IRC channel logs


back to list of logs

<swedebugia>fps, unfortunately I don't understand how to do what I want as the examples are not near what I want. I think I have to use (map (match-lambda and that is not among the examples...
<fps>you have a two lists, a list of packages and a list of blacklisted items?
<fps>and for each package in the list of packages you have a list of dependencies?
<fps>i'm not a scheme hacker of any capacity, just trying to clarify the problem :)
<fps>so, do you have working code using map-lambda that returns a function which matches fine for an individual package?
<fps>you might have one indirection too many. you could just (map (your-matcher ....) list-of-packages), no?
<fps>nah, just ignore what i just said.. ;)
<swedebugia>fps, thanks for trying to help :) I posted to the list. We will see what comes out. I have a function in npm.scm which calls a proc: blacklisted? that returns #t if the package in question is in the blacklist.
<swedebugia>Now I want to regex match against the list so rollup matches rollup-plugin because they begin the same.
<swedebugia>goodnight :)
***jonsger1 is now known as jonsger
<Swedneck>alright, i've got guix installed in my fedora 29 VM :D
<enderby>has anyone else's 'guix pull' command taken 20+ hours
<bavier>Swedneck: great!
***catonano_ is now known as catonano
<apteryx>enderby: hmm, it used to be very long, but 20 hours, hell no :-)
<apteryx>on what kind of hardware is that?
***rekado_ is now known as rekado
<rekado>enderby: it also depends on what it’s doing.
<rekado>enderby: if you end up compiling large amounts of build dependencies on slow hardware it can take quite a while.
<rekado>if that happens you should double check that substitutes are enabled and that you are fetching substitutes from as well (because that’s where substitutes for the “guix pull” derivations are built).
<enderby>rekado: thanks. I just ran it again and it finished in a normal amount of time, so idk what it was doing last time
<bavier>rekado: do you mind if I update the neovim package?
<rekado>bavier: I don’t mind. (I’m not using it myself.)
<bavier>ok, just saw you had added it, np :)
<rekado>heh, a user at the institute had asked for it back then.
*rekado rebases the wip-gnome-upgrades branch
<jbw1111>Hi, I've lost my root password for my guixsd machine so I'm trying to get into single user mode, but I'm failing. Appending "single", "1" or "init=/bin/sh" to the linux line in grub doesn't seem to change anything, and still ends up with shepherd booting up normally and leaving me at the login prompt.
<jbw1111>What do I have to do to get into single user mode in Guixsd? Alternatively I could boot from a liveusb and chroot.
<fps>jbw1111: i'm not sure. but take a look into the shepherd manual
<fps>it's guixsd's init system
<jbw1111>I managed to boot into an early boot guile shell, but it was pretty impossible to change my root password from there. I ended up using the installation usb to chroot and change the password
<civodul>Hello Guix!
<rekado>wip-gnome-upgrades no longer builds fine.
<rekado>A couple of packages with meson-build-system don’t include the subdirectories of $out/lib any more.
<rekado>patching those now.
<civodul>rekado: oh, is this due to changes that were made in the meantime?
<rekado>seems so
<efraim>nss doesn't build on master anymore and I gc'd the one I had on my aarch64 board
<rekado>it used to build on the old core-updates that it was initially based on.
<rekado>just rebased it to current core-updates
<rekado>I hope I’ll get to test the newer GNOME soon.
<efraim>i'm working on removing ffmpeg-2.8 from core-updates, just have to test that the freerdp update didn't break vinagre
<efraim> speaking of meson, I modified the "dump the testsuite.log" code from gnu-build-system for meson
<civodul> <- armhf-linux
<civodul>efraim: i pushed a fixed for nss in core-updates
<civodul>a few days ago
<civodul>efraim: BTW, i re-plugged the overdrive that you host in berlin, and it seems to be working well so far
<civodul>could you check once in a while if networking is okay?
<efraim>civodul: yep, I see it every time I go up the stairs. I turned it back on about 2 hours ago
<efraim>it seems to have heat issues
<rekado>civodul: I asked IT to give me the complete firewall rules so that we can compare them with our requirements.
<civodul>rekado: cool, thanks! for now the armhf builds take place on the overdrives
<civodul>efraim: if you find that it becomes too unreliable, please tell rekado or myself so we can remove it from machines.scm before it becomes a problem
<vagrantc>civodul: do you set the personality to linux32 or something? there are a number of things that will build incorrectly with the wrong kernel
<civodul>vagrantc: yes the daemon uses personality(2)
<civodul>like what would build incorrectly?
<vagrantc>some programs will, say, enable 64-bit features even on a 32-bit build...
<vagrantc>or at least try to
<vagrantc>some silent, evil bugs
<vagrantc>there was literally a post earlier today/yesterday about one
<civodul>vagrantc: do you have a link?
<civodul>i mean, how is it different from i686 builds on x86_64?
<vagrantc>civodul: different specific problems, same types of issues :)
<roptat>hi guix!
<vagrantc>civodul: the i386 builds running an amd64 kernel for debian's reproducible builds has found a few issues relating to embedding the kernel options, but maybe setting the personality fixes that
<vagrantc>any idea off-hand how much disk space linux-libre 4.19.5 takes on aarch64 ?
<vagrantc>i mean the build process, not the resulting /gnu/store/ ... thingy
<civodul>vagrantc: re kernel details in the build output, these are bugs in general for Guix
<civodul>i don't know how much disk space is needed for linux-libre
<civodul>probably quite a lot :-)
<vagrantc>after days, i've got guix pull to a recent guix, and u-boot-pinebook built, and building linux-libre ... i might actually be able to test guixSD on the pinebook eventually :)
<vagrantc>i should just build a test image on a faster machine ...
<rockandska>Hi Guix
<rockandska>I previously asked on ML if there was any possibility to read a manifest file from stdin but didn't had an answer, is there anyway or plan to implementing it ?
<civodul>vagrantc: soonish we'll turn on builds for the "modular" jobset (i.e., guix pull) for aarch64/armhf
<civodul>so hopefully you'll be able to pull much more quickly :-)
<vagrantc>and that will also bring in a handful of basic dependencies for other packages too
<vagrantc>did the substitute servers used to have more aarch64 builds?
<civodul> has aarch64 substitutes for core-updates but not for master
<civodul>hydra has armhf but not aarch64
<vagrantc>so if i were living even more on the edge, i could get substitutes :)
<civodul>exactly, that's the deal :-)
<brendyyn>Nix's full system tests, like running bittorrent tests seem super cool!
<civodul>wait, Guix system tests are super cool too ;-)
<brendyyn>It would be great to develop similar tests for guix, maybe even a sort of high level fuzz test, like specifying parameters and generating random virtual machines with various network configurations etc and defining some automated test
<brendyyn>haha, I don't think I've seen them yet
<civodul>it's improved since that article
<brendyyn>cool. maybe ill come up with a test do define one day
<rockandska>civodul: Hi, any thoughts about my previous question ?
<civodul>hi rockandska
<civodul>rockandska: ~/.guix-profile/manifest cannot be passed to "guix package"
<civodul>there's no concrete plan to implement it
<civodul>instead i'd suggest using the --manifest option that already exists
<rockandska>civodul: yeah i know, we discuss about it a few months ago, that was not my question here :)
<civodul>oh so i misunderstood your question
<rockandska>civodul: but even with --manifest, we can't provide it through stdin or a fiel descriptor
<civodul>ooh, got it
<civodul>i guess we could support it, yes
<civodul>do you feel like it's an important feature?
<rockandska>civodul: for my little project / use case yeah. I already explained it on ML but could provide you the link where i explain my use case
<civodul>well no, i'll read the message when i get there
<rockandska>civodul: it was explained when we discussed about the confusion between --manifest and ~/.guix-profile/manifest
<roptat>civodul: no news from the TP?
<civodul>roptat: no; you did receive my second email tho, right?
<roptat>yes I did
<civodul>maybe i need to change the subject line
<g_bor>hello guix!
<civodul>roptat: resent
<civodul>hi g_bor!
<civodul>roptat: BTW please take a look at the TRANSLATORS comment in guix.texi since you won't see it in the pot file ;-)
<roptat>ok, c'est reparti
<roptat>wrong chan ^^'
<civodul>thoughts on --with-branch?
<janneke>civodul: very nice!
<civodul>janneke: cool, thanks
<civodul>colleagues also asked for --with-commit so i'm adding it now
<civodul>it's never enough ;-)
<janneke>hehe :-)
<jonsger>civodul: --with-branch=guile-gcrypt=master looks a little strange, maybe --with-branch=guile-gcrypt:master is better
*janneke adds `mv' command to gash
<civodul>jonsger: oh, that was for consistency with --with-input
*rekado reconfigures with the new GNOME
<rekado>I just pushed a new version of wip-gnome-upgrades, rebased on top of core-updates. Feel free to test this on your own systems!
<baconicsynergy>hello my dear compatriots :)
<baconicsynergy>is hydra back up and running?
<fps>i think so
<baconicsynergy>coolio, i'll try installing guix again
<baconicsynergy>yay everything is working now :)
<baconicsynergy>parabola just wasn't complete without guix :p
<rekado>I hit another problem with tensorflow.
<baconicsynergy>I'm so proud of my laptop and software right now ;_; 100% RYF hardware, 100% free software. Full disk encryption with libreboot, bleeding edge software with parabola and reproducible builds with guix. so happy
<rekado>The cmake build system doesn’t install any libraries; worse, it doesn’t even build shared libraries at all.
<rekado>everything is built as objects that are then used to statically link useless executables.
<rekado>the cmake stuff is deprecated; going forward they will remove all cmake files and make it only work with bazer.
<rekado>…which we cannot build from source yet.
<rekado>guess I’ll have to try to make progress on bazel then :(
<rekado>this might contain a solution:
<rekado>does globbing work when using “invoke”?
<rekado>I have this: (invoke "ar" "-r" "libtensorflow.a" "*.o")
<rekado>or should I splice a find-files list?
<rekado>rebuild tensorflow again. Let’s see what happens.
<civodul>rekado: nope, there's no shell involved so no globbing
<civodul>instead you have to use find-files, indeed
<rekado>civodul: do you know where I should submit my talk proposal for FOSDEM?
<rekado>I accessed penta, but I can only see FOSDEM 2016 there.
<rekado>on the devroom page I see no contact either.
<civodul>rekado: not sure exactly, it should be on Penta
*civodul hasn't submitted a talk yet
<ng0>where do we keep the binary gdmd?
<ng0>maybe I'm approaching this the wrong(most complicated) way, for now. Is there a better way to process swf video files outside of a browser than to dissasemble them and work the host specific swf into your own fork of youtube-dl?
<rekado>ng0: have you tried
<rekado>ng0: what kind of thing should “gdmd” be?
<ng0>yes, but last time I tried gnash was many years ago and it wasn't usable. now flash is on its way out, maybe gnash got better.
<ng0>idk, something a compilation of RADCDAsm with rdmd is throwing at me
*civodul doesn't know what gmd and RADCDAsm are
<civodul>gdmd, even
<rekado>rdmd is a disassembler (framework), no?
<ng0>got angry at the flash site a while back and I started looking at this:
<rekado>(there are so many tools that use this arrangement of letters)
<rekado>it’s probably easier / more productive to try gnash and then make a few changes if the file isn’t fully supported.
<rekado>the readme gives you the answer
<rekado>“Assuming you have git and a D2 compiler, such as dmd or gdc installed, compiling should be as straight-forward as: …”
<rekado>“Substitute dmd with gdmd if you're using gdc”
<rekado>so: it is the executable of the D compiler.
<civodul>oh, that one
<ng0>hmh.. ok. thanks
*janneke has now reached scheme-only bootstrap install phase of mes
<civodul>i need to look at the bootstrap tarballs again
*civodul finally upgrades whole user profile to core-updates \o/
<civodul>speaking of which, is anyone still using Conkeror?
<rekado>I thought it died with the new Firefox?
<rekado>with the birth of the new Firefox, that is
<civodul>yes, so i keep an old version, which is obviously very bad
<civodul>i guess i'll have to stop, but it saddens me
<civodul>it also means we may have to remove the package altogether?
<rekado>I guess so.
<baconicsynergy>noob question: where should I place "export PATH="/home/foo/.guix-profile/bin${PATH:+:}$PATH" to make it persistant?
<baconicsynergy>well, i know several places, but whats the most appropriate?
<janneke>civodul: i haven't found a more keyboard-friendly graphical browser than conkeror
<civodul>oh, tell me!
<janneke>browsing the web, i usually do M-x eww, conkeror, *enable touchpad* icecat
<civodul>baconicsynergy: you should ". ~/.guix-profile/etc/profile" from .bash_profile, see
<baconicsynergy>civodul, thanks yet again!
<janneke>tell you what? conkeror has emacs-like bindings and can be used without a mouse
<civodul>i meant tell me which browsers you use
<janneke>oh :-)
<civodul>i use emacs-w3m and conkeror mostly
<janneke>yeah, if those fail, fallback to icecat and mouse-shoving
<janneke>oh, i dropped emacs-w3m for eww, it has improved
<rekado>I use eww only very rarely.
<rekado>I’m again between browsers. I use *both* epiphany and icecat (because I have used tabs in both), but I’m not happy with either of them.
<fps>i wonder how much duplication is in nars in a typical system..
<baconicsynergy>I wish icecat had more people working on it. it's a noble goal for a web browser
<rekado>can we “just” wrap webkit in
<rekado>(like it’s done in the example)
<fps>i.e. how much space savings would we win if we would push the binary distribution [e.g. via ipfs] to a lower level than nars.. i.e. distribute the nar contents on ipfs which would do the deduplication..
<rekado>ooh, nice: “The following grafts will be made” — I like it!
<rekado>fps: that’s a good point.
<rekado>I’d guess that there’s quite a bit of duplication there.
<fps>[the ipfs-nar could be a meta file describing the CIDs of the contents]
<fps>i wonder what precisely this tells me:
<fps>du cache/ -h
<rekado>store deduplication saves me dozens of GB.
<fps>8.4G cache/
<fps>du -sh ~/.ipfs/
<fps>4.5G /home/fps/.ipfs/
<fps>and the complete content of cache is added to ipfs
<fps>oh, that's even on the level of blocks with nar files.. without even splitting out the contents
<fps>[where cache is the directory i store the narinfos and nars i pulled from berlin in]
<fps>rekado: there's store deduplication in the store already?
<fps>redundant redundancies are redundant :)
<civodul>fps: what throughput can one hope for over IPFS?
<civodul>is it supposed to be super-mega-fast like Bittorrent?
<civodul>(at least for popular content)
<fps>civodul: that's the idea.. let me try something.. :)
<nly>I tried adding contents of my /nix/store to /ipfs/ which must have been 15Gb-ish, left it overnight and it didn't finish, sadly, before i killed it :(
<civodul>nly: oh, good to know :-/
<fps>the 8G are added in a couple of minutes in my case.. hmm
<fps>locally i retrieved a file of 138M in 18s..
<fps>from the same ipfs node i added it to.. which tells us not much ;)
<nly>rate of 8GB in minutes is awesome :D
<nly>what kind of hardware that is?
<fps>some vm on a hetzner root server
<fps>i completely fail to get the same CID from my home machine right now though... the swarm has size 1 though ;)
<fps>civodul: interesting. right now after you asked it seems ipfs doesn't provide any throughput AT ALL ;)
<fps>brb, groceries..
<rekado>bleh, the tensorflow build failed at the last moment because of a missing -fPIC somewhere.
<rekado>so close!
*civodul cheers rekado
<civodul>you'll make it, i'm sure you'll make it! :-)
<civodul>and you'll make a lot of people happy, too
<rockandska>civodul: did you had the time to take a look at the need to have the possibility to read the manifest file from stdin and/or open an issue in case someone will have the time to take a look at it please ? :)
<rekado>rockandska: is there a bug report about this?
<rekado>that might be helpful
<roptat>civodul: we have the new po files available \o/
<ng0>btw: guix is hip enough now to be recognized by neofetch :D
<ng0>Packages: 1435 (emerge), 131 (nix), 733 (guix)
<nly>I made a change in gnu/packages/audio.scm but do i need to run ./bootstrap; ./configure; again ?
<civodul>rekado: re FOSDEM, you need to go to
<civodul>roptat: yay!
<civodul>roptat: you should blog about our i18n efforts!
<roptat>civodul: I'll think about it :)
<roptat>we still have no idea how to localize the website, right?
<fps>hmm, it seems i drank the coolaid a bit too early.. downloading content from my cache via ipfs is abyssimally slow.. since i'm the only one requestiing the content and a single node is providing it it should boil down to a point to point connection
<fps>spectacular fail..
<rockandska>rekado: just talk about it on devel when we talked about the confusion with profile manifest and --manifest, but i will write about it specifically in a bug report :)
<fps>[ports are forwarded just fine on both ends]
<rockandska>I need to go, cu next time :)
<fps>i wonder how the ipfs people modified the original kademlia XOR routing to make it so bad :))
<fps>slightly jumping to conclusions here..
<civodul>roptat: i thought we did? IIRC someone (Florian?) came up with a complete solution a while back
<civodul>my memory is fuzzy though
<nly>'./pre-inst-env guix package' is not showing the changes in my checkout?
<bavier>does our icecat include the "Tor Browser Button" by default now?
<rekado>hmm, the new gnome-shell fails to start
*rekado packages emacsy
<bavier>I have a project here that says some files are under gplv2+, but that the project as a whole is under gplv2. Does that make sense?
<bavier>can anyone give a concise summary of the differences between ffmpeg and libav?
***slyfox_ is now known as slyfox
<mbakke>bavier: 'ceph' has a mixture of GPL2/GPL2+ too.
<mbakke>I think the libav fork is no longer maintained, no?
<mbakke>Oh, is still active.
<yrk>bavier: mixing gplv2-only and gplv2+ is fine. in that case the work, as a whole, can only be distributed under gplv2
<yrk>bavier: the gplv2+ parts can be taken out, upgraded to gplv3+ and included in a separate gplv3 work
<bavier>yrk: oh ok, makes sense
<yrk>bavier: this is because the so-called "+" merely provides the option of upgrading to a later version under, in this case, section 9 of GPLv2
<lsl88>hi! :) this is a quick message to say that I successfully installed guixSD on bare metal. My retroPC found and OS for her ;)
<buenouanq>awesome lsl88
<buenouanq>welcome to the rest of your improved life~
<swedebugia>I succeded in adding the blacklist function for recursive import of npm packages! Hooray!
<nckx>lsl88: Awesomeness.