IRC channel logs


back to list of logs

<noobly>is lvm support still absent?
***jonsger1 is now known as jonsger
<apteryx>rekado: 504 gateway time-out
<aaron____>I'm trying to build mysql which is version 5.7.21 which is supposed to be built against boost-1.59. It looks like databases.scm has code to use that version of boost. However when I do the build it uses boost-1.66 which causes the build to fail. Has anyone else seen this?
<hulten>When I first do a 'guix pull' as one user, and then 'guix pull --commit=...' as another using the commit from the first, could I instruct it to pull from my local machine?
<Laalf>hulten: afaik you cannot. but you should take a look at "invoking guix publish" in the manual
<Laalf>that might speed it up a bit
<hulten>I'll do that, Laalf, thanks!
<Laalf>does someone have some qt stuff packaged? qt5-styleplugins for example?
***rekado_ is now known as rekado
<rekado>apteryx: the timeout happens when Debbugs fails to respond.
<rekado>apteryx: I want to work around this by adding a long-term cache for every email fetched from Debbugs.
<rekado>over time we would not need to hit Debbugs each time we request to have bugs displayed.
<brendyyn>I'm getting errors about having no hard drive space despite having 80GiB free, however tune2fs says I have 10's of millions of inodes left.
<jlicht>hey guix!
<jlicht>I just learned discovered iwd (, which seems really cool. What would be the criteria on which to judge this to enable it in GuixSD-provided kernels?
<roptat>jlicht: "The page isn’t redirecting properly" :/
<roptat>ha my bad, I didn't accept cookies from the page
<jlicht>roptat: Sorry I probably already accepted them. I'll try to aim for sharing accessible links in the future :)
<roptat>I guess you'd have to make sure it's free software
<roptat>then, we can provide the necessary modules (maybe we already have them?)
<thorwil>hi! i'm looking into packaging autotalent. which is gpl2, exceot for a part with pure-data license
<thorwil>which is bsd-style ... and not listed on
<thorwil>to my reading, that combination seems acceptable
<thorwil>if it is, i guess i should add pure-data to guix/licenses.scm, or is there a norm to skip that for one-offs?
<Gamayun>iwd seems to be LGPL
<thorwil>hmm, grep be praised, i found this construct: (license (list l:lgpl2.0+ (l:fsf-free "file://COPYING")))))
<thorwil>wonder if "fsf-free" is the rigjht choice here, though
<thorwil>this should do: (license (list license:gpl2 (license:non-copyleft "file://COPYING-mayer_fft")))
<apteryx>rekado: makes sense. Thanks for explaining.
<civodul>snape: hey!
<snape>hey civodul :)
<civodul>snape: our Cuirass db on berlin has become huge again: 1.5G
<kmicu>Huge Data™ ;p
<kmicu>(Though 1.5GB sounds like a little DB.)
<civodul>depends on the viewpoint :-)
<quiliro>hello. is there a way i can make a mirror of all sources that guix refers to?
<civodul>hello quiliro!
<civodul>more or less, yes
<civodul>you could run "guix build --sources=all $(guix package -A |cut -f1)"
<civodul>you'll need quite a bit of disk space :-)
<efraim>'guix build --sources=transitive $(guix package -A | cut -f1,2 --output-delimiter=@)'
<efraim>Less than 100 Gb
*janneke is working towards the new bootstrap milestone after "reduced binary seed" we'll have a "scheme seed" bootstrap, with bootstrap-guile as "scheme seed" initially
<janneke>what are the chances of getting a /bin/guile (or /bin/scheme) some time soon?
<janneke>the conventional /bin/sh is nice, but during bootstrap that won't be available initially
<civodul>janneke: there's no /bin in the build environment, so no problem? :-)
<civodul>hmm looks like EMMS 5.1 sometimes loses track of its sub-processes and i end up with two players running at the same time
<civodul>anyone else experienced this?
<snape>civodul: I don't know what can be done about the db size
<snape>except rm -rf /var/lib/cuirass and restart ;-)
<efraim>download failed "" 503 "Backend is unhealthy"
<jlicht>civodul: I struggled with this as well
<jlicht>I was never able to figure out what went wrong
<jlicht>How does one keep the code reference part of the texi manual in sync with the source code?
<efraim>keeping my ppc g4 up and running isn't easy :/
<yrk>civodul: which backend?
***hugh_marera_ is now known as hugh_marera
<Laalf>efraim: why?
<Laalf>do you have guixsd on it?
<efraim>debian sid, something keeps on messing with the video driver
<Laalf>i should get my powerbook out
<Laalf>i loved the keyboard
<efraim>for 32 bit ppc, the options are gentoo, debian sid, adelle (sp?) and the BSDs
<Laalf>or macos. still unix. can do a lot of stuff with that
<Laalf>i need to replace the thermal paste before doing anything with it. its been laying here since 2012 or so. untouched. never had the paste replaced
<efraim>10.4 is unmaintained, so there is that problem
<Laalf>i have a 1,67ghz one
<Laalf>so 10.5 is easily doable. but unmaintained as well. thats 10 years old now
<Laalf>i wanna say i am old but i am really not
<janneke>civodul: hehe, yes that's right
<janneke>so i can prolly just use #! /bin/guile and watch patch-source-shebangs to fix it for me
<luciddreamzzz>hydras up?
<luciddreamzzz>not bad
<civodul>janneke: yup!
<civodul>janneke: i like the idea of bootstrap Scheme!
<civodul>janneke: BTW, i still have to upload the binary seeds, right?
<janneke>civodul: good! i was a bit hesitant, but rutger convinced me to go bootstrap-scheme (=Guile) first, then migrate towards bootstrap-scheme (=mes OR bootstrappable Guile)
<janneke>civodul: yes, please do!
<janneke>we're going to have core-updates merged rsn i figure?
***luciddreamzzz is now known as luciddreamz
<civodul>RSN, for sure!
<civodul>we've never been this close
<aaron____>Hi, I'm trying to install mysql-5.7.24 which requires boost-1.59. I installed boost-1.59:
<aaron____>boost 1.59 out /gnu/store/978nayjys3wdhsc5zg9yravp2yf34r3c-boost-1.59
<aaron____>Then I tried to install mysql using that version of boost but get a package not found error:
<aaron____>guix package -i mysql --with-input=boost=boost@1.59
<aaron____>guix package: error: boost: package not found for version 1.59
<aaron____>What am I doing wrong?
*janneke goes afk for a bit, bbs
<efraim>ok, forget fixing f2fs-tools@1.10.0 on core-updates, 1.11.0 builds just fine
<civodul>aaron____: you're doing nothing wrong!
<civodul>current Guix provides 1.66 only
<civodul>you can see that by running "guix package -A boost" or at
<aaron____>I installed boost 1.59 from a tar.gz I downloaded from boost. What is the proper way to get guix to recognize that version so I can use it with --with-input?
<luciddreamz>wow and mirror up
<Laalf>aaron____: you should build your own package with inherit
<civodul>aaron____: --with-input only accepts package specs for currently defined packages; what packages are installed on your system doesn't matter
<civodul>what you could do is do that from an older Guix that has boost 1.59
<civodul>so you would 'guix pull --commit=X', where X is the commit that contains boost 1.59
<efraim>iirc core-updates re-introduces boost@1.59
<efraim>commit 7cbf06d8c2935abfc6c688cf3f9b99e0e5393960 on core-updates brings it back for mysql
<aaron____>Thanks for the help. I'll try that.
<efraim>civodul: I have a follow-up commit to my lbzip2 and sharutils commits on core-updates
<efraim>I'm getting the wrong hash on `guix build -S linkchecker'
<thorwil>i managed to get /gnu/store/68pnvpin/gnu/store/68pnvpinwj91vhihc14llln8nf6hg0lb-autotalent-0.2/lib/ladspa/
<thorwil>but ardour doesn't find that plugin
<davidl>When you install a manifest, is there an option like --don't-build-if-no-substitute-available ?
<thorwil>i added ladspa to my profile to have `listplugins`. with no args, it prints nothing except "Warning: You do not have a LADSPA_PATH environment variable set."
<thorwil>so how are ladspa (and lv2) plugins supposed to found on guix?
<nckx>davidl: No.
<davidl>nckx: ok. Are there any plans for something like an all-substitutes-available branch of guix?
<Laalf>shouldnt that be master in theory?
<rekado>thorwil: we set the environment variables.
<davidl>Im just having a few laptops that I wish I could run guix on but Im unable to keep them updated because of the builds.
<Laalf>davidl: invoking guix offload
<rekado>davidl: policies like that can be implemented with “guix pull” and channels.
<rekado>davidl: it’s a bit rough, but civodul demonstrated it on the mailing list.
<davidl>rekado: ok cool. I'll see if I can find it.
<thorwil>rekado: so LADSPA_PATH will be a concatenation of all the store addresses?
<davidl>Laalf: not sure what you mean. I think I need to read about guix offload first.
<thorwil>in any case the question remains: what do i have to do that the autotalent plugin can be found?
<rekado>thorwil: no, you can just set it to $HOME/.guix-profile/lib/ladspa (?)
<rekado>that’s a union directory
<thorwil>ok, listplugins output looks good now
*thorwil sends poylphonic material through autotalent and grins ear to ear
<pkill9>what's autotalent?
<pkill9>nevermind, i searched it
<thorwil>pkill9: ever heard of autotune? autotalent is a simple version of that
<thorwil>heh, already taped it
<thorwil>typed, even
<pkill9>ah ok
<thorwil>i will add LADSPA_PATH to .profile, then
<quiliro>civodul, efraim: thank you. will try efraim's solution once Guix is installed.
<Laalf>pkill9: what magic do you have to do again to get whiskermenu to get detected?
<pkill9>Laalf: im not sure, I just install it normally i think
<pkill9>i need to submit it as a package sometime though
<pkill9>Laalf: this is the package definition i use for it
<Laalf>pkill9: doesnt work for me. now integrated into the config for extra autism and unneeded hackery
<Laalf>but you should probably submit your packges, yes;)
***slyfox_ is now known as slyfox
<pkill9>i haven't built it in awhile actually
<pkill9>oh hmm it's already built on my guix commit
<civodul>rekado: after 'guix gc' (with no arguments), we have only 1.4G left (probably largely due to the new GC root registration in Cuirass, but still)
<Laalf>pkill9: i know that it doesnt theme.
<pkill9>Laalf: what doesn't work with it?
<pkill9>does it fail to build?
<pkill9>or just doesn't theme
<Laalf>pkill9: its fine, but doesnt theme
<Laalf>xfce-dusk doesnt get applied
<pkill9>do you have a bug where if you drag a window on the bar, the whiskermenu icon gets filled witha black box?
<pkill9>well, the whole button, not just the icon
<Laalf>cannot reproduce
<pkill9>Laalf: have you tried restarting the panel when you change the theme?
<pkill9>oh interesting
<pkill9>wonder why it happens fr me
<pkill9>might just need to update or something
<Laalf>ill reboot for you
<pkill9>Laalf: you don't need to reobot
<Laalf>its simpler
<Laalf>and it doesnt work
<pkill9>just run `pkill -USR1 xfce4-panel` to restart the panel
<Laalf>doesnt theme
<pkill9>well atleast it's functional
<pkill9>is down?
<Laalf>502 Bad Gateway
<Laalf>do you need anything build?
<pkill9>is the guix server down or just the web portion?
<pkill9>just guix itself
<pkill9>but i have code in channels.scm that checks for latest built evaluation of guix n berlin and pulls the commit for that
<Laalf>can i delete old system configurations yet?
<rekado>cuirass is back up
<rekado>civodul: just saw your mail. Should I leave cuirass stopped?
<rekado>we have 13GB left right now.
<rekado>I’ll try to figure out why we can’t boot with the external storage.
<Laalf>pkill9: xfce4-popup-whiskermenu doesnt work
<Laalf>"no such file or directory". isnt that the message you get when you execute a binary from another system?
<Laalf>wait... it calls xfce4-panel?
<Laalf>xfce4-panel is in the input. shouldnt it theoretically be able to call xfce4-panel then?
<pkill9>oh yeah it didn't work for me either, i temporarily fixed it somehow though
<Laalf>that is not a binary
<pkill9>oh yeah, that's a wrapper script with incorrectly written gnu store paths, just use the command at the bottom of the wrapper without the store path
<pkill9>xfce4-panel --plugin-event=whiskermenu:popup:bool:$ATPOINTER
<pkill9>where $ATPOINTER is 'true' or 'false' depending on if you want it to appear at your mouse pointer
<Laalf>great... why doesnt it theme though?
<pkill9>i dunno
<pkill9>i'll see if there's any issues about that upstream
<pkill9>doesn't look like it
<pkill9>it might be missing an input
<civodul>rekado: yeah i removed old generations here and there, but 13G is really the maximum we can get
<civodul>rekado: we'd consume it in ~10 hours or less
<rekado>civodul: I just removed a few old system generations
<civodul>there aren't that many left now ;-)
<rekado>I need to figure out why we can’t boot when /gnu is on the external storage.
<civodul>we can't?
<rekado>I reconfigured berlin on Friday and rebooted
<Laalf>shouldnt everything the kernel needs be on /gnu?
<rekado>right after GRUB it complained about some paths with were /store/foo instead of /gnu/store/foo
<Laalf>initial ramdisk on /boot if i understood that right, but modules on /gnu?
<rekado>so I changed them in the GRUB console
<rekado>then the next problem was that it wanted to mount the external partition, but said that a partition with this UUID didn’t exist.
<rekado>looks like it tries too early to mount it.
<rekado>we need to boot the kernel first
<rekado>(19GB free on berlin now)
<civodul>uh so now GRUB can see the device?
<rekado>it cannot
<civodul>so how can it complain?
<rekado>it tries to mount the partition, because that’s what the system configuration tells it to do.
<civodul>GRUB or Linux?
<rekado>in berlin.scm I added a file-system declaration
<civodul>ah ok so that's the initrd code trying to mount that partition, right?
<rekado>(device (uuid "827fabbc-c12b-43f9-bd77-…") (mount-point "/gnu"))
<rekado>yes, I think so.
<civodul>yes, definitely
<rekado>I left off “needed-for-boot?” because I know that’s not working
<civodul>but so before that you copied the kernel and initrd elsewhere where GRUB can access it, right?
<rekado>yes, the store on the local disk is the same as the store on the external storage.
<civodul>ah ok
<rekado>I reconfigured and then synced the store.
<civodul>makes sense
<Laalf>pkill9: the pulseaudio plugin doesnt theme properly either
<civodul>rekado: and the store on the external storage device is at /store, not /gnu/store, right?
<rekado>that’s confusing. I mount the ext storage at /mnt; there is a directory /mnt/store.
<rekado>I think that’s how it should be.
<civodul>oh you don't mount it at /gnu?
<rekado>while syncing I mount it at /mnt
<rekado>the system configuration says it should be mounted at /gnu, obviously
<rekado>but this never happens.
*civodul looks at the diff on berlin
<civodul>this never happens because of the UUID, right?
<civodul>note that you need the file system UUID, not the GPT partition UUID
<civodul>(normally 'reconfigure' checks this)
<rekado>yes, it doesn’t happen because we don’t even get to boot linux. The initrd tries to find the partition by UUID and fails.
<rekado>I got the file system UUID with blkid /dev/sdb1
<civodul>"mount UUID=a6455b66-59d2-40bd-bddb-0c572bb62a2f /mnt" works
<civodul>so this is the right FS UUID
<civodul>(currently commented out in the config)
<rekado>(why is there a UUID and PARTUUID in the blkid output?)
<rekado>I can try to reconfigure with that UUID, sync the store again, and then reboot.
<civodul>PARTUUID is a GPT thing and not what we want here
<civodul>didn't reconfigure warn about the UUID?
<civodul>oh maybe it warns only for the root fs
<rekado>guix system: warning: only 2.1% of free space available on /gnu/store
<rekado>hint: Consider running `guix gc' to free space.
<civodul>it looks friendly and all, but sometimes you'd like to yell at it ;-)
<Laalf>pkill9: do you know why i cant suspend on xfce? it works on bspwm and in slim
<pkill9>i dunno, what happens when you try to suspend it?
<Laalf>sorry, when i close the lid
<Laalf>it suspends and then comes back instantly
<pkill9>that happens to me often
<pkill9>i might switch from xfce tbh
<Laalf>xfce can handle this itself
<Laalf>maybe if i disable that stuff in elogind it will work
<pkill9>Laalf: are you running the linux-libre kernel, or have you changed it?
<pkill9>(just to rule that out)
<Laalf>pkill9: linux-libre
<pkill9>but yeah i often find my laptop wakes up immediately after i close the lid
<pkill9>so maybe it's xfce
<Laalf>it handles this stuff on its own
<pkill9>oh i see
<Laalf>so maybe when elogind tries to suspend it wakes up because linux