<Digit>just saw the release announcement on lwn. :) v happy for ya. if i hadnt just done a gentoo install, i'd likely have jumped on this as a cue to install guixsd on my main rig here. oh well... spose i should update guixsd on my laptop. :) <j-r>Hello Guix! I'm preparing a package for an emacs restclient, should the module be placed at the end of ../gnu/packages/emacs.scm? It seems these aren't in alphabetical order. <paroneayea>j-r: there's not really a specific guideline iiuc <paroneayea>j-r: my general approach is: if there are other things it's similar to, put it near them <M-geemili>gpg --keyserver pgp.mit.edu --recv-keys 3CE464558A84FDC69DB40CFB090B11993D9AEBB5 <M-geemili>Meant to say: Hello, I'm trying to build a disk-image for the raspberry pi. It fails saying coreutils is `x86_64-linux'. <M-geemili>Has anyone else created a guix image for the raspberry pi? <catonano>so what happens is that all the .go files iin my current guix (.conig/guix/latest) are unloadable because they were produced with a previous version of Guile <catonano>should I boot from a USB pen with the last GuixSD on it and re-init thhe system rom scratch ? <catonano>or can I do anything rom within the current installation ? <janneke>catonano: can you access/run something like /var/guix/profiles/system-22-link/profile/bin/guix <catonano>I can access both /run and /var/guix/profiles/ <catonano>the current guile is 2.2.2 and the previious one was 2.0.13 or something <rekado_>catonano: all I had to do was update, reconfigure, make. <catonano>rekado_: in calling "guix environment guix" I run into the same problem <catonano>guix environment: error: guile-ssh-rexec-bug.patch: patch not found <rekado_>the patch was added with commit 5a3429bcc52a47ca9c879b9ace092eecddeed300 <rekado_>have you tried “./pre-inst-env guix environment guix”? <rekado_>if all of this fails I suggest using “guix pull” <catonano>;;; ERROR: In procedure load-thunk-from-memory: File o directory non esistente <catonano>;;; WARNING: loading compiled file /home/catonano/.config/guix/latest/gnu/packages/wine.go failed: <catonano>;;; ERROR: In procedure load-thunk-from-memory: File o directory non esistente <catonano>;;; WARNING: loading compiled file /home/catonano/.config/guix/latest/gnu/packages/xnee.go failed: <rekado_>catonano: it will not be able to load the .go files, but interpret the sources instead <rekado_>catonano: it will take a little longer but you will end up with a good environment. <rekado_>catonano: in that environment you have a new guile, so you should run “./configure --localstatedir=/var” and then “make” <rekado_>oof, I’m trying to merge master into core-updates. Lots of conflicts. <catonano>rekado_: with "in that environment" do you mean in ".config/guix/latest" ? <catonano>should I run conigure and make inside a folder that's in the store ? <catonano>rekado_: that's ok, take care of your merge. My machine is working, ater all, I'm not isolated <rekado_>catonano: do you normally use Guix from git or do you use “guix pull”? <rekado_>when using “guix pull” .config/guix/latest will point to the current guix in the store. <rekado_>if you use a git checkout you want .config/guix/latest to point at the checkout. <catonano>I am a bit frightened to move to a checkout <rekado_>okay, well then you don’t need to bother with “guix environment” I suppose. <catonano>rekado_: I don't. I mentioned it because I was also checking how it went with the local checkout <rekado_>just running “guix pull”, then reconfigure, then “guix pull” again should be sufficient. <catonano>but the last one is the one about the missing patch <catonano>guix pull: error: guile-ssh-rexec-bug.patch: patch not found <rekado_>I see that patch in dist_patch_DATA, so it should be fine :-( <catonano>I wrote about this on the help ml. We can discuss this asynchronosly, if you preer <catonano>rekado_: as a side note, if "guix pull" doesn't work, "sudo guix pull" DOES work <rekado_>catonano: if “guix pull” does not work, but “sudo guix pull” does then maybe you could just swap out the target of .config/guix/latest <rekado_>so that your account’s .config/guix/latest points to root’s .config/guix/latest <brendyn>ACTION wonders if ludivic tested the installer <rekado_>brendyn: you may want to try again after a while. This just means that the cache doesn’t yet hold that item. <catonano>rekado_: ok, I made ln -f -s -T ~/projects/guix/ ~/.config/guix/latest <catonano>should I want to go back to gix pull, could I ? <rekado_>catonano: yes. You can just delete ~/.config/guix/latest and use some other guix to run “guix pull”. <brendyn>I'm installing 0.13 now with --fallback <brendyn>the bootstrap-binaries were not in the cache so they are being built <solene>that guy on the mailing list seems to be weird <brendyn>Who has the power to delete posts from the mailing list? <brendyn>solene: My icedove email client marked one of his posts as spam, amusingly <brendyn>civodul: Can you delete Fox's thread on the mailing list? <rekado_>ACTION just merged master into core-updates <brendyn>I think many internet communities have died out in the past because the original members were too kind to censor fools. I don't want that to happen to Guix. <rekado_>I don’t think there’s a need to do anything in this case. <rekado_>there’s better things to spend time on <brendyn>There is a mathjax patch on the mailing list. Do you agree with putting it in /share/webapps/mathjax ? <rekado_>baijum: I think you could just regularly label all items under /gnu/store <brendyn>Seems arbitrary, I guess some choice has to be made, but it's different to both Arch (/usr/share/mathjax) and debian (/usr/share/javascript/mathjax) <baijum>rekado_, Thanks. let me check how to do that <rekado_>The topic of how to label store items for SELinux requires some more thought. <rekado_>baijum: but your bug report seems to be fixable by changing the label of that one file to default_t <rekado_>baijum: would you like to work on this? <baijum>rekado_, I don't have much idea about SELinux. I am just playing around with Guile & Guix now <rekado_>your problem on Fedora is simpler than what I was planning for. <rekado_>you don’t need a full policy for just that one file. It’s enough to label it with default_t <rekado_>baijum: you may find that this just gets you to another problem. Just report that here and we’ll see how to move on from there. ***Piece_Maker is now known as Acou_Bass
<rekado_>catonano: did you get your problem fixed? <catonano>the tricks related to .config/guix/latest should be discussed in the manual, to some extent, in my opinion. I will open a thread about this ***boegel is now known as boegel|afk
***Piece_Maker is now known as Acou_Bass
<M-geemili>Do I need an armhf-linux machine to build coreutils for armhf-linux? <rekado_>M-geemili: I think we have a working cross-compiler for armhf <rekado_>M-geemili: so you could probably just do “guix build --target=armhf-linux coreutils” or similar <rekado_>M-geemili: the argument to target should be a proper triplet, I think <rekado_>I don’t know the exact invocation, unfortunately <M-geemili>I was trying to build a disk-image for armhf before, but it complained about being on x86_64 <rekado_>note that we don’t have GuixSD support for armhf. <rekado_>you can use Guix as a package manager, though <M-geemili>rekado_: Can I use Guix to build an image for armhf? <M-geemili>Using something like `guix system -s armhf-linux <file>`? <rekado_>M-geemili: maybe, but it probably won’t be bootable. <civodul>rekado_, M-geemili: to cross-compile to ARMv7, you should use: guix build --target=arm-linux-gnueabihf coreutils <civodul>"armhf-linux" is not a valid GNU triplet <civodul>ACTION just posted a tentative road map for 0.13.1 and 1.0 <rekado_>it’s on the mailing list and should appear in the archives shortly <rekado_>and we have some unaddressed CVEs for PCRE <rekado_>hmm, I get a bad signature on the 4.6.4 release of samba <efraim>I saw a samba thing on Debian-security but they tagged it stable onlu <rekado_>ah, they sign the *unpacked* sources <efraim>One of those where they sign the tar and distribute tar.xz or tar.bz2? <rekado_>samba still hasn’t finished building <rekado_>I wonder why it relinks a lot of binaries during the install phase <solene>I'm not sure to understand the grub specific installation for UEFI system <solene>the doc now says "Make sure the grub-configuration form refers to the device you want to install GRUB on. You also need to specify the grub-efi package if you wish to use native UEFI boot. " <solene>as someone new to guix, I have no idea what to do... :( <solene>I may try to add grub-efi to packages list but I'm not sure that would do the trick <brendyn>guix system init appears to be displaying "substitute: updating list of substitutes from ... 100.0%" over and over forever. <brendyn>Usually this comes up many times anyway for some mysterious reason <arodlinux>I need help with network card installing GuixSD <efraim>As far as I understand the updating lists, it checks for some things, and then based on what answer it gets it asks about other substitutes <solene>I just installed guixSD successufilly on my uefi laptop with ng0 workaround, but using luks, once guixsd starts after asking the password, it fails finding "my-root" (which is the label of the root partition) <solene>I don't understand in mapped-devices what target means <rekado_>solene: target is the name of the device as in /dev/mapper/the-root-device <solene>now, from the usb system I modified my /mnt/etc/config.scm, how can I do guix system reconfigure ? it doesn't takes 2 args like init. Do I need to chroot and then do a reconfigure ? <solene>hmm, manual doesn't explain that /mnt/boot/efi should be mounted to init the system with efi (at least it fails when installing grub) <Apteryx>Why would: (assq-ref 'set-paths %standard-phases) returns #f ? (%standard-phase from gnu-build-system module). <Apteryx>If I use just assq, I get: (set-paths . #<procedure set-paths (#:key target inputs native-inputs search-paths native-search-paths)>) <Apteryx>But I just want the cdr, which I understand assq-ref should return. <brendyn>Apteryx: Actually, haven't you got the arguments backwards? <paroneayea>classic mistake; assoc and assoc-ref have arguments reversed <paroneayea>(get-val sequence key) is clearly the best order :) <catonano>paroneayea: I suuppose this is one of the things you don't like of Cloure ;-) <paroneayea>catonano: heh, I'll admit that I found some of it a bit inconsistent but <janneke>catonano: think our messages crossed :-) <brendyn>get looks much simpler, and even allows one to change the not-found value <janneke>catonano: cross is not the right term i guess <catonano>well, I'm goiing out for a ride, now. Later I will try to explain why I was enthusiastic about those data ;-) <janneke>catonano: you intend to play with and move intention to npm while i'm loosing interest <noobie_>hey, guys! I`m looking for software installed by default in GuixSD. Such as web browser, terminal emulator etc. <janneke>noobie_: i don't understand your question, what's installed by default you determine yourself when you first stay guix system init <rain1>janneke: ive been following the hex assembler, and copying it <brendyn>noobie_: If you use the example desktop setup, you'll get GNOME with Abrowser and Thunar. You can then install whatever you want <rain1>rather than the VM instruction set <noobie_>janneke, I assume theare are such a default user environment in GuixSD, like in ubuntu, isn`t it? <janneke>i have added two [ascii] output stages to mescc: output the parse tree first and also an output mescc-specific object format that can be linked later <noobie_>brendyn, thanx! Just what I looked for. <janneke>rain1_: as my next step for mescc, i wanted to create something close to OriansJ's stage2 labeled hex...but i was wondering how to do that <janneke>this cat3.hex86 may be closer/easier [for now]? <brendyn>noobie_: I'd like to create default setups for a variety of options like KDE, but currently we just have a few simple examples like that, and most people use GNOME, so that's the easiest to get working atm <rain1>yes its built up very gradually. at first we just have raw hex assembler (hex0), then labels and various ways to refer to them (hex2, I found that I needed to add some new sigils), and then we can add nmenomics/shorthand macros for small pieces of hex code (thats all hex86 is) <rain1>ideally that should be enough to create a small virtual machine, which more advanced languages can be interpreted ontop of <rain1>but ive just been reading the hex stuff at the moment, i need to start reading mes in detail next! <janneke>i'm going to look at cat3 labeled hex86 and stage2 labeled hex, i'm looking for something labeled-hex to migrate mescc's o to <rain1>maybe it'd be possible for mescc to emit relocatable elf object files, which could then be assembled and linked <rain1>right now i have just been emitting executables straight off, no linking stage <janneke>i like the idea of using ascii output for every stage... <rain1>yeah its really nice to be able to easily inspect them :D <rain1>I have been digging around the elf format for a while, just got .bss sections figured out (which should let me do programs with global vars) <janneke>also, mescc runs real slow on mes, but is pretty doable when run on Guile. when the output is ascii and the same, that really helps during development. <rain1>ah do you know why it is slow? the biggest slowdown in my scheme was macro expansion (and i have an idea how to fix it) <janneke>rain1: no, i don't know. i have done quite some profiling earlier and made quite some progress. <janneke>alist/environment lookups has always been a big bottleneck <rain1>perhaps a quick pass to resolve scope, and create closures could help <janneke>yes, something like that -- and/or add real modules <rain1> http://ix.io/uLw this is how i do that, then instead of variables simply being a symbol they are one of (var tmp|loc|env|glo <name>) <janneke>i have focussed my attention to compiling TCC, ie, closing to bootstrap loop first <rain1>building tcc will be awesome!! <janneke>oh, uLw is your scheme compiler in scheme? i'd very much like to rewrite mes.c in some kind of scheme/lisp <rain1>yeah, this is a rewrite i'm doing though <noobie_>How guix go along with other PM`s for locally installed software? I mean such soft as rvm. <rain1>i learned so much i can do it a lot better now ^^ <noobie_>janneke: it`s for ruby lang management <noobie_>when you can install different version of ruby and switch between them in different projects <janneke>noobie_: guix has an importer for gem -- that's ruby right? <noobie_>janneke: yes, ruby packages named gems. <janneke>the idea of guix is to have it manage /all/ your software <noobie_>janneke: and what about guix importer? Is it stands as wrapper or...? <janneke>noobie_: you say `guix import gem ...' to generate a package description, that you then can build and install using gux <brendyn>Yeah but it seems like it'll take much effort for it to take over things like extensions <noobie_>janneke: and to perform anything like `guix import gem ...` that gem should be in guix index ( base, or so...)? <janneke>noobie_: no and i guess that many gem packages are available already <noobie_>janneke: it is more about gem versions. Because we can find ruby gems in almost any package manager, but versions would be very very different. <janneke>noobie_: i don't know much about ruby, sorry. all i know is that guix can help you create separate environments where you can install any combinations of packages <janneke>and reproduce those environments anywhere <noobie_>Maybe I should ask about it in ruby/rvm channel :) <noobie_>bb, guys and gals. have a nice time. <randomfoobla>sneek: later tell catonano: I pushed your patches to gnunet.org, excellent work massive reduction in texinfo failures. if I forgot anything, send a mail <acthel>Quick question: why does Guix compile some packages when installing them but not others. Shouldn't subsitutes be available for all packages? <davexunit>acthel: ideally, yes, but in practice our build farm may be behind. <acthel>davexunit: forgive me if the answer is obvious but why let a client guix know that there's a new version before the build farm has finished building the package? <davexunit>acthel: because it's unrealistic to expect everything to be built. there are thousands of packages. <davexunit>downloading binaries is an optimization, not a necessity. <davexunit>so usually there's a "good enough" point reach <davexunit>but the more resources the build farm can acquire, the better the experience will be. <acthel>davexunit, bavier: thanks, I was wondering because I am used to the binary distribution style of package management <bavier`>eventually gnunet/ipfs/etc would be cool so that users could distribute substitutes amongst themselves <bavier`>e.g. often a developer has built a package locally before submitting or updating a package. it would be nice if they could distribute that to others easily <quiliro>/gnu/store/bdzxdpdw25k8v6lz54clz42bilx47srj-grub-2.02/sbin/grub-install: warning: Sector 50 is already in use by the program `FlexNet'; avoiding it. This software may cause boot or other problems in future. Please ask its authors not to store data in the boot track. <quiliro>Installation finished. No error reported. <quiliro>guix system init /mnt/etc/bare-bones.scm /mnt --fallback <davexunit>acthel: the devs try to maximize the amount of packages that have substitutes available, but sometimes things aren't as good as they could be. <davexunit>especially if you use 'guix pull' to get the latest version of the master branch. <quiliro> "netdev" is required group for users? <efraim>Who was it that did all the work on ceph and rocksdb? I don't remember if I tried before but ceph is currently building on my aarch64 board <quiliro>if the installation of grub is bad, can i fix it by booting from live guixsd usb? <rekado_>solene: I think the efi manual is about to be updated. It didn’t make it into the release by mistake. <rekado_>solene: or rather “the efi section in the guix manual” <rekado_>solene: about your reconfigure problem: I don’t really have a good way to fix this. <rekado_>solene: it should be possible to reconfigure with a chroot, but it’s a little difficult <rekado_>it would be nice if the USB disk image made recovery possible by allowing to reconfigure a system that’s under some mount point. <mbakke>efraim: I'm pretty sure ceph will fail due to some intel-specific stuff that has been fixed in 12.0.3 <mbakke>but there are other problems with 12.0.3, so haven't been able to update yet <efraim>i don't have a use case for it yet, but it is good to know it should be working on aarch64 <mbakke>nice that rocksdb tests worked on aarch64 too <mbakke>efraim: I have a dream of a ceph cluster consisting of 100s of single-disk ARM OSD nodes, and a couple intel/ppc monitor nodes <rekado_>gazon: I don’t understand your question. <rekado_>you just submit something there and get a URL back that you can then share. <gazon>rekado_: thank you. just me being utterly daft. <gazon>This http://paste.lisp.org/+7G2W is the result of a guix pull in GuixSD in a qemu vm: guix substitute: error: corrupt input while restoring '/gnu/store/wak3m4kdkgw010qn1ksnqlggvklp4b24-gmp-6.1.2/lib/libgmp.a' from #{read pipe}# <rekado_>gazon: this means that there was a problem with the substitute (either in transit or in our cache). <rekado_>gazon: you can fall back to building this from source with “--fallback” <gazon>rekado_: i tried it a copule of times yesterday and a couple today; so i think it unlikely to be corrupt in transit. <rekado_>yeah, it’s pretty common for older substitutes to be corrupt <rekado_>I patched the CRAN importer to be a bit more robust and to support more package types – and then proceeded to import as many packages from Bioconductor as possible. <rekado_>I’m now sitting on 40k lines of new code to review <gazon>rekado_: will do, thank you. <catonano_>does anyone know who randomfoobla is ? I' d like to send an email but I have no clue as to whom to <sneek>catonano_, you have 1 message. <sneek>catonano_, randomfoobla says: I pushed your patches to gnunet.org, excellent work massive reduction in texinfo failures. if I forgot anything, send a mail <catonano_>sneek: later tell randomfoobla yes, thanks. I also sent a couple of patches taht reduce the warnings in the dev file too, i don' t see them in the log <catonano_>sneek: later tell randomfoobla also, I have no email address of yours <catonano>say I want to create a brand new python based project. Is there a widely used tooling for python projects ? <paroneayea>catonano: you can use virtualenv and pip, but lately I've been trying to use guix environment and just ship a guix.scm with the package <catonano>paroneayea: exactly: I want to use Guix to set up my dev env <catonano>so in thhe guix env there has to be python itself and the libraries I need and nothing more ? <paroneayea>catonano: yeah you want the dependencies in the propagated-inputs slot <rekado_>catonano: local-file is a gexp and is defined in (guix gexp) <ng0___>catonano: sorry, picked a random username earlier today.. I thought you sent no one else patches on documentation :) <catonano>I think I sent 2 more patches that fix (mostly) the dev file too <catonano>ng0___: yes, I saw them. There should be 2 more <ng0___>or was it on the mailinglist, then I can filter <ng0___>have you consulted the original documentation books to see if something was forgotten? it's not that important, there will be at least one more person double checking it before it is merged into gnunet.git <ng0___>when I checked, I had some open tables where there were images <catonano>ng0___: no I didn't consuult anything. I just ironed out as many warnings as I could <catonano>I'd be really happy to have Gnunet properly packaged, one day <ng0___>I'll test my spectrwm package now. I hope to get back to services soon <catonano>package init file 'distutils/__init__.py' not found (or not a regular file) <catonano>error: package directory 'distutils/command' does not exist <ng0___>distutils should be setuptools, right? <ng0___>you do use the python-build-system?