could try sending an e-mail to guix-devel asking if anyone is willing to co-mentor <cel7t>Parametrized packages would be really cool to have ( ´ ω ` ) <rekado>devmsv: no idea what that’s about. Is the daemon still running? <devmsv>rekado: seems to be something about the git ssh channel and packages I have... will figure it out (I hope) <AleQu[m]>Hey hey, I've got a version update patch waiting in line for python-graph-tool <AleQu[m]>if anyone has some time to approve it... Cheers! <jpoiret>anyone know how to make install-locale work in build systems? <jpoiret>i added the locale input from the standard-packages to the native-inputs of my build, but still a warning <AwesomeAdam54321>jpoiret: I think you need to set the LC_* environment variables in a new build phase <jpoiret>rather I think it's because gcc-toolchain is not pulled in as an input, so the search path GUIX_LOCPATH is not set properly <mirai>does CI not keep the logs of builds <lechner>Hi, thanks to mirai we now have a Cachefilesd service. It was tested by a heavy consumer of NFS drives. If any reviewers are present, I'd appreciate merging v2 in #41180. Thanks! <gabber>mirai: that confused me yesterday, too (with my dnsmasq service build failing) <gabber>nvm my comment. i was talking about qa.guix <lechner>mirai / i rebooted and don't have a service for it yet. just started it <lechner>sorry, i did not realize anyone was using it <lechner>Hi, is Mathieu Othacehe here sometimes? <AgentKilo[m]>Hi Guix! Do you people use Icecat with Vimium? It seems the recent updates broke it? <AgentKilo[m]>Vimium keeps saying "vimium is not running on this page" for every web page I visit. Is there any new related settings in Icecat I should care about? <jlicht>AgentKilo[m]: I don't, but I can confirm that it doesn't work for me either on current master <sneek>mothacehe was last seen in #guix one month and 29 days ago, saying: roptat: hey! i cannot connect to lieserl, is it offline?. <AgentKilo[m]>hmmm, meanwhile Vimium works properly on my other Firefox installs, so I <winter>AgentKilo[m]: you can try to bisect with time-machine to catch the problematic commit <AgentKilo[m]>Ah thanks! That's an option. Guess I'll try out the time-machine command for the first time. <AgentKilo[m]>I hope I don't have to compile Icecat again everytime I do a bisect? It's a potato machine.... <ellysone[m]>How long is too long for a phase? I have a phase that is supposed to regenerate some go code but it takes 70 seconds ~ <gabber>ellysone[m]: i don't think there's a general answer to that question. the `build` phase of linux-libre takes quite a while.. <ellysone[m]>also guix lint complains about some lines being too long, but it is guix style that squashes my lines into one very long line (it's a string that I need to write for go code gen to work) <gabber>you can manually break your long lines into shorter ones by appending a backslash to the proposed end of line <gabber>guile concatenates lines ending in a backslash with the following ones <florhizome[m]><florhizome[m]> "devmsv actually it just doesnt..." <- Still gnome problems <gabber>ellysone[m]: i don't think i understand.. why/where do you need to copy/paste things? the Guix linter only checks whether Guix package definitions are written in a (somewhat) uniform style <gabber>maybe you want to paste the package definition the linter doesn't like? <ellysone[m]>I need to give you the full context, you mind me sliding into your dm so we don't pollute here ? <apteryx>lechner: good question, that's what I do since the channels I work with are mine <RavenJoad>Hey Guix! My minor documentation patch has been sitting around for a while now with no feedback. Could someone look into it? Issue #60661 <unmatched-paren>In procedure struct-vtable: Wrong type argument in position 1 (expecting struct): #f <civodul>unmatched-paren: it's deprecated, don't use it :-) <unmatched-paren>showing you the old way of doing things, and why it's not as good as gexps are <civodul>unmatched-paren: alright, that's a valid excuse ;-) <civodul>maybe it's #:guile-for-build that's #f? <unmatched-paren>> When guile-for-build is omitted or is #f, the value of the %guile-for-build fluid is used instead. <unmatched-paren>$6 = #<derivation /gnu/store/s16ric8wwr46cbs0i6r7bj47n7jmnlih-irssi-symlink.drv => /gnu/store/yfgp4z01g97zh34k5mfmlcr7kcx3aphm-irssi-symlink 7f47772e6c30> <eisbaer[m]>Hi. I am trying to build a linux kernel with debug symbols. I am using this tutorial for that: <eisbaer[m]>I am trying the %default-extra-linux-options method. (the second one) I wrote a slightly altered version of the first snippet (the one that starts with define %default-extra-linux-options) into my config.scm was that correct? My second question is where to actually put the snippet that is supposed to be added to the make-linux-libre configure script? I don't have a file with that name on my computer. <civodul>eisbaer[m]: hi! make-linux-libre is a procedure defined in the (gnu packages linux) module <civodul>if you run "guix edit linux-libre" and then scroll around, you'll see it <eisbaer[m]>civodul: thanks for your response. So both snippets go into the file that opens for edit when calling "guix edit linux-libre" ? <zacchae[m]>Im trying to run an executable which runs via the interpreter /lib64/ld-linux-x86-64.so.2. Any idea what package would supply that? <lechner>Hi, should Guix embrace the use of Zram? We have a service for it and could use it for tmpfs and maybe even swap? It may help to sidestep memory issues for installations on resource-constrained systems, such as a 1 GB Linodes (which I have not had, with 2 GB swap) <lechner>zacchae[m] / yeah, two weeks or so. i'm trying to advertise because it's just my system. with enough submissions, i will eventually show the entire available file system that one would have when installing everything <lechner>it also shows you that i have a working copy of Gocryptfs, which is not yet available in Guix <lechner>in a little while, it will also show the link target. that would have helped a lot in your case <zacchae[m]>rekado: Thanks! I have made progress. I had to symlink the relavant file from glibc into /lib64. <civodul>eisbaer[m]: not really: you could use that make-linux-libre function from your OS config.scm file and use the custom kernel from there <civodul>perhaps the cookbook needs to expound on that <Guest74>I added cups service and enabled the interface. on the interface I added my printer which has a .local address. now if I print it says unable to locate printer name.local <lechner>Guest74 / did you also enable the avahi-service? <Guest74>yes is done automatically, found it, was .local resolution missing via name-servce-switch <Guest74>(name-service-switch %mdns-host-lookup-nss) fixes it <Guest74>now lp works but firefox does still not detect the printer <evilsetg[m]>I am trying to install guix on my laptop and when I reboot after the installation the initrd fails. I have a lvm on luks setup and I suspect, that cryptsetup fails because /dev/mapper is empty besides control. Does anybody know of a problem like this? I tried searching online but did not really find anything. The initrd drops me into a guile repl and when I try to run cryptsetup manually with system*/tty it asks for my password but after entry <evilsetg[m]>fails with return code 6. The cryptsetup man page does not document that return code though. <oriansj>evilsetg[m]: I have a guix install procedure and example configuration I know works with luks <jpoiret>evilsetg[m]: what is the order of your mapped-devices field? <jpoiret>it is undocumented but the lvm map should go before the cryptsetup <RavenJoad>When doing a 'guix system -L <path> reconfigure', I get a Git error I have never seen before. Git error: object not found - no match for id (d936649fb0c6dea2960363af26f97977656a07b5) <jpoiret>RavenJoad: are you using a guix pulled from a local checkout? <RavenJoad>jpoiret: I am running with the default channel guix, but a checkout of my personal channel. <jpoiret>what is the guix describe commit for the guix channel <RavenJoad>I found it. I switched back to the upstream version of my channel and I get an appropriate "not a descendant" error from guix pull. I think I rebased a branch in the channel, removing a commit hash from existence. <RavenJoad>The pull error matches with the commit ID. guix pull: error: aborting update of channel 'synnax' to commit 622e1d79facf100dcf24ccd65792a44b7a6ec068, which is not a descendant of d936649fb0c6dea2960363af26f97977656a07b5 <jpoiret>but why was there not a pull error before <RavenJoad>The guix system build and guix system reconfigure were run before I re-pulled with the corrected channels. The system build worked, but the reconfigure failed. <evilsetg[m]>jpoiret: I had put the luks entry before the lvm one. I will try it the other way around. Thank you! <ieugen[m]>hi, I just got a Dell 5580 (seond) laptop and I plan to use it to explore guix system. Anyone have experience with this? <ieugen[m]>It has Windows and I might keep it for dual boot since I don't have any other windows systems around - and it might be usefull once/twice a year when I do taxes <jpoiret>ieugen[m]: we now have opentaxsolver :) <ieugen[m]>thanks jpoiret but I'm from Romania (Europe) <Guest7477>ieugen[m] if it is a simple software you may just use a vm and do taxes on that. would make it easier with partitions and so on since you do not need to think about dual boot <ieugen[m]>yeah, in this case I will try dual boot. guix should have a good experience with dual boot IMO. <ieugen[m]>my main concern is hardware support right now - I know guix is hard on GNU / Libre hardware <ieugen[m]>anyway, I just got it and started tinkering - will take some time to prepare