<xavierm02>rekado_ : I can't find a way to add that to my bug report, but I tried replacing the texlive-union in the propagated inputs by the full texlive, and it didn't seem to help.
<xavierm02>And I also tried telling LyX which pdflatex to use, which didn't help either. After telling lyx which pdflatex to use, it gives the error support/Systemcall.cpp (276): Systemcall: '/gnu/store/qlg22dwl8i70qx68jz5p4j0ky0h941mw-texlive-20180414/bin/pdflatex "TTT.tex"' finished with exit code 1
<xavierm02>but /gnu/store/qlg22dwl8i70qx68jz5p4j0ky0h941mw-texlive-20180414/bin/pdflatex "TTT.tex" in a terminal works just fine
<xavierm02>so my next guess is that it's about some environment variables
<poet`>Hello. Is anyone able to tell me what I've done wrong? I'm on Debian 10, and no matter which guix command I enter, I get this:
<lfam>Also, you didn't share the timestamps of the files I asked about, which are the scm / go pairs from your initial picture
<poet>Okay. Lemmie explain first: I'm on a Lenovo Yoga 2. I believe it has a bad motherboard, because when the download speed gets too fast, everything freezes. I ran the shell script to install guix, and it froze... so yes, you're right, I did install some things by hand.
<lfam>Do you mean you moved the files into /gnu/store by hand?
<dafdafdafs>I want to use Guix on my non-root Android phone. Can it put the prefix in a private directory of /data like gentooprefix?
<nckx>will`: GDM itself is buggy and crashes a lot, but I haven't heard of it freezing (although I haven't been around a lot lately). Care to file a bug report at email@example.com?
<will`>nckx: Ok! do I need to attach some logfile?
<nckx>I'm afraid I wouldn't know as I don't use GDM.
<nckx>will`: If /var/log/gdm exists it might have goodies.
<will`>nckx: BTW, actually GDM doesn't get frozen at all. If I move to TTY 1 and then comeback to TTY 7, the login screen appears again (like comming back from the sleep). If I text my username and password again, the same happen again (just keep in there without take me to the desktop environment)
<str1ngs>will`: could be a Xsesson issue have you checked $HOME/ for .Xsession error logs?
<will`>nckx: Thanks! yeah! there is a log file in that path. Thank for you help! I'm goint to report the bug now
<str1ngs>nexgen: the issue is if say you use openrc then you need a way to create services from scheme. due to functional nature of guix. because shepherd is configurable using scheme. it make more sense just to use it
<efraim>For the most part the services create config files which are passed to the daemons of the services. Like I said, if you're interested you're welcome to work on it.
<rain2>shepherd has code from systemd in it: elogind
***scs is now known as Guest61143
<nexgen>hopefully after getting more popularity GUIX will get alternatives for RC system
<nexgen>or will be forked the same way as Devuan is continuously being forked from Debian
<str1ngs>though maybe a vm would be better if you want to test services and file structures now that I think about it.
***scs is now known as Guest71829
<ngz>When I want to recompile Guix, from an appropriate environment, I get "In procedure abi-check: #<record-type <operating-system>>: record ABI mismatch; recompilation needed".
<ngz>How could I fix that? I tried recompiling, but the error is the same.
<reepca>'make clean-go' would probably do it, but it's a longish recompile
<reepca>also, how exactly does libstdc++ in (gnu packages commencement) have *any* inputs? It has #:implicit-inputs? #f, and empty inputs, native-inputs, and propagated-inputs. Are there implicit-implicit inputs?
<ngz>reepca: Trying a recompile after a make clean-go. Thank you for the tip.
<reepca>zaab: the module is (guix packages), with an 's'.
<reepca>%standard-phases has different bindings for each build system, and is generally 'build-side' code. So for example, you could (use-modules (guix build gnu-build-system)) and then evaluate %standard-phases and it would print out the list of name-procedure pairs.
<reepca>Or if you're just giving it a one-off glance, you could also just evaluate (@ (guix build gnu-build-system) %standard-phases) in the repl.
<reepca>It might help to pretty-print it like so: ,pp (@ (guix build gnu-build-system) %standard-phases)
<reepca>also, there's another (understandable) reason your snippet is wrong: #:phases is most likely not bound to a procedure, but as it currently is it will be evaluated as a procedure. You can quote it to prevent evaluation, like so: '(#:phases (modify-phases ...))
<reepca>gah, I'm an idiot, didn't notice the inputs field of libstdc++
<__Myst__>does guix support the propietary nvidia drivers?
<NewUser-57387>Hi, I installed Guix System a few hours ago. I would like to declare my to-be-installed packages in a file. I am thinking the globally-visible ones. Should I change /etc/config.scm and run `guix system reconfigure /etc/config.scm` from the root user? Or can/should it be done from my normal user? Thanks
<ngz>Globally visible packages should belong to /etc/config.scm.
<ngz>__Myst__: Guix relies on Linux Libre kernel, so, out of the box, there is no support for such hardware. However, it is possible to use any kernel you want.
<__Myst__>ngz: what do you mean by "no support" here?
<__Myst__>i'm on a dual-GPU laptop, as in I have an iGPU and an Nvidia 930MX
<ngz>I mean that out of the box, no proprietary driver, or firmware, is supported.
<reepca>__Myst__: Check h-node.org to be sure about your hardware, but most recent AMD and NVIDIA cards require proprietary firmware for pretty much any acceleration (and to make it worse, the firmware is signed, so it's impossible to develop free replacements). It's possible rending unaccelerated to a single video output might work, though.
<__Myst__>reepca: I'm fine with NVIDIA propietary drivers for the simple reason that no good opensource exists or AFAIUI can exist
<NewUser-57387>ngz: Thanks. Is there a standard location of a manifest file for normal users, for user with `guix package --manifest`, or should I just choose my own "random" place for it?
<HolyPanda>Hello, can someone please explain to me the following snippet from the manual (section 8.1): "One should never have to touch files in /etc or to run commands that modify the system state such as useradd or grub-install. In fact, you must avoid that since that would not only void your warranty but also prevent you from rolling back to previous versi
<HolyPanda>ons of your system, should you ever need to."
<HolyPanda>What is this warranty mentioned? What do I have a warranty on? I thought all GNU software has no warranty, _to the extent permitted by law_.
<reepca>HolyPanda: not a literal warranty, but a more general use of the phrase to mean "it's not meant to be used that way and stuff may break".
<bandali>HolyPanda, afaiu that’s just a play on words :) or a common expression, if you will
<reepca>although of course I'm not a lawyer and one might be able to twist that wording
<bandali>the point is that you should configure guix through a configuration file (usually /etc/config.scm but can be placed anywhere)
<HolyPanda>Okay, yes, I understand that. I was just confused by the use of the word warranty. Thanks for clearing that up.
<reepca>I've tracked down the libstdc++ mis-build to gcc-boot0-wrapped providing different answers to 'g++ -print-multi-os-directory' depending on whether it was built in test-env or not. Anyone have any idea why that would change?
<ngz>During asymptote package compilation, I get this failure: http://paste.debian.net/1095624/ It looks like a missing texlive package or some such. Anyone have an idea about what could be wrong?
<NewUser-57387>Is there any reason to not use globally visible packages for everything, if I only have a root and one normal user?
<reepca>NewUser-57387: changing globally-visible packages (adding or removing) requires root access and a full reconfigure. If you're keen on updating your system extremely frequently, I guess it's not a huge problem.
<reepca>ngz: I dunno much about texlive, but it mentions asymptote.log and CAD.log. Did you use --keep-failed?
<rekado>ngz: the texlive-epsf package should provide the missing file.
<vagrantc>NewUser-57387: occasionally something will fail to build, and having a minimal system configuration increases the liklihood of always being able to at least upgrade the system or the user's profile
<rekado>ngz: and ltxguide.cls is provided by texlive-latex-base.
<reepca>NewUser-57387: it may or may not, depending on how much has changed since you last reconfigured (for example, if you've ran 'guix pull' and the kernel was updated, that will be updated as well). But it will always at least build a new system profile if anything has changed.
<NewUser-57387>reepca: Okay, makes sense! So should I both run 'guix pull' on my normal user, and 'guix pull' on root (less often)?
<ngz>rekado: I added texlive-latex-base to native inputs, but ltxguide.cls is still missing =/
<vagrantc>NewUser-57387: sudo -E guix system reconfigure ... should use the user's guix, so you only have to pull once
<sebboh>hello guix. How should I install quicklisp? The upstream project's installation method is to download and execute (load) quicklisp.lisp. The result is the creation of a directory in $HOME and optionally modification of your common lisp implementation's rc file. (so, .sbclrc or whatever.)
<rekado>zaab: yes, you can. Look for special-files-service-type in the manual.
<zaab>rekado: wheres a good place for the symlink to be in guix?
<zaab>I am failing to make a symlink in my system definition. In the services I try to do (extra-special-file "/bin/vim" (program-file "nvim")). Problem is I do not know how to locate nvim from the store
<zaab>this is what I could find in the info page, but I do not understand any of the lingo yet :-)
<mbakke>zaab: you need something along the lines of (file-append neovim "/bin/nvim")
<ngz>While building a new version of Asymptote, I have this TeX error: http://paste.debian.net/1095662/ about missing "ltxguide.cls", even though texlive-latex-base belongs to the texlive-union in the package definition. Any idea?
<ngz>"ltxguide.cls" doesn't seem to belong to texlive-latex-base
<ngz>hhmmm. Actually, it belongs to some texlive-latex-base incarnations in the store, but not all of them. Odd.
<sebboh>Hello! My system takes a really long time to `guix package -i $anything`. It seems that it is spending some large static amount of time "building XDG MIME database...". I kinda know what that is and I am sure that I don't need to automatically update it after installing any package. (This machine does not *have* X windows. Even if some app I installed is xdg aware, I'd rather manually update that DB on
<sebboh>demand, the first time I discover that I need to reference it.)
<ngz>rekado: So, in texlive-latex-base, I grab the list of files in /tex/latex/base that are not generated, bind it to, e.g., files-to-copy, then, in install phase, do (for-each (cut install-file <> target) files-to-copy) ?
<pen14641>When using 'guix system vm' should i avoid running the returned script with sudo? (If I don't use it I get an error about KVM, which I have no idea if I can solve without just using sudo)
<sebboh>Also, frankly, MIME types are kinda not my style. I generally invoke `$program $document`, not `xdg-open $document` or whatever magic. :)
<sebboh>anyway if I can remove or disable any XDG-related 'hooks' in my guix, I think I'd be better off. But, where are these 'hooks' defined?
<sebboh>"try installing openjdk:jdk" Neat. How do I get a list of "outputs" for a given package, or across all packages?
<bavier>sebboh: output are listed in 'guix package --show=...' and 'guix package -A ...'
<rekado>ngz: simpler. You can copy-recursively the source directory.