IRC channel logs


back to list of logs

<davidl>i'm having hydra mirror issues again.
<davidl>and my guix init command fails with Network is unreachable despite other pings working fine.
<davidl>I was thinking thath this is perhaps a timeout issue? Since it works fine for quite a while until this happens.
<davidl>probably there's a timeout issue with hydra mirrors since a simple reboot with my usb install medium and now it all seems to download again. I guess a very lightweight initial configuration is the way to go.
<brendyyn>Holy cheeesus GuixSD 0.12 successfully installed. GNOME running
<brendyyn>but no networkmanager
<DoublePlusGood23>brendyyn: it uses wicd by default
<DoublePlusGood23>look in the bottom left
<brendyyn>oh, why is that there instead of the top right?
<DoublePlusGood23>brendyyn: Not sure. maybe a gtk2 v. gtk3 thing?
<brendyyn>Hmm well I'll see if anything changes when I do a system update now that I've at least got it installed
<brendyyn>Any idea why network manager isn't used?
<DoublePlusGood23>brendyyn: I'm not sure. It was in the repos.
<DoublePlusGood23>Maybe it didn't like shepherd? Wicd seems to work (didn't pick up my 5GHz network though) OK, so I haven't poked around much.
<brendyyn>wicd seems more of a stand-alone program. I've always used it, but it's always seemed kinda ugly
<DoublePlusGood23>What's the protocol of adding new package expression files? ex: there's a generic admin.scm, but also a mc.scm purely for midnight commander...
<DoublePlusGood23>brendyyn: yeah, it doesn't integrate nicely with GNOME either.
<b11111000000> ERROR: In procedure load-thunk-from-memory: not an ELF file
<b11111000000>;;; WARNING: loading compiled file /root/.config/guix/latest/gnu/packages/wdiff.go fail
<b11111000000>on `guix pull`
<b11111000000>what that fact?
<b11111000000>(then it update subsitutes, loading,compiling etc)
<brendyyn>b11111000000: Guix has updated to guile 2.2
<b11111000000>need I to do something?
<brendyyn>Not sure, I think it might go away after you fully upgrade
<brendyyn>Interesting. I run guix pull and it downoads some packages, then I run it again and it downloads even more
<brendyyn>Anyway I better see if I can get Calibre upgraded before 0.13 is out!
<DoublePlusGood23>brendyyn: yeah, I run it perodically - it fails - I shake my head, then run it again til it works
<brendyyn>But it didn't fail, it succeeded
<DoublePlusGood23>brendyyn: actually it's `guix package -u` that seems to be giving me problems lately
<b11111000000>brendyyn +1, I run `guix pull` from root before, and then I run it from user - both use /gnu/store, both downloads something and compile it again
<b11111000000>can it do that faster? and only once?
<brendyyn>b11111000000: Well it should only only happen once I think, unless the dependancies installed on each profile are different. You'd have to ask civodul
<reepca>b11111000000: see if you haven't already
<b11111000000>reepca: thanks
<DoublePlusGood23>reepca: are you knowledgeable on package expressions at all?
<reepca>DoublePlusGood23: Only what a quick glance in the manual can tell me
<DoublePlusGood23>reepca: hmm. trying to package a small program and I'm not sure how to proceed. I think it's a similar to a problem I had last time due to the program lacking a configure script
<DoublePlusGood23>gives me an error that it can't find ncursesw/curses.h even though I have ncurses in the inputs.
<reepca>I don't see any -I flags for the compiler that tell it where to search for header files. So it's probably searching in the "default" locations (I think /usr/include maybe?). Anyway, somehow you need to tell the makefile to tell gcc where to look for the header files. I'd guess this would involve patching the makefile.
<DoublePlusGood23>reepca: was hoping I could avoid that. Thanks for the tip
<reepca>Oh wait, there's hope after all!
<reepca>That should tell it where to look, as long as make doesn't unset environment variables or anything like that.
<haz1>Hi, it is seems that new version of a package definition overwrites previous versions. What is the rationale on not keeping all previous versions of a package by default?
<haz1>relevant on nix dev:
<DoublePlusGood23>reepca: currently looking at `rpath` flag. will also try that :)
<DoublePlusGood23>reepca: might actually be related to !
<reepca>My installed version of gcc apparently has a man page containing only "timestamp". Checked in ~/.guix-profile/share/man/man1/gcc.1.gz and it indeed just says "timestamp". Weird
<DoublePlusGood23>reepca: guix is amazing
<DoublePlusGood23>reepca: haha!! it works
<DoublePlusGood23>reepca: :O
<rekado>brendyyn: you can easily change from wicd to network-manager
<rekado>brendyyn: it’s just that the default services include wicd, because it has been in Guix for longer.
<rekado>(I’m still using wicd, because I didn’t get around to playing with network manager)
<reepca>so I'm trying to look at the contents of a package object in guile, and trying to use "display" within the inspector is causing it to say "invalid arguments to display", but "help display" just says "print the current object using `display'."
<reepca>(doesn't mention any arguments)
<rekado>when you say “inspector” do you mean the REPL?
<reepca>,inspect gcc-7
<rekado>could you show us the code that’s failing?
<rekado>I’ve never used ,inspect with Guix.
<rekado>I just use things like (package-foo gcc-7)
<reepca>in gcc.scm gcc-7 inherits from gcc-6 which inherits from gcc-5 which inherits from gcc-4.9 and so on, so I'm trying to figure out what the actual stuff in gcc-7 is directly.
<reepca>(trying to figure out why my man page is MIA)
***Apteryx_ is now known as Apteryx
<davidl>After I have installed with guix init my partition is no longer a valid LUKS-device, happened twice - why is that?
<catonano>efraim: did you read the suggestions about the Emacs/Guix setup that myglc and I wrote to you in reply to your inquiry on the help mailing list ?
<b11111000000>how to setup automounting in guix without gnome? can we to distribute pmount etc?
<rekado>b11111000000: You may need to add gvfs to the system-wide packages.
***Piece_Maker is now known as Acou_Bass
<janneke>i've inadvertently allowed postgres to be updated to 9.6.2, now it doesn't start anymore
<janneke>yeah, state is evil...
<janneke>i wonder how we want to handle this in guix? `you can always revert'--yeah right
<davidl>Anyone who has a clue what might cause guix init to invalidate a luks-device?
<tct1>Hi :-)
<tct1>How can one get all the source code for the packages in GuixSD?
<rekado>tct1: you can get the source of any package with “guix build -S name”
<brendyyn>'Why is there such inconsistency with #:use-modules ((guix licenses))? sometimes it has #:select (gpl3) etc, sometimes it has #:prefix license:
<rekado>brendyyn: history happens
<tct1>Thank you, rekado!
<brendyyn>rekado: Should we change them all to #:prefix license:?
<rekado>brendyyn: no, we change them when we touch the code there anyway
<rekado>otherwise there’s no point in changing it.
<rekado>doesn’t hurt anyone.
<brendyyn>rekado: to license:?
<rekado>brendyyn: to whatever makes sense. In small modules there’s really no need to complicate things with an extra prefix.
<rekado>we only do this because “zlib” is both a package and a license.
<brendyyn>Upstream keeps changing things faster than I can package them -.-
<rekado>The last step in the java bootstrap is really hard
<rekado>I have classpath 0.99, jamvm, ecj, a javac-like wrapper for ecj, and wrappers for other classpath tools running on top of jamvm
<rekado>but building icedtea 7 remains a challenge
<rekado>it’s hard enough to get it right with the official tools
<rekado>looks like the classpath isn’t right at compile time.
<rekado>this shouldn’t be hard to fix, but it’s hard to find out where and why
<solene>hello, I'm new to Guix, I finally got it installed on one laptop but encountered issue with 2 packages (python 2.7 and mesa), I had to use guix package --fallback -i to install them
<solene>is it a known issue ?
<rekado>solene: yes.
<rekado>“--fallback” means that Guix should build from source when the build farm cannot offer a binary.
<solene>rekado: I wasn't able to find it on the bugtracker
<solene>rekado: yes I understood the fallback option, but I wasn't able to understand why the problem was hapenning.
<rekado>it’s not a bug
<rekado>it’s just a problem with the infrastructure
<catonano>solene: the build farm is not exactly a rock. It's frail and sometimes it misses some important binaries
<rekado>we’re trying to get a new server up soon, but we’ve been having hardware problems
<solene>okay :)
<solene>i'm trying to install openssh and start sshd but I can't find how to start ssh. herd status | grep ssh doesn't show anything
<solene>I'm trying to understand how to configure services but I fail to find it in the documentation. I have a /etc/config.scm that I made for the installation, is it still necessary ? I wasn't able to find anything else looking like a configuration file in /etc
<rekado>solene: yes, it’s still necessary.
<rekado>the configuration file is how you define your system
<rekado>each time you want to change something in the way your system works you would edit the configuration file
<rekado>and then run “guix system reconfigure /etc/config.scm” as root
<rekado>(note that this would use the root user’s version of guix, so it’s advisable to run “guix pull” as root as well)
<rekado>solene: to add an ssh service you need to add “(service openssh-service-type)” to the list of services in your configuration.
<solene>rekado: I will check this, the reconfigure commands seems what I was lacking
<solene>rekado: something like (services (cons (service openssh-service-type) %desktop-services)) ?
<rekado>solene: I recommend reading the Info manual, section “Invoking ’guix system’”
<solene>rekado: thanks for you help, I will study this. The manual is quite big I didn't know where to "start". I'm the kind of people which learn by practicing and doing mistakes, I can't read the whole book to start :D
<davidl>solene: for ssh service see here: under (services ..
<davidl>so I managed to install on a btrfs subvolume that I used "btrfs set-default" on (using dos instead of GPT) however, now the boot-configuration is shit cuz my system is placed in /sv_root instead of /.
<davidl>except when I mount from the usb, then the system is directly under /mnt.
<davidl>is anyone successfully using btrfs with guix?
<efraim>catonano: yeah, I saw the emails, about to try them out now
<paroneayea>which would be a good place to put a thermald package?
<solene>davidl: indeed I found this page to enable ssh but I did not know how to use the config file, rekado then told me to use guix system reconfigure, I have ssh running now :)
<solene>when I type guix, I have a lot of warning and errors
<solene>just before, I did guix pull and guix system reconfigure config.scm, tried to reboot but still the problem, did I break everything ?
<reepca>solene: see
<reepca>it's a transitional thing caused by that particular "guix pull".
<solene>reepca: ok, so if I undertand, i just need to run "guix pill" to fix this
<reepca>solene: worth a shot, I don't remember exactly what I did but it turned out okay ¯\\_(ツ)_/¯
<castilma>after hours and hours of .. I don't know, compiling, configuring, downloading substitutes and more, guix system init /mnt tells me, there is no more space left on the device. I made a 5 gb image as recommended in the docs. how big is your installed guixsd?
<civodul>castilma: what kind of install is it? GNOME?
<janneke>guix pull: error: build failed: some substitutes for the outputs of derivation `/gnu/store/5ch78mg064h0fcn141kmwqm7cvqrijxv-gcc-4.9.4.drv' failed (usually happens due to networking issues); try `--fallback' to build derivation from source
<janneke>$ guix pull --fallback
<janneke>guix pull: error: fallback: unrecognized option
<civodul>janneke: this has been fixed recently
<civodul>to get the fix, simply run... guix pull :-)
<janneke>i'm trying to install guix 0.12.0 on an arm account i got [in paris]
<paroneayea>ok I'm adding thermald to admin.scm
<paroneayea>that seems like the right fit
<castilma>civodul: are 5 gb enough for the lightweight desktop conf?
<janneke>maybe i'll wait until 0.13
<janneke>it's a debian box
<civodul>castilma: i would think so, though i think the example includes xmonad, which means ghc and all
<civodul>so i don't know but if 'guix system init' ends up building the Haskell stuff, instead of getting substitutes, that can take quite a bit of disk space
<castilma>what ist ghc and what do I do now? reformat? or can I somehow make the install script use the already installed packages? how could I remove the unused ones?
<castilma>garbage collect wont work, right? atomical actions...
<civodul>castilma: you could remove "xmonad" from your config.scm and rerun 'guix system configure init config.scm /mnt'
<civodul>(no need to reboot or reformat)
<castilma>civodul: guix system configure: unknown action
<castilma>I think the 'configure' shouldn't be there
<castilma>but without it getting a no space left on device
<civodul>sorry, i meant "guix system init"
<civodul>janneke: so there are corrupt or missing substitutes for arm?
<civodul>janneke: what was the error in the first place?
<civodul>note that you can also run "guix build /gnu/store/whatever.drv --fallback"
<castilma>how can I tell guix gc to use another guix store than /gnu/store. (e.g. /mnt/gnu/store?
<janneke>civodul: yes, the bash-static thing i guess
<janneke>i'm doing: guix environment guix
<janneke>on a 0.12.0 on debian jessie
<janneke>now building bash-static...
<Apteryx>Any idea why bash completion isn't working in M-x shell (emacs)?
<Apteryx>For example, git TAB TAB --> nothing.
<Apteryx>Nevermind. The same situation arises on Ubuntu (not a Guix problem).
<Apteryx>I can this is because the Emacs dumb shell doesn't run a real bash process. M-x ansi-term does so it works.
<slyfox>there is 'M-x term'
<slyfox>(ah, likely the same as 'ansi-term')
<Apteryx>slyfox: I think they are different (code base) but mostly alike in functionality.
<paroneayea>I thiiink I have thermald packaged ok :)
<castilma>why is it so common to get crc errors on substitutes?
<davidl>I wonder that too. I got it on gmp package.
<ng0>I am preparing a darkhttpd-service right now.. I just have to do many services at once it seems to understand debugging services better.. I hope darkhttpd just works.