<janas>apteryx: thanks for the links, I was away from the computer for a while so I didn't see them earlier! <tsarfox>Would anyone be willing to tell me what the '%' prefix to some variable names means? <tsarfox>I've assumed it's just a stylistic thing for global constants but wanted to make sure <samplet>Sometimes it means something like “this is a low-level interface” or “this is a system-level implementation detail”. <noobly>what image do I want for a virtualbox (not qemu) install? <brendyn>I just noticed my grub says it will auto boot in 4 seconds but then the countdown never starts until i press a button <nckx>noobly: There's no ‘pre-installed’ image for VB, you'll have to install Guix onto the virtual drive using the installer. <nckx>At least that's what people who use it tell me… <noobly>nckx: seems accurate, I couldn't find anyone hosting a VB image elsewhere either <nckx>janas: emacs-idris-mode was already on my ‘stuff modified in-place’ list; should be fixed on master now. The files are identical, only the tar time stamps changed. <nckx>janas: If there's a better place than Melpa from which to fetch it, please go ahead and submit that patch. *nckx → 😴, it's getting light here. <apteryx>rekado_: I've sent the improved pypi importer patches <noobly>I'm surprised exwm isn't a desktop env option on the GUI installer :-) <buenouanq>so I can't just plug a printer in, I have to add a cups service and reconfigure? ***MinceR_ is now known as MinceR
<montxero>hi is it possible to create a list of files I want to install in emacs-guix and have them installed? More to the point is it possible to manage my installed programs this way? Note I am using Guix on a foriegn distro <str1ngs>montxero: yes you can use manifest files <str1ngs>montxero: M-x: guix c p =m . select your manifest <str1ngs>also guilx packaget -m ~/path/to/manifiest <rekado_>apteryx: great, I’ll take a look today <civodul>rekado_: can you sort things such that failing builds come last? ;-) <civodul>filtering by system would also be nice <civodul>maybe qemu-binfmt missing or without (guix-support? #t), as we discussed last week? <rekado_>hmm, I’ll run the install-berlin.scm script again to reconfigure the machines. Odd. <rekado_>civodul: yes, I still wanted to add a way to filter by system <civodul>rekado_: indeed, if i do "ps aux| grep guix-daemon" on one of the first machines, i see a short command line, whereas it should contain lots of --chroot-directory flags <civodul>for now i've commented out the x86_64->qemu-armhf machines <wingo>it decided it needed to build `file` and has been running sed for the last 11 minutes in the `install` phase <wingo>though it's odd that it spends 90% of time in system :/ <civodul>there are substitutes for /gnu/store/cc2x806srmavyx2cngjpjr8s2igdwrzk-file-5.33, but maybe it's a variant of that that's needed? <wingo>maybe there are substitutes now but i'm not pulling, it might make me build gcc again :) <civodul>i built the binary tarball for ARM yesterday and fortunately i didn't have to build anything but Guix <wingo>i finished the port of armhf for lightening btw <wingo>it's been nice using guix environment -s <wingo>only weird thing is that you need to gdb through some kind of socket <wingo>as qemu doesn't emulate ptrace <civodul>ah true, so it's not too painful to use? <wingo>well you need to run the binaries like "qemu-arm -g 1234 path/to/binary" <wingo>then you need to run gdb inside "guix environment -s armhf-linux --ad-hoc gdb --pure" <wingo>because otherwise it's not built to know about arm <wingo>and then in the gdb you run "target remote :1234" <wingo>so it's a bit of a run-around but not so bad <rekado_>civodul: sending this to our admins now <roptat>if I give it the sha1 of a source file, can software heritage tell me what origin(s) have the file? <roptat>because right now a dependency of kotlin I'm trying to build is a zip file downloaded as a raw file from the right commit in a github repository... <roptat>and the zip file has no reference to a version, and it seems to have been modified compared to any upstream version I could find (it looks like a compiled source archive, instead of the actual sources) <g_bor[m]>guix.info still has 1.0.0, do you see the same there? I believe we should update that. <g_bor[m]>ok, thanks for that. I am now downloading the new version to give it a test ride :) <rekado_>espectalll[m]: what’s wrong with it? Is there a bug report about it? <espectalll[m]>rekado_: As far as I know, no bug report, I planned to months ago but only told over here <espectalll[m]>basically, I know that at the very least, my Mac's EFI freezes when it finds multiple partitions with the same UUID <espectalll[m]>for obvious reasons as well, UUIDs are supposed to be unique <civodul>espectalll[m]: hmmyou see three partitions with the same UUID? <espectalll[m]>yeah, but specifically with GNOME Disks, and editing them does make the image boot up, even if not work <civodul>yeah if you fiddle with the image, things will probably go wrong <civodul>anyway, too bad you didn't report the bug <g_bor[m]>I am quite sure that I am doing something wrong, but can't find out what. I was trying to do the following: qemu-system-x86_64 -cdrom guix-system-install-1.0.1.x86_64-linux.iso <g_bor[m]>espectalll: :) the error message was not very helpful... <g_bor[m]>Kernel panic, unable to mount root filesystem <espectalll[m]>expect lots of those in Linux, they weren't made for the average final user <g_bor[m]>espectalll: :) I was trying to simplify the command line libvirt creates for me... Simplified it too much :) <degrees380>hello, it seems ci.guix.gnu.org is down, I'm trying to ping it from 3 different nets and get no answer; I was installing Guix and got stuck in the middle of the installation (guix sysyem...) with guix trying to download <jonsger>degrees380: try "curl ci.guix.gnu.org". It doesn't response to ping, but it's still up :) ***apteryx_ is now known as apteryx
<Marlin[m]>Can i install emacs through guix without getting dependencies? <rekado_>what do you mean by “without getting dependencies”? <jlicht>Marlin[m]: you could install the `emacs-no-x' package to have a console only version of Emacs, if that is what you want <Marlin[m]>I'm on debia with the binary install of guic <Marlin[m]>i want to install emacs with guix because it's newer <Marlin[m]>But it tries to pull stuff like xorg and drivers, which will not even be used, because debian is the one handling that <amz3>I think there is package of emacs without x <amz3>Marlin[m]: maybe it is required for compiling? <Marlin[m]>I want to use debian's X without bringing everything else <Marlin[m]><amz3 "Marlin: maybe it is required for"> i guess <rekado_>Marlin[m]: Guix packages will not reuse any libraries on the host system. That’s by design. <rekado_>Marlin[m]: Guix describes how to build from source and we do so on our build farm. If you trust it you can substitute a local build with a binary from the build farm. <amz3>Marlin[m]: yes it has binary substitutes <amz3>Marlin[m]: the package definition compile from source, but you can allow binary substitutes if you trust the farm <amz3>Marlin[m]: it is describe in the manual :P <amz3>Marlin[m]: authorize something <rekado_>the build farm is authorized by default <rekado_>if you don’t get a binary from there it’s likely that it hasn’t been built yet. <apteryx>civodul: in case you missed it (I replied to an email from Ricardo), I've finally gotten around to reorganize the commits of the updated pypi importer. <amz3>Marlin[m]: new packages :) <civodul>Marlin[m]: adding or updating packages is a good way to get started with code, but you might also be interested in other activities such as web site development, translation, etc. <amz3>Marlin[m]: I don't like packaging, what I do, is develop (guile programs) using guix so I have to add a few package from time to time, even if they are very specific. <roptat>but that's probably lower priority <donofrio_>they proessed by request and now on my ds8a host I type guix and it's there now how do I onmy ds8a host I type guix and it's there now how do I enforece/restrict everything I do to /app forever <donofrio_>like first think I want to install is inxi just because it's like a hello world but diffrent just require everything I do to stay within the app directory to be happy....lookinf for checks to ensure it's only in /app <_tibbe>Hi, is there a proper way to stop the running display manager / X session on GuixSD? Stopping the Xorg service in shepherd leaves me with an empty prompt without a shell. <civodul>_tibbe: "sudo herd stop xorg-server", and then switch to, say, VT1 with ctrl-alt-f1 <_tibbe>civodul: Doesn't work for me sadly... When I switch my OS configuration to gdm it hangs in an endless loop on startup. I can switch to any terminal and issue commands but when I want to restart the X server to debug things everything locks up <Luke|Skywalker>I've been trying without sucess to run exim as mail service. After many attempts, studies and frustations I came to realize it`s most likely a bug in guix regarding the MTA <Luke|Skywalker>it wont load the configuration when passed in config.scm and even when not setting a config file for exim (let the default from scheme) <Luke|Skywalker>mails are received but not directed to users, they stay locked in input folder and it seems that exim cant deliver because it can`t write (permission) anywhere <rekado_>Luke|Skywalker: please send an email to help-guix@gnu.org with your config and a description of your problem. <roptat>donofrio_, it's hard to understand what you're trying to do... did they install guix from the binary tarball? is the daemon running? can you run "guix build hello" for instance? <Luke|Skywalker>my time is up, if no solution present today I will get back to slack <rekado_>It’s unlikely you’ll get a solution for your problem here. <roptat>I use opensmtpd for my emails, not exim... so it's hard to help you :/ <rekado_>I suggest getting your email back in order first and if you’re still interested to send a description of your problem to help-guix. <Luke|Skywalker>but does opensmtpd reveice emails too? if that the case I may try it <roptat>by default it only listens to localhost, but with some simple config, you can have a decent mail server <roptat>in case you need something to start with :) <roptat>but if we have a bug in our exim service, we should fix it... <Luke|Skywalker>thanks roptat, will be handy... And yes, everything I did this week past made me belive there are at least 2 bugs with exim <roptat>so, if you manage to get your emails in order, I strongly suggest that you send a bug report, with your config and error messages <roptat>how do you tell it what config file to read? <roptat>I really don't know how exim works, but the default config file seems a bit small: it only defines a user and a group, then includes /etc/exim.conf <roptat>the service also creates the exim users (the exim daemon is supposed to run as this user) and creates /var/spool/exim owned by exim <Luke|Skywalker>and mails are held stuck in /var/spool/exim/input and wont deliver from there <roptat>I don't know how exim is supposed to handle delivery... how does it normally gets enough permission for that? <roptat>should it run as root? sounds dangerous <roptat>although opensmtpd might do just that too, I don't know... <Luke|Skywalker>I dont think so, actually never used it too, but seemed to be the only MTA on guix <Luke|Skywalker>the default config file for exim in guix also indicates the deliver to be done in /var/mail/$user <roptat>is that because of permissions, or something else? <Luke|Skywalker>cant say, ive tryed setting permissions manually, didnt seem to affect. <donofrio_>roptat, "guix build: error: failed to connect to `/var/guix/daemon-socket/socket': No such file or directory" <roptat>donofrio_, so the daemon is not working? <roptat>I don't know, you're on a foreign distro, so maybe systemd? <jonsger>"sudo systemctl status guix-daemon.service" if you are on systemd based distro <rekado_>you can also check with “pgrep -fa guix-daemon” <donofrio_>ok they started it and the build hello command is running but I'm paniced, is /gnu/store just a cache location that gets cleaned up alot, if not it needs to go in /app/snu/store, and how? <rekado_>why are you compiling everything from source? <rekado_>what’s with /app? Guix doesn’t use it. <donofrio_>yah why is /gnu at 1gb, it's added gcc and etc into /gnu I need it in /app <efraim>Their use case needs it to be in /app <donofrio_>rekado_, I'm restricked to only one dirctory for my middleware use - /app <rekado_>donofrio_: then you need to build Guix from source and configure it to use /app as the store. <rekado_>donofrio_: this means that you will need to build *everything* from source as you can’t use the binaries we build for /gnu. <donofrio_>oh no, can I instead haev linux teammake a symlink that is /gnu points to /app/gnu cause this feels like ity should be the answer we've been dreaming of <donofrio_>we compile everything already but are weeks behind cert alerts <donofrio_>this was suppose to get us to hopfully alomost nights package control of /app in regards to apache http apr prce jboss tomcat etc <donofrio_>a hard symlink, or even a diffrent mount point cause we cannot take up 2gb within gnu directroy and 3gb within /app directory (or is everything using gnu stock (like we should change our /app to be /gnu?) ***jonsger1 is now known as jonsger
<donofrio_>eeek it's still going........did m4 now it;s doing gcc5.5..... <Luke|Skywalker>couldnt put opensmtpd to work too, i dont know what to do anymore <Luke|Skywalker>tryed with the config (modified) that you provided. tryed without any config (guix default) and with modified default (listen on all). none seem to receive any mails... <nckx>(service opensmtpd-service-type (opensmtpd-configuration (package opensmtpd-next) (config-file (plain-file "smtpd.conf" "include \"/etc/guix/mail/smtpd.conf\"\n")))) <nckx>has worked for me for 4 years. <nckx>(dovecot-service #:config (opaque-dovecot-configuration (string "!include /etc/guix/mail/dovecot.conf\n"))) <Luke|Skywalker>will send the configs as a last resort on help, if you can, if not I'm all set to go back to slackware <Krafter_>How could I configure autologin to desktop? I am not sure how to balance the services form. <Krafter_>I get a xorg-server provided more than once, presumably due to the thing I am trying to provide already exists in %desktop-services. <nckx>Krafter_: Congratulations on finding the oddest bug of the month! <nckx>Luke|Skywalker: That's not even a valid opensmtpd configuration file… Guix is not your problem; Slackware won't solve that. <str1ngs>Krafter_: you will need to modifiy the service %desktop-services <str1ngs>Krafter_: at the tail of your service modify %desktop-services. using modify-services <str1ngs>Krafter_: I have not tested that. but it should give you a general idea <Krafter_>str1ngs: Cool, thanks. I will try that. Where did you read that that function existed? <donofrio_>so I now have 2.0gb in /gnu directory and not in /app (should there have been a --prefix that I should have run before guix build hello? <donofrio_>is this project like homebrew (I used that 15 years ago on osx now it's for linux) <str1ngs>donofrio_: packages in guix are stored in /gnu . <str1ngs>donofrio_: the prefix for a package is autogenerated to a location in the store aka /gnu/store/<hash>..hello <str1ngs>donofrio_: when invoking guix build the finally output will display the prefix for the package in the store. <str1ngs>donofrio_: once the package is built, you can install it to your profile with guix package -i hello. which would also build the package if it's not built already. so guix is more like stow in this regards then homebrew <nckx>donofrio_: I haven't been following your (mis)adventures, but: to change the store directory you'll need to (at least!) ./configure Guix with ‘--with-store-dir=/app’. This is a pretty big decision. <str1ngs>yes, that would require bootstrapping all of guix ***jonsger1 is now known as jonsger
<thomassgn>Hi, getting build errors from ghc-8.0.2 and I can't figure out what is causing them. The build stops at "make[1]: *** [compiler/ghc.mk:585: compiler/stage2/doc/html/ghc/ghc.haddock] Killed" and I see there are a few messages about "changed semantics in GCC 4.4" before the build stops. Here's the log (it's a little large for a log file at 8.3M): <jje>has something changed recently, why would guix-1.0.1 tell me that "(file-systems (cons (file-system" is an invalid field specifier? <str1ngs>jje: I think you want file-system-label now <str1ngs>jje: sorry I mess read that. that actually looks right <str1ngs>but my guess is device need file-system-lable <donofrio_>str1ngs, so I need to compile guix basic with /app to have --with-store-dir=/app that way all these hundreds of builds after that would all use /app instead of /gnu? <str1ngs>donofrio_: yes, but this is not worth doing. you need to completely rebuild ever package. and bootstrap everthing <donofrio_>so I should just move from /app to /gnu and consider that my new home for all of our jail'esk like work? <str1ngs>donofrio_: yes, but /gnu is just a store. you main what to use a profile or environment or container <thomassgn>str1ngs: it could be, I had a message from icecat about a stopped script now. But it should be possible to build ghc@8 on 3.8Gi mem and waaaait a minute, my swapfile isn't active...? let's try with it on... >P <str1ngs>thomassgn: yes, you need a swap other wise it will OOM if it goes even slighly over physical RAM <thomassgn>indeed, no idea why my swap was inactive. rechecking my system config now... <str1ngs>thomassgn:of ghc 3.8G might be too low. that thing is massive right? <str1ngs>well try with a swap atleast. that way it has a buffer and won't OOM <str1ngs>I find 8G minimum is suited for large builds. and I'm not sure if there is a way to tune this <thomassgn>but I have 9G swap on SSD. Hope it's enough, but maybe I'll have to increase it. I'll see in a little while :) <thomassgn>thanks for asking so that I actually checked :) <str1ngs>still, if it goes to far into swap. it's just going to slow the system to a halt. the swap is only useful as a small buffer <rekado_>civodul: I haven’t yet been successful in using install-berlin.scm to update all nodes. I’m lacking a mechanism to spawn a remote inferior based on channels. <rekado_>it’s going to take me a little while longer. <civodul>rekado_: that's ok, there's no rush on that front <civodul>speaking of which, were you able to get your OverDrive up and running, nckx? :-) <civodul>rekado_: BTW, could you update the web pages at guix.gnu.org? <pkill9>does nayone have a script for uploading to paste.debian.net? <bavier>pkill9: there's the 'wgetpaste' package? <pkill9>wgetpaste doesn't do paste.debian.net <pkill9>I'm looking for something to run from the commandline *vagrantc was running guix on an overdrive but the power supply seems to have retired itself <str1ngs>pkill9: pastebinit supports paste.debian.net <str1ngs>though probably easier to just curl it <pkill9>str1ngs: i would curl it but i don't know how, that's why i was asking if someone has a script <rekado_>civodul: I was looking at your documentation generation build script, and well, you mentioned that you couldn’t get the modular texlive to do the right thing… <rekado_>… so instead of using it to update the website right away I plunged into debugging this… <rekado_>I took tomorrow off, so I guess I’ll have some time to first update the website and then take my time in figuring out what’s wrong with makeinfo + texlive. <rekado_>(it’s difficult to debug; I had the same problems with the matplotlib or numpy documentation) <civodul>yeah it's terrible, just a single vague error message <civodul>i encountered it before, but i couldn't remember how i got around it <civodul>mit-scheme uses 'texlive-union' just fine for its Texinfo manual