IRC channel logs


back to list of logs

<ng0>well i used to use some hybrid version of mutt until I got stuck with mutt and moved on to emacs because of 'vanilla' mutt. before that I used vim
<reepca>mark_weaver: I applied that patch and used ./pre-inst-env to reconfigure the system and now neither the ethernet interface nor the wlan interface are showing up in ifconfig.
<reepca>/var/log/messages now says "b43 ssb0:0: Direct firmware load for b43/ucode5.fw failed with error -2" twice before it says the same thing about b43-open/pcm5.fw. The /*DEBLOBBED*/ thing no longer shows up.
<reepca>bavier: are all the necessary firmware files included in that package?
<bavier>reepca: should be
<mark_weaver>reepca: now it's looking in b43 instead of b43-open?
<mark_weaver>(for ucode5.fw) ?
<reepca>but it's still looking in b43-open for pcm5.fw
<reepca>I'll boot from the old system configuration and then upload my /var/log/messages and it'll be the second boot from last
<reepca>... but first I need to hurry to eat supper with my brother
<mark_weaver>best to report this to #linux-libre
<mark_weaver>okay, enjoy!
<mark_weaver>bavier: pcm5.fw is not in the firmware package, and afaict, it's not free source, nor is the source even available to us.
<mark_weaver>but apparently it has been usable in the past without pcm5.fw
<bavier>mark_weaver: yeah, I see that
<mark_weaver>but this is a discussion for #linux-libre. I doubt this is a Guix-specific issue.
<mark_weaver>bavier: what's your relation to this process? do you have a b43 device?
<mark_weaver>(just curious)
<bavier>mark_weaver: nope :). I just jumped in and packaged the firmware quick
<bavier>at one point another newcomer, quilirio IIRC, had asked if guix had it available
<mark_weaver>I guess we need a champion to help debug this
<bavier>that would be nice, yes
<bavier>ACTION needs to catch bus
<Digit>so, i just issued a package install command (still running) n then realised i didnt want it, nor the hundreds of packages it's pulling in. how do i undo that?
<Digit>guix package -m --roll-back ?
<Digit>no m. ok. idk if that worked. more time making mistakes necessary to gain familiarity with guix. ^_^
<iyzsong>Digit: I'll just ignore it.. if you want, run 'guix gc' will remove unused items in store (include the ones you just download but not install).
<iyzsong>ACTION out for meal
<Digit>thnx, that's helpful too. answers the question i didnt ask, but would have, had i known to. :)
<reepca>So bavier, it looks like either something's messed up with the package and we're missing firmware (currently only have b0g0bsinitvals5.fw, b0g0initvals5.fw, and ucode5.fw) or something bizarre is happening with linux 4.8.1.
<reepca>ah man, is there any way I could leave a message for him?
<janneke>reepca: yes, sneek is back
<reepca>I assume that's a bot? What's the syntax?
<janneke>reepca: yes
<janneke>sneek: later tell reepca, thanks!
<sneek>Will do.
<sneek>Welcome back reepca, you have 1 message.
<sneek>reepca, janneke says: thanks!
<janneke>yes, it's good to have them back
<reepca>sneek: later tell bavier, it looks like either something's messed up with the package and we're missing firmware (currently only have b0g0bsinitvals5.fw, b0g0intivals5.fw, and ucode5.fw) or something bizarre is happening with linux 4.8.1. I'm gonna try manually building it and see if I get the same files as output
<sneek>Got it.
<janneke>sneek: botsnack!
<jmd>What is an "activation-service-type" ?
***fps_ is now known as fps
<rekado>I need the freepascal compiler fpc for a game I’m packaging. fpc is self-hosted, so it needs fpc :(
<rekado>I wonder if GNU Pascal could be used for bootstrapping.
<janneke>ACTION has unrealisticly high hopes that Mes will add some awareness to all the bootstrapping mess
<bill-auger>i just installed guixsd in a qemu VM which took almost 24 hours - is that normal?
<bill-auger>i ask mostly because it seems to have failed and im not sure if it is recoverable
<janneke>bill-auger: that seems awfully slow, did you enable kvm in qemu?
<rekado>gah, this is too much work. GNU Pascal didn’t have a release since 2002 and to build it I probably need to build an old GCC first. And only then I’d be able to try if I could use GNU Pascal to bootstrap Freepascal :(
<rekado>Freepascal is written in Pascal and used to be built with Turbo Pascal until it became self-hosting.
<random-nick>hello, I am having trouble building guix on another distro (arch)
<random-nick>at GUILEC guix/nar.go I get an error from boot-9 "no such language tree-il"
<random-nick>I have guile 2.0.13 installed on my system
<bill-auger>janneke: sry - yes i have kvm enabled but i set my cpu governor to 'powersave' because compiling in qemu always makes my CPU overheat
<janneke>hmm...can't say i actually did full installs in a vm
<bill-auger>i did not realize guixsd was a source ditro - i was just wondering if i did something wrong to force all these packages to build
<janneke>i do produce full vm images now and then, and that may take minutes
<janneke>a full `guix system init /mnt' on non-virtualize iron typically also takes a couple of minutes -- (apart from any build time)
<janneke>hmm, bill-auger: did you enable binary substitutes?
<bill-auger>i dont think so -
<bill-auger>herd start cow-store /mnt then copied the desktop config to /mnt/etc/config.scm then guix system init /mnt/etc/config.scm /mnt
<bill-auger>but it did print a whole bunch of lines that mentioned substitutes at the start
<bill-auger>so this was not typical behavior - maybe i should start again fresh?
<bill-auger>so is there a prebuild qemu image i could try?
<bill-auger>im willing to try again because it is quite possible i messed it up myself - but is useful to ask if i should expect it to take more than one hour
<bill-auger>i would consider installing on hardware except that i dont have a USB stick large enough - and that seems not to be the only way to install (not possible from a CD?)
<bill-auger>a CD burnable ISO image would be ideal - soes anyone know a way to transform the installed img to a CD ISO ?
<jmd>How do I specify a configuration parameter to service-extension?
<jmd>(Yes, I have read the manual.)
<jmd>(Yes, I have looked at the source.)
<jmd>(Yes, I have looked at the examples which seem most relevant.)
<quigonjinn>kicad, which i try to package, when using the opengl canvas, is working on a guixsd system correctly(though it's "laggy", while non-opengl canvases are fine). Using some functionality in opengl mode, crashes my non-guixsd system (it does not crash a guixsd system). any insight on how to proceed with debugging that?
<ng0>so wmalossless (.wma) is a format we can not build in ffmpeg? I just had a lot of glitching in a very old playlist, pointing out that this format is not completely supported
<mekeor>where does guix get its packages from? is there an official guix server? could we combine guix with gnunet so that everybody could share packages in a decentral, peer-to-peer, secure, possibly anonymous way?
<catonano>mekeor: there's a server. "Official" is probably a too big word. Yes, Guix could use Gnunet, it's in the roadmap and there are some roadblocks, so for now there's no figure about when
<catonano>specifically, ng0 is working a lot on bringing Gnunet in Guix and he declared he has plans to move Guix to some Gnunet based service
<catonano>Also, ludo expressed concern about some issues with Gnunet.
<catonano>it's quite foggy, right now
<catonano>the best news is that the next version could be available in Guix almost as soon as it gets released. Because ng0 has a working version of the packaging
<catonano>I hope you appreciate my synthesis
<mekeor>wow, that's super amazing and exciting! this is revolution happening here, seriously.
<ng0>what's so difficult about singular they
<ng0>but, yes
<ng0>I predict that moving ALL of guix to gnunet-fs will be a long research thing, but we can get there.. specifically because I decided that I need to pay attention to myself and not burn myself out on this. so autumn 2017 was.. optimistic... but I get work done
<ng0>one concern i have is that gnunet-fs, at its current size and speed does not make file available immediatelly for the other nodes.. but it will take certainly away some pressure from a purely 1 / many servers system, where when the servers harddrives go boom you loose all build binaries
<ng0>but you still need nodes which are onoine 24/7
<ng0>or something
<ng0>I have lots of notes on that, not publkic though
<ng0>got frustrated with server maintenance and canceled them for now
<ng0>mekeor: Nix is doing a similar thing with ipfs
<ng0>got the link in some git.. if you search their issues you can find a 9 months and still going discussion
<lfam>Twolame (an FFMPEG dependency) fails on core-updates, so I asked upstream for advice
<lfam>Any Perl people have advice about that test failure?
<lfam>The Talos Secure Workstation (POWER8 hardware) is fundraising now:
<mekeor>very very interesting
<mekeor>is there anybody here who lives in Munich, Germany?
<lfam>This seems like a promising example of how to workaround that twolame issue:
<catonano>ng0: sorry about the pronoun. I'll pay attention in the future.
<jmd>catonano: What were ludo's concerns?
<catonano>ludo's concern were that the release rythm of Gnunet is not adequate for guix being based on it. Ng0 reported those concerns (I don't remember where I read that, in some chat log I think).
<ng0>we talked about it offlist.. and if I stay at the system-integration level, I should try to address automated/semi-automated releases at some point. the way releases are done right now does not work in favor for regular releases
<ng0>at least that's what cg told me
<ng0>I have to dig out the conversation again
<ng0>but it would help
<ng0>I share the concerns
<lfam>Interesting read:
<bill-auger>i would like to attempt install guixsd again - could someone plz confirm these assumptions for me?
<bill-auger>it seems the only way to install guixsd onto hardware is with a USB stick (not possible from a CD?) and noone knows a way to
<bill-auger>to transform the installer into to a CD ISO?
<bill-auger>there no prebuilt qemu image i could try?
<bill-auger>if i start again fresh installing into a qemu VM then i should expect install to take not more than an hour on any machine?
<lfam>bill-auger: I haven't looked, but have you searched for information about how to convert a raw disk image into an ISO?
<bill-auger>if the install mentions substitutes or takes more than one hour then i have done something wrong
<lfam>I don't know what makes an ISO but it seems that others will have looked into this question before
<bill-auger>no i have not tried to do convert it yet
<lfam>The time your install will take depends on a what packages your new system wants to install, the speed of your hardware (disk, CPU) and network.
<ng0>lfam: afaiu civodul, there is not much work done to make an iso
<lfam>You can pass '--dry-run' to `guix system init` to see what it will build and what it will download. If you are unsure about the meaning of the dry-run, share it here and we can help you decide what to do
<ng0>left, not done
<bill-auger>ok but it took almost 24 hours yesterday only to end up failing - that is probably excessive
<lfam>That's usually not normal but, like I said, if your hardware or network is very slow, it could be possible. Especially if you are using a VM without KVM acceleration
<bill-auger>so i do not wish to repeat that mistake - and i do not know what the mistake was
<bill-auger>i have kvm yes - maybe i should try a leaner install instead of the "desktop" config?
<lfam>I recommend using a bare-bones system for the system initializiation. The initializiation is sadly still a brittle thing. Once it is completed, the system is solid and you can reconfigure it into a more full-featured desktop easily and safely
<lfam>Make sure you pass -enable-kvm to your QEMU invocation
<lfam>If you are using QEMU on a spinning disk, try not to use the disk outside of the VM or you will get terrible performance
<lfam>I always recommend against initializing with a desktop. If you are at all comfortable on the console, use the bare-bones configuration example and then reconfigure into the desktop
<bill-auger>ok yea thats what i was thinking - i was following the instructions in the docs which say to install the desktop initially
<lfam>They don't say to install the desktop. They say how to install the desktop :)
<lfam>I think we should reword that sentence.
<lfam>It leads new users astray quite often
<lfam>If `guix system init` fails partway through (for example, due to a network failure), you can re-run the command and it will re-use the work that it's already done
<lfam>Just FYI
<bill-auger>ok ty - i will try the bare-bones install first
<lfam>Okay. Try the --dry-run first. It should report that you will mostly be downloaded pre-built binary substitutes. There will be some things to build from source, but not the majority.
<bill-auger>oh yes that sound good
<lfam>I have to go now. Good luck!
<Dawgmatix>Have a couple of questions about guix
<Dawgmatix>a. any pointers on how to make my own libraries / executable available as packages?
<Dawgmatix>b. i would like to apply some local patches to python. any pointers on how to apply the patch and have the resulting built package be installed?
<jmd>a. If you are new to guix, the best way is to copy some existing program and use that as a template.
<jmd>b. There is an "inherit" procedure which allows one to make packages based upon some existing other package.
<jmd>If I were to accidently delete /etc/config.scm is there any way to recover it?
<jmd>In the guix manual all instances of predicates (predicate?) are misrendered.
<jmd>Have we done something wrong or is this a texinfo bug?
<efraim_>which browser are you seeing the error with?
<jmd>both with standalone info and emacs
<Dawgmatix>just tried to install guixsd on qemu and get a grub failure (grub-install: error : will not proceed with blocklists)
<Dawgmatix>Any pointers?
<jmd>Is there any policy in which module a service belongs?
<civodul>jmd: nope!
<sneek>Welcome back civodul, you have 1 message.
<sneek>civodul, ng0 says: femtolisp source I can identify: FreeBSD, OpenBSD, OSX, win32. I also found that some files are placed in the public domain so i have to adjust the license field. I can't identify more.
<jmd>Good. I can do what I like then!
<ng0>sneek: you are verrrry slow
<ng0>sneek: no botsnack for sloths
<efraim>our mutt uses urlview to view urls in emails but apparently no longer bundles it and/or is dead
<civodul>oh sneek is back
<efraim>another option is urlscan, which is written in python, and would increase mutt's size from 165 to 226 ... i think
<efraim>the top line of mutt stays about the same, the total at the bottom changes from 165 to 226
<ng0>neomutt is also an option (though doesn't solve this), but I wonder where that disapeared to after I published my supposedly buggy version
<civodul>30% of mutt's closure size is perl...
<efraim>i don't want to add python to the closure just to get this feature though
<jmd>let's write a guile replacement.
<efraim>thats starting to be my go-to answer for a lot of problems
<ng0>there's an app for that... eh I mean there's a guile for that :)
<ng0>washing machine is broken.. let's write a guile replacement :P
<ng0>cat waking me up at 4 AM is showing weird sideeffects. I'll be afk now
<Dawgmatix>While installing is there some reason why the stage with messages like "copying /gnu/store/*" is so slow? Any pointers on how to speed it up?
<jmd>guix has it's own passwd command?
<ng0>it's just passwd
<reepca`>Dawgmatix: I noticed something similar, the copying process (total amount copied ended up being under 2GB IIRC) took a long time, but I just attributed it to it being to a crappy flash drive (someone else suggested the write patterns might just be really weird for the flash drive).
***reepca` is now known as reepca
<lfam>Wow! Apparently the reporter of that c-ares bug that I linked to earlier (fixed in our tree) will receive 100k USD from Google as a bug bounty. Now there is some motivation!
<lnbb>hey, I'm trying to install guix sd on a gigabyte ga-g41m-es2l running libreboot. Grub recognizes the usb stick and shows the menu entry "GNU with Linux-Libre 4.5 (beta)"
<lnbb>linux seems to boot, but the computer resets pretty soon. Even using a serial cable I can't see any error message.
<lnbb>the last line displayed is "[ 0.208002] .... node #0, CPUs: #1"
<lnbb>any help is appreciated
<lfam>lnbb: I wish I had an idea :) Have you tried asking on #libreboot?
<lnbb>lfam: not yet
<lfam>I'm not saying it's not our bug, but maybe somebody there has seen this issue and will have some advice
<lfam>Can you share the GuixSD OS configuration file that you used?
<lnbb>trisquel gnu/linux boots without issues from usb and I already got that installed
<lfam> is a good site to share it on, if you want to share it
<lnbb>I just used the usb stick image for 0.11.0
<lnbb>trisquel uses isolinux, maybe grub chaining is also a problem
<lfam>Oh, this happens when you boot the installer? I had thought you had already installed
<lnbb>no, it happens when booting from usb, yes
<lfam>Anything else interesting before that line?
<lnbb>I'll upload the log from serial out
<lnbb>serial log now uploaded here:
<lnbb>sorry, I didn't clean it up, it also contains the whole libreboot output
<lnbb>guix sd installer boot starts around line 1276
<lfam>Well, nothing jumps out at me, but I'm no expert. Hopefully somebody else spots the problem, or has a suggestion for debugging
<lnbb>that would be great. It could also be a kernel bug or incompatibility, as the guix sd kernel is way newer than the kernel used by trisquel
<lfam>True, could be
<lfam>I'd ask on #libreboot too, just to reach more people :)
<lfam>If you don't figure it out, please email
<ng0>there's something about the videomode noth being found
<ng0>wrong line
<Dawgmatix>reepca - i am in fact installing in kvm from a disk image and the machine i am running on has a pretty fast ssd, so we can rule usb drives out
<efraim>is libreboot up to date?
<lfam>go-1.4.3 is failing on all architectures on core-updates :/
<reepca>So I'm not sure how to go about trying to get the wireless working, now that I know to use ifconfig -a to check for its presence as an interface I've tried the suggested process in the manual, I'm not too sure how to get the wpa_supplicant.conf right though, there isn't any network security as far as I'm aware, so far I just have the ssid entered.
<lfam>Failing Go: <>
<lfam>"cannot load DWARF output from $WORK/net/_obj//_cgo_.o: decoding dwarf section info at offset 0x4: unsupported version 0"
<lfam>Does anyone have any idea what's going on there?
<jmd>efraim: Libreboot is of course somewhat uncertain right now.
<efraim>i'd check the go source to see if there's a clue there, I saw that and closed the tab
<Dawgmatix>Any pointers to my question about grub earlier? I cannot get grub to install
<lfam>efraim: I guess I'll email Matthew about it
<Dawgmatix>Should I be creating a seperate partition for grub?
<Dawgmatix>So far I tried both /dev/sda and /dev/sda1 in config.scm
<Dawgmatix>and for each of them the install failed
<efraim>iirc it should be /dev/sda
<lfam>Dawgmatix: I missed your earlier questions. Did you share your OS configuration and the error messages?
<jmd>lfam: I think he's installing in QEMU
<lfam>Hm, that should work. I've done it before
<jmd>Does anyone know how to unlock the screen in slock ?
<Dawgmatix>Installing on qemu with kvm enabled. Error message was along the lines of (grub-install:error: will not proceed with bloclists)
<lfam>jmd: Type your user's password and press <ENTER>
<lfam>Or visit the slock bug tracker and find a crashing bug ;)
<lfam>Dawgmatix: Did you find anything useful when searching the net for that error message?
<jmd>lfam: I tried the first idea.
<lfam>jmd: Huh, that really should work! I use slock
<lfam>You can also switch TTYs and kill slock
<jmd>Yeah that's what I did in the end.
<Dawgmatix>lfam I found some place which suggested that instead of specifying /dev/sda1 (the partition) I should specify /dev/sda (the drive)
<Dawgmatix>I tried that, but it didnt resolve the issue
<reepca>and the bootloader field of your config.scm is (bootloader (grub-configuration (device "/dev/sda")))?
<Dawgmatix>fwiw i created the partitions using cfdisk
<Dawgmatix>and the disk is using the gpt scheme
<lfam>Dawgmatix: What does `fdisk -l /dev/sda` tell you?
<lfam>I believe that '/dev/sda1' should start at 2048
<Dawgmatix>Yes it does start at 2048
<lfam>I would ask on #grub, since none of us seem to know how to interpret that error message
<lfam>Hm, I can't build the latest libgit2 on master
<efraim>i have it updated locally
<lfam>Oh, then please push :)
<efraim>that was from before I subscribed
<lfam>I'll check if those fixes are in 0.24.2
<lfam>I was trying to build 0.24.3
<lfam>Oh, no I wasn't. There is no 0.24.3 yet :p
<lfam>Sorry for the confusion!
<efraim>i saw there are a couple more qemu CVE patches incoming
<lfam>QEMU is probably the most common package on oss-sec. I guess it's big, complex, C, and popular
<lfam>After QEMU, Linux
<lfam>efraim: libgit2 0.24.2 doesn't build on master :(
<lfam>Same issue I was having
<efraim>hmm, it built for me :/
<lfam>Really? Interesting...
<lfam>I'm on x86_64
<efraim>parallel-build? #f maybe?
<lfam>ACTION tries it
<efraim>it says succeeding on hydra
<efraim>for 0.24.1
<lfam>It actually fails in the tests, so I'm trying with parallel-tests? #f
<lfam>Once it builds, I'll try to cherry-pick this:
<efraim>well, i was actually about to go to bed, leaving my computer to spend the next 6+ hours building qt-5.7.0
<lfam>I'll work on it, don't worry :)
<lfam>efraim: Can you revert the update in the meantime?
<efraim>for libgit?
<lfam>Actually, no
<lfam>efraim ^
<lfam>If it works for you then we can leave it and see what Hydra does
<efraim>ok, sounds good
<lfam>This is the problem I'm seeing:
<lfam>So I'll try cherry-picking that bug fix and the object parsing hardening patches
<ng0>I think my vim-full, I forgot to actually add the patch files
<lfam>Heh, that happens to me too
<lfam>`git add -u` won't catch new patches :)
<ng0>untracked.. so i guess so :D
<efraim>lfam: are you building on tmpfs?
<lfam>efraim: No, on btrfs
<jmd>No wonder it doesn't work. slock is not linked against !!
<efraim>my /tmp is on ext4
<lfam>Okay, I have to go offline for a little while. I'll be back to work on libgit2
***jackhill_ is now known as jackhill
<efraim>i'm going to assume the freebsd failure is on zfs
<lnbb>quick update on my libreboot/guix sd problem: Parabola GNU/Linux boots fine (isolinux), they used Linux 4.6.4
<efraim>maybe its a COW filesystem thing
<lnbb>can I somehow try to boot the installer using a different kernel?
<ng0>ACTION tries lolcode interpreter again
<jmd>Should I check-in gxi into a wip-branch or wait until it's (more) mature?
<efraim>qt-5.7.0 failed, might dig into it later
<efraim>wait, no it didn't
<efraim>that was the 12 minute configure phase
<ng0>I wonder if I spaced out during first time building lci succesfully
<ng0>now if fails
<ng0>hm... I think I contributed the correct version to the list
<lnbb>lfam: just tried booting the installer using the parabola gnu/linux kernel - that seems to "work", although I get a guile prompt/backtrace, but no reboot
<lnbb>I'll post on the mailing list
<ng0>efraim: thanks for spotting the mistakes in vim-full, I'll send updates tomorrow in the new thread.