<RX14>having tried nix/nixos guix seems a lot nicer, on the commandline at least <buenouanq>they took what they learned working with nix and went beyond <lfam>I use both and I also think our command-line interface is easier to learn and use <buenouanq>ignoring everything else, choosing guile is worth it by itself <buenouanq>I'm wanting to connect a GoPro camera to a machine running GuixSD. It uses MTP though which I know nothing about. <buenouanq>What do I need to install to get this readable? <lfam>buenouanq: `guix package -s mtp` suggests gmtp. On Debian I use jmtpfs on the command-line but we don't have it packaged <lfam>The Nix expression for jmtpfs suggests that it would be easy to package <buenouanq>I don't exactly know what my question should be here... <mekeor>buenouanq: it would be helpful to know what led to the error message. <buenouanq>I did the dumb thing and installed it to root now and it works. <mekeor>you might want to install gmtp as user and then use sudo <buenouanq>you know, I'm not sure why using sudo didn't even occur to me... <buenouanq>but I don't want to have to elevate permissions for something like this <lfam>You can use it as an unprivileged user assuming your kernel supports FUSE <lfam>You unmount with `fusermount -u $mount-point`, which comes from the FUSE package <lfam>Assuming there is only one MTP device connection, you mount it with `jmtpfs $mount-point` <sneek>OriansJ``, you have 1 message. <sneek>OriansJ``, janneke says: i'll be working to remove hex3 from mescc altogether and directly produce hex2..small steps :-) <OriansJ``>janneke: I'm also generating some more examples for M1 for you. That way you'll see it works just fine for x86 and AMD64 <OriansJ``>janneke: also, how are you liking the test structure in MESCC_Tools? <mekeor>i made "overlayed" ghc-xmonad-contrib with my own package-definition which uses (local-file "/path/to/my/xmonad-contrib") where i added some functions to a module. but after installing it or creating an environment using that package, those functions still aren't available :( <mekeor>(and yes, i did export the new functions: the new functions are even listed in the /gnu/store/…-ghc-xmonad-contrib-0.13/share/doc/ghc-xmonad-contrib-0.13/html/….html documentation file) <mekeor>$GHC_PACKAGE_PATH is holding the wrong path <mekeor>aaah, maybe it's because i have installed xmonad system-wide and per-user <mekeor>so, i have to care about the order of sourcing /etc/profile and ~/.guix-profile/etc/profile <janneke>wow, you're going fast there -- i am aming as a big next target mes 0.8 release with hex2 x86 <janneke>about the test setup...ime tests are very hard to get right: when you discover you need to make changes, refactoring a test-suite is not trtivial <janneke>mescc's test "suite" is mainly scaffold/t.c--with hex2 that can be refactored into something more sensible <eacces>have i got this right: `guix system reconfigure blah.scm` is how one normally upgrades system wide things [eg: kernel] <reepca>eacces: aye, though you'll need root access <rekado_>buenouanq: and you may need to install gvfs into the system profile <rekado_>this works for our devices that use MTP <rekado_>gmtp is terribly slow, so I don’t recommend using it. <rekado_>buenouanq: gmtp reads from the usb device file, which is owned by root. I suppose you could change this with a udev rule. <rekado_>mekeor: is a different version of ghc-xmonad-contrib pulled in by a different package? Or do you only have your variant in the profile? <rekado_>does anyone here know how to build libreboot from source with Guix? <rekado_>I’m not even trying to package it yet; I just want to create an environment with all the dependencies. <rekado_>hmm, so many substitutes are missing. <rekado_>I’m compiling arm-none-eabi-toolchain, qtbase, all of the icedtea packages, gtk … <janneke>rekado_: icedtea is a dependency of libreboot? <rekado_>no, it’s needed for f-seq, which I wanted to push. <rekado_>I also had to build sablevm, so the complete bootstrap :-( <rekado_>probably because sablevm-classpath needs gtk+-2 <reepca>anyone else getting a lot of "secure connection failed" page load errors with icecat recently? For example, this page I just tried accessing from stackexchange: https://oeis.org/A000458 <RX14>loads for me on chromium outside guix <reepca>So is icecat being zealous for my safety, paranoid, or broken? <catonano>it happens oftn that Icecat won't load some resources associated to a web page, such as styleseets or javascript sources <espectalll>Quick question – I have downloaded and setup the GuixSD VM image onto Virtualbox, expanded to 20GB HDD and 1'5 GB of RAM <catonano>so many sites are broken in Icecat bt they workk like a charm on oter browsers <espectalll>I'm trying to do a task as simple as creating an user, but there is no `useradd` command <reepca>espectalll: the idea is to handle that through your config.scm <espectalll>oh wait, the `config.scm` keeps stored after installation? <reepca>espectalll: yeah, it's analogous to a .emacs file - you keep updating it and it grows to describe your system <reepca>only this time instead of "guix system init", you use "guix system reconfigure" to enact it <reepca>well, the install manual recommended copying it to /etc <RX14>reepca, is there any reason why there isn't a standard location for the file? <espectalll>But I installed my system via the precompiled image at gnu.org <rekado_>RX14: it’s because no standard location is required. <rekado_>RX14: I keep it under /etc/config.d/config.scm so that I can put all related files under /etc/config.d/ <reepca>espectalll: no, that's actually a guile module used internally by guix for configuring stuff at compile-time <espectalll>hopefully the system is empty enough so creating my own won't break any previous configuration <reepca>Err, hang on, you're using the vm image? <reepca>ah, well the vm image if I understand it correctly is basically an installer. You followed the instructions in 6.1.6 "Installing GuixSD in a Virtual Machine"? <espectalll>it's a working Guix image that even recommends you to set a `root` password and expand its size when booting <espectalll>(I can use it to install a GuixSD system as it has the package manager, duh) <reepca>Oh, I see, I was looking at "installing GuixSD in a VM", not at "running GuixSD in a Virtual Machine". <espectalll>So, is there even a template I can use, as my system has currently no editor other than `nano`? <espectalll>Or do I make random guesses by copying other existing `config.scm`'s and then apply them? <RX14>there's a good few example config.scm files in the docs <RX14>i found them quite helpful <reepca>In the guix repository, we've got gnu/system/examples/vm-image.tmpl, I'm not sure if that's what's used in the official vm image, but probably worth a look <espectalll>If I'm successful with this, I'll try to install GuixSD natively <espectalll>I know my hardware is compatible and have been following the project for years already, I just need to know I can use tools like IntelliJ IDEA <rekado_>espectalll: we don’t have a package for that. <rekado_>if you have the jars for it you should be able to use our icedtea packages with it. <espectalll>(yup, it's written with Java, and has an open-source version) <reepca>RX14: they're all hiding in gnu/packages/, and each file usually has many packages. <RX14>oh yeah i see there's a lot of packages in each file <RX14>is it easy to get a package in there? <reepca>jamesrichardson: I finally got around to installing and starting libreoffice - the fonts look okay to me <RX14>i was more asking how likely it is my patch will get accepted <RX14>if i packaged some random thing that noone else needed would it get rejected for being obscure for example? <RX14>i'm just trying to gauge if you want every package ever to be in the one repository or not <reepca>someone else probably will though <janneke>RX14: why would you want to package something noone will ever use? <RX14>well i meant apart from me <RX14>so that I can install it i guess <janneke>RX14: getting a package good enough for inclusion always takes some more work than doing the bare minimum for yourself <janneke>RX14: it's easy to keep your own set of patches with packages on top of guix git, or you can use GUIX_PACKAGE_PATH for those <janneke>RX14: i don't think i've ever seen the 'this is too obscure' but please use your own judgement here <RX14>i'll have a go at packaging crystal-lang once I have my system set up <RX14>no it's not i was just asking for the sake of asking <RX14>crystal itself uses make although it might be useful for packaging shards <espectalll>keep in mind Guix is packaging Python modules, Rust crates and so on <RX14>although actually this might be a lot harder than I thought given that crystal is a bootstrapped compiler <mekeor>on GuixSD, any idea why my system (e.g. `date`) thinks it's two hours later than it is where i am? <RX14>you might have the wrong timezone set <reepca>mekeor: timezone not properly set in config.scm? <mekeor>it is properly set to (Europe/Berlin)... <mekeor>maybe i have to decrease the time in BIOS <RX14>windows sets the system clock to localtime but linux treats it as utc but if you don't have windows then thats weird <RX14>just enable ntp and let it sort itself out i guess <espectalll>probably because the system used to have Windows installed? <rekado_>mekeor: Do not install ntp, use the ntp service. <rekado_>mekeor: and allow it to correct the time even when it is far off. <mekeor>i.e. #:allow-large-adjustment? #t <mekeor>ah, i already have ntp-service in my system config.scm, i guess i only need to add #:allow-large-adjustment? #t <RX14>i just ran guix pull a few times and then added openssl to my services and tried to reconfigure <mekeor>RX14: that error was reported on the mailing list <mekeor>RX14: i think XFCE current is broken because it depends on different versions of gtk or so <mekeor>RX14: see bug#27467 on the bug-mailing-list or on debbugs <RX14>i'm only going to be using i3wm on my real system anyway <RX14>so i might as well remove both and use a non-desktop config <RX14>because thats what the default desktop cfg comes with <RX14>there's a lot of building going on whenever I do stuff with guix <RX14>it seems to fail to find substitutes pretty often <OriansJ``>reepca: also, you'll have LSHIFT and RSHIFT shortly <janneke>OriansJ``: finally got some progress eradicating hex3 <OriansJ``>still waking up but this should be a good day <espectalll>Hey, I tried to apply my new `config.scm`, but I get the following error <RX14>does guix system reconfigure need to be run as root? <RX14>is that connman-service-type doesn't have a default value of connman-configuration a bug? <RX14>i need to include an empty (connman-configuration) where in other service types i don't <catonano>RX14: I don't know. It depends on how that service is defined <espectalll>catonato: Can I do a POST to upload a file via `curl` or `wget`? <RX14>catonano, well i guess I can send a patch and see what people think? <OriansJ``>reepca: I must say, you are doing an excellent job improving the stage0 project FORTH <cbaines>espectalll, I think I see a problem in your configuration, the services section is a bit jumbled <cbaines>cons* takes a number of things, and then a list <cbaines>the third argument (%desktop-services) is a list, so that really wants to go at the end <cbaines>espectalll, I'd suggest replacing %base-services with %desktop-services <espectalll>OK, I was suspicious about that, but I don't know Scheme that much <cbaines>I believe %desktop-services includes %base-services <boskovits>I have seen a discussion thread from 2015/04, but it seems work was not finished on this. <cbaines>out of interest, what is interesting about LVM boskovits? <cbaines>I've only used it when its been automatically setup by the Debian installer for a encrypted disk <cbaines>espectalll, Yeah, that is looking better <cbaines>The * in cons* is essential, as cons only works when you have two arguments <espectalll>oh I see, by default Lisp langs only accept two arguments? <cbaines>a list in Lisp is a linked list, made up of pairs, so to add an element to the start of a list, you can use cons. e.g. (cons 1 '(2 3)) => '(1 2 3) <cbaines>but if you want to add more than one element, then cons* comes in useful e.g. (cons* 1 2 '(3 4)) <boskovits>I mostly use it to extend existing partitons to new physical disks. <boskovits>I also use it to create lvm volumes dinamically to allocate it as block storage. <cbaines>espectalll, I'm not sure what syntactic sugar exactly means in the context of scheme, but you could replace any use of cons* by multiple cons calls. e.g. (cons 1 (cons 2 '(3 4))) <cbaines>boskovits, I've heard that GuixSD supports Btrfs, which I've used to do similar things before, but I'm not sure how well tested that is, if you're brave, it may be interesting? <espectalll>which I think fits in the definition of "syntactic sugar" <boskovits>And anyways, I would like to help, so it can be removed from limitations:) <boskovits>Yes, I'm definietly into btrfs, and I'm glad that it is supported now. ***jonsger1 is now known as jonsger
<cbaines>Supporting LVM sounds great, have you used GuixSD yet? <boskovits>I am currently in a debian envrionment where I use btrfs over lvm, and I would like to repoduce the setup on GuixSD. <cbaines>Well, I don't really know much about the LVM bit, but that sounds like a great start :) <cbaines>I guess trying to make it work in a virtual machine would probably make sense? <boskovits>What I would like to know, why the work on lvm support stopped, so <OriansJ``>reepca: and RSHIFT & LSHIFT are implemented, patches will be up shortly <OriansJ``>janneke: the really good news is that reepca is well towards making a rather complete ANS FORTH with all the improvements going into the stage2 and stage3 FORTHs <earthfail22>hello guys. I was wondering has anyone successfully installed guixsd on a pc? cause it isn't working for me <cbaines>earthfail22, I managed to finish an installation this morning, what specifically isn't working for you? <janneke>OriansJ``: i've seen bits of reepca's and your work there <janneke>i have no feeling for forth or where it will be used yet <rekado_>earthfail22: many people have installed GuixSD on many machines. <earthfail22>cbaines: the gnu documents were some what vague and first time I tried my keyboard didn't connect the other times there was a problem with something couldn't build in the gnu/store <rekado_>earthfail22: I have it on many x86_64 and i686 systems. <cbaines>earthfail22, while my install over the last few days didn't go very smoothly either, I can assure you it is possible <cbaines>are you trying to install GuixSD at the moment? <earthfail22>yes and it would be really great it someone could guide me a bit. I've made a bootable usb drive <OriansJ``>janneke: the best part is the people who want FORTH to be the key to the bootstrap problem are responsible for that work. Personally I prefer the lisp approach and am still formulating how to best attack it. <cbaines>earthfail22, sounds good, I'd recommend just posting in IRC if you have questions <rekado_>earthfail22: have you read the instructions from the manual? <rekado_>earthfail22: which of the steps causes problems? <rekado_>ACTION decided to sell the old IBM P70 (with Intel 80386 DX) <earthfail22>rekado_: yes , it failed when trying guix system init and when rebooting <janneke>OriansJ``: okay, anything is possible -- good! <janneke>fwiw, i'm quite happy to see all this bootstrap-enthusiasm <earthfail22>rekado_: it said something about failing to build something in gnu/store and should try --fallback which I did but didn't work and could not remember what was the second message <OriansJ``>janneke: well at this point we have rain1, reepca, myself and you all making improvements and pushing in the right direction. I call that a huge improvement <janneke>OriansJ``: yes, my thoughts exactly! <janneke>OriansJ``: and we have mwette doing great work with Nyacc <earthfail22>rekado_: anyway I will try it again cause it will probably give me the same message now <OriansJ``>janneke: very true and that work is essential for our goals <civodul>OriansJ``: make sure reepca doesn't lose sight of the GSoC work on the daemon though :-) <OriansJ``>sneek: later tell reepca that I've added a new primitive PRINT which takes a pointer and from there prints all characters until it hits a NULL and there will be a proper abort message when insufficient memory is allocated. <civodul>every time i run a long verbose build in shell-mode, Emacs eventualy eats up all my memory <civodul>and the kernel doesn't react in a timely fashion, so i end up doing a hard reboot <civodul>it used to be better, i think (like emacs would slow down until you kill the offending buffer) <janneke>civodul: the hard reboot luckily not; but emacs eats all, lately <janneke>civodul: 32038 janneke 20 0 8787.4m 8.330g 0.0 54.5 38:58.87 S `- .emacs-25.2-rea <cbaines>I've had similar problems... although I think my laptop just slows to a halt <janneke>i used to live in the same emacs for weeks... <rekado_>I now build in urxvt, which is much less fun. <RX14>guixsd actually hangs on shutdown in my vm, every time <civodul>janneke: yeah it really sucks that emacs has become unstable <janneke>civodul: yeah, otoh it might mean some real developments, i'm not in touch there <rain1>I finished the qcode based scheme compiler - it takes 51 seconds to bootstrap. a bit disappointing if you add up the time to compile each stdlib function it is slower than the previous bytecode based one <rain1>so unless i can sort this i am not sure this stuff will be able to help accerelate mes ***jonsger1 is now known as jonsger
<OriansJ``>rain1: but your scheme compiler is still potentially useful <rain1>I hope so.. i will get this chaotic mess tidied up and put something online <OriansJ``>rain1: with some work and a few enhancements in stage0 lisp, it could give us a lisp solution to the mes bootstrap. <janneke>OriansJ``, rain1: that's what i'm hoping for <quiliro>what packages do i need to install in guixsd in order to use emacs for navigation, email and making new packages for guix? <quiliro>espectalll: haha, thank you....is there any specific config you would suggest? <espectalll>…Emacs comes bundled with Gnus, Eww, and Guix mode, right? <quiliro>and is there a way to watch videos, and listen to audios n emacs? <espectalll>To listen to audio, there's a few stuff out there, never tried it out <espectalll>and the Guix version of Emacs comes with a frontend for the package manager, so you can handle Guix stuff from Emacs <espectalll>Scheme is a programming language heavily used by Guix <quiliro>espectalll: yes, i remember now...thank you for your guide <espectalll>Oh, and the Eww browser may not be enough for you <quiliro>i will just install emacs then plus vlc <espectalll>there's luckily now a pre-bundled way to run a full, WebKit browser <espectalll>dunno if Guix has the compile flags enabled for it tho <espectalll>OK, the WebKit mode for Emacs is called `xwidget-webkit` <rekado_>quiliro: for music and video I suggest emms <quiliro>rekado_: so emacs with emms and mvp? <quiliro>espectalll: how do i use all these things in emacs? <rekado_>espectalll: yes, I’m actually working on it. <espectalll>If you're new, I would highly suggest going to the Emacs manual for help <espectalll>You can find it by either searching on the web or doing C-h r (Ctrl-H followed by R) <quiliro>espectalll: i have used emacs only once <espectalll>I could also suggest the book "Mastering Emacs" by Mickey Petersen <rekado_>I second the recommendation for “Mastering Emacs” <espectalll>at first you can do most stuff just by using the menu buttons and M-x (Alt-X) followed by your command <espectalll>But actually learning how to Emacs can be really, really valuable <espectalll>I, at the very least, find its customability and documentation to be too valuable to ignore <quiliro>thank you both...i will find where to download it <OriansJ``>and patches to the stage0 lisp for proper behavior when insufficient memory is allocated are up <espectalll>Is it normal that Guix just keeps compiling stuff? <espectalll>I have Hydra subsitutes enable, yet Guix just keeps going for hours <RX14>espectalll, i've noticed that, it's not all packages <RX14>but say one in 10/20 seem to fail to find substitutes <espectalll>So not all packages are still being precompiled by the Hydra build farm? <lfam>They aren't compiled yet. But it's okay, because Guix will automatically switch between downloading the binaries from Hydra and building locally :) <lfam>espectalll: You're using the 0.13.0 release image, right? <lfam>Did you run `guix pull` yet or not? <RX14>if you run guix pull aren't you on master <lfam>espectalll: If you ran `guix pull`, then you are using the latest Guix code. It's possible we haven't finished building everything yet. <lfam>RX14: That's correct, `guix pull`, by default, updates to the latest commit of the master branch. <espectalll>in fact, trying to browse the queue gives me an 504 error <RX14>wouldn't know if it was built befause the hydra queue times out <lfam>Yes, that machine is quite overloaded <lfam>We are working on replacing it <espectalll>if only I could have an effective way to throw money at the screen <espectalll>(since you can use the package manager itself to create a repo) <RX14>i meant which services lol <lfam>It's not really a money issue. It's a time issue. <RX14>also surely the builds are done on seperate machines to the master <lfam>espectall: How did it fail? <lfam>espectalll: Was it the grub_cmd_set_date test? <lfam>Yes, that test is flaky. We thought it might be caused by a race condition in the tests but it's occuring even after we disabled parallel testing. I'm going to submit a patch to disable that test. <civodul>lfam: i spent time on that test but i forgot what the outcome was <espectalll>I'll check tomorrow if GuixSD can be my daily driver then <koosha>rekado_: I emailed the mailing list , it's one day and no answer . <lfam>espectalll: The test will probably pass if you run the build again. <lfam>civodul: The test is still failing even with sequential testing <civodul>lfam: oh right, just saw your message there <lfam>We should probably disable it and report the issue to GRUB <lfam>I can do that if you think it's safe to disable <lfam>So many upstreams to tell about so many problems :) <lfam>Mo packages, mo problems ;) <lfam>Hopefully more contributors too! <civodul>i wonder what's up with core-updates <civodul>perhaps we'll have to apply the glibc suid/LD_LIBRARY_PATH patch there? <lfam>civodul: I remember you said you weren't sure if we were affected, due to how "_is_trusted_path" works. However, I don't know enough to evaluate it. <lfam>So, I pushed for the patch, just in case <civodul>lfam: i looked at the libc code because i was surprise LD_LIBRARY_PATH was honored for setuid libraries in the first place <lfam>We could graft it on core-updates. There is at least one other graft to cherry-pick to core-updates (expat). <civodul>what i found is that it's honored, but only entries that pass 'is_trusted_path' are taken into account <civodul>and is_trusted_path means libc's $libdir, which is immutable in our case <civodul>however, the way the LD_LIBRARY_PATH attack works, i'm not sure it matters <civodul>long story short: better apply the patch <lfam>Indeed! I suspect we'll see more vulnerabilities from this bug class :( <lfam>I mean, indeed it's complicated! <civodul>what upstream did is disable LD_LIBRARY_PATH for setuid binaries altogether, right? <lfam>The subject of the patch is "Ignore LD_LIBRARY_PATH for AT_SECURE=1 programs" <lfam>I'd have to read more code to know what AT_SECURE is <lfam>I'm currently idle at work, so I can't really read anything carefully this week <lfam>This environment is too distracting <lfam>And I sent a bug report to GRUB <civodul>glad you're idle at work, it's good for Guix :-) <lfam>Heh, I can do stuff that doesn't require much focus <lfam>"Sponsored" work on Guix ;) <efraim>I got an email from Bruno Haible of clisp, letting me know that libffcall 1.13 was released with arm64 support :) <rain1>OriansJ``, janneke, update it now takes 1 second <earthfail22>hello guys, I have a bios 32bit system any suggestion how to partition my drive. I've read the manual but I'm still new and didn't understand what to do with the bios <civodul>earthfail22: if you're not familiar with GNU/Linux partitioning, the simplest way is to have just one partition ***quiliro is now known as quiliro1
<quiliro>ACTION congratulates himself and all the people that guided him. <quiliro>civodul: my first step to become a guix hacker! <quiliro>how can i access the guix info from emacs <sneek>Welcome back reepca, you have 1 message. <sneek>reepca, OriansJ`` says: that I've added a new primitive PRINT which takes a pointer and from there prints all characters until it hits a NULL and there will be a proper abort message when insufficient memory is allocated. <civodul>quiliro: though of course, using an IRC client in Emacs (even if it rocks ;-)) is optional to hack on Guix :-) <quiliro>to type commands such as guix package -i emacs-guix <quiliro>or what part of the guix info covers that? <reepca>The way I do it is M-x guix-packages-by-name and then there's an install button you can push RET on <reepca>If you've installed emacs-guix, it should be in the top-level info directory (it's in a separate node from the info pages that come with guix) <quiliro>reepca: it is not installed...i just guix system reconfigure lightweight-desktop.scm <reepca>ah, well you'll have to guix package -i emacs-guix to install the emacs interface to guix <quiliro>i cannot even configure the keyboard <quiliro>i am installing claws-mail now via emacs-guix...i feel so great! <quiliro>and it is such a simple thing...haha 