IRC channel logs

2019-04-27.log

back to list of logs

<roptat>meiyopeng, if you can read me, could you translate "version.texi" in the manual to "version-zh_CN.texi" and send it to the TP please?
<roptat>civodul, I pushed a small fix, it should be fine now
<roptat>good night guix :)
<civodul>night roptat!
<civodul>meiyopeng: also the changes from commit 3068509ac12df6f2fc6e500019a4d8bbf75d32a2, in addition to what roptat wrote above ↑
<rekado>Hmm, Bioconductor 3.9 will be released on April 29. This will be a day of mass updates.
<civodul>hey rekado!
<civodul>i suppose we could update a day or two later?
<vagrantc>this whole concept of a rolling release with releases is interesting, at least :)
<rekado>is 1.0 planned for the 30th?
<civodul>yes, or May 1st
<civodul>(or 2nd)
<civodul>vagrantc: true!
<civodul>vagrantc: so how 'bout a Debian package for 1.0? :-)
<vagrantc>i keep meaning to get to it... and mes, and ...
<vagrantc>1.0 is just a reduced seed, or is it down to that minimal hex?
<vagrantc>er, guix 1.0 will be just ...
<civodul>1.0 will still have the biiiig seed
<vagrantc>ah.
<vagrantc>both easier and harder to package
<civodul>yeah
*civodul -> zZz
<civodul>good night/day!
<Blackbeard[m]>ison: I don't think copyright is trivial
<nckx>rvgn: Oh dear, now you've got 3 identical replies to your Transmission question ;-) Seems like replies aren't being deliverd to everyone… or it's just a coincidence.
<swedebugia>Good morning!
<swedebugia>Today I am gonna try again working on the quicklisp importer.
<buenouanq>installing guixsd on an old machine and I'm getting something I've never seen before about initrd modules and sata-sil
<buenouanq>the suggested modules seems to have worked
<buenouanq>I've just never had to have an initrd-modules section before.
<reepca>Hm, I'm getting a match-error when 'guix system reconfigure' attempts to activate user-homes, looks like this: https://paste.debian.net/1080151/. For the record, I don't explicitly specify the root account in my config. I wish these error messages would give some information about where the error is, or at least which file it's in.
<nly>It's incredible, Elogind broke after a power failure. What should i do?
<brendyyn>40 minutes and no message back from guix-patches using my new email address. hmm
<nly>you might have to subscribe to the mailing list
<nly>you already know this i think but issues.guix.info has a web interface for checking patches sent to guix-patches
<brendyyn>hadn't seen that webpage before
<brendyyn>Recent activity
<brendyyn>504 Gateway Time-out
<brendyyn>nginx
<reepca>anyone else had issues with user-homes activation recently?
<nly>The web ui is working no? http://issues.guix.info/issue/35423
<brendyyn>yep
<brendyyn>Just the Recent Activity section doesn't work
<brendyyn>I don't even know where to subscribe. there is no link in the archive. i recall there was some mailman page
<nly>reepca my machine always warns about user-homes activation failing after guix system reconfigure, but i faced no issue
<reepca>nly: with the red text and error message and everything?
<nly>no red text, but idk, i only use eshell
<reepca>actually I see you joined right after my previous message - here's what I'm seeing, for reference https://paste.debian.net/1080151/
<nly>brendyyn search for "gnu mailman mailing lists"
<reepca>I did a test in the Guile REPL, the third argument match-error is thrown with isn't normally quoted
<reepca>which means that most likely something is trying to match a struct and is being given the string representation of it instead
<reepca>nly: have you reconfigured recently?
<nly>Can't see the paste, even if i could i wouldn't be of use apart from confirming if it's the same error
<nly>a week earlier, yes
<reepca>herd: exception caught while executing 'start' on service 'user-homes':
<reepca>Throw to key `match-error' with args `("match" "no matching pattern" "#<<user-account> name: \"root\" password: \"\" uid: 0 group: \"root\" supplementary-groups: () comment: \"System administrator\" home-directory: #<procedure home-directory (x)> create-home-directory?: #t shell: \"/gnu/store/qn1ax1fkj16x280m1rv7mcimfmn9l2pf-bash-4.4.23/bin/bash\" system?: #f>")'
<nly>No it's different
<nly>reepca how do you "test in guile repl"?
<reepca>in emacs: M-x run-guile, then enter (use-modules (ice-9 match)), RET, then (match 'foo ('bar 3)) and look at what the third argument with the 'match-error' reported is
<reepca>though that supposes one has geiser installed
<reepca>you could also just run guile directly and enter the same stuff
<nly>Thanks
<reepca>the error that shows up has the <user-account> record in quotes, which means somewhere match is actually getting passed the string representation for some reason
<snape>hi
<sneek>Welcome back snape, you have 1 message.
<sneek>snape, civodul says: hope you'll like this: https://issues.guix.info/issue/33515 :-)
<snape>I have issues building guix, things related to zh_CN translation. Did it happen to anyone?
<nly>`guix gc --verify` outputs nothing, i don't think it fixed anything
<nly>`guix gc --verify=repair,contents`*
<nly>:/
<nly>not to me
<snape>oh it works now. I had to 'git clean -xdf'. 'make clean' wasn't enough.
<snape>oops. spoke to fast. same error
<snape>too
<snape>is there a way to *not* build the translations?
<brendyyn>subscribed and sent my email, but still no response. I've got a custom domain with mailbox.org so maybe there is something wrong with it or the gnu servers dont like it
<snape>brendyyn: if it's the first email to a gnu list, it's normal to wait, there's human moderation
<brendyyn>oh
<brendyyn>i got no message back saying there would be moderation
<snape>yes, but still, there is :) And it happens in the US, so it's a bit early now
<brendyyn>Good morning USA
<reepca>would any of you brave souls be willing to try reconfiguring with latest guix? I'm wondering if this user-homes weirdness is exclusive to me
<snape>what is the weirdness?
<reepca>snape: this https://paste.debian.net/1080151/ happens when I try reconfiguring or manually running 'sudo herd start user-homes'
<rekado_>snape: human moderation? I don’t think so. It’s just grey listing.
<snape>rekado_: nope, it's human moderation
<snape>the first time
<rekado_>how do you know?
<snape>I talked to them
<rekado_>reepca: this is normal when updating from an older Guix. There’s been talk about this on the mailing list before.
<brendyyn>I was suprised that we never get any spam. I guess that's the reason
<snape>they can be contacted through listhelper-moderate@gnu.org, they are very nice people
<snape>hmm actually, that's for the Guile mailing list
<reepca>rekado_: I can't seem to find it in my mail archives just searching for 'user-homes', could you spare a link?
<snape>but seeing https://lists.gnu.org/mailman/admin, it seems that it's not the same for all mailing lists
<snape>so maybe for Guix it's another story
<brendyyn>Ok well I got back messages for my subscription, so it seems my email is at least workign
<snape>which might be greylisting, as rekado_ says
<snape>brendyyn: you don't need to subscribe to post to the list
<brendyyn>snape: well I sent an email to guix-patches@gnu.org but it didnt go through
<brendyyn>twice
<brendyyn>once before registering and once after
<snape>I've sent tons of emails and I never subscribed
<brendyyn>how do you get the emails without subscribing?
<snape>web interface, I click "reply"
<snape>(or before I subscribe with another email address)
<snape>*subscribed
<brendyyn>Hmmm ...
<snape>git clean -xdf + guix (freshly pulled) environment -C guix + ./booststrap + ./configure --localstatedir=/var --sysconfdir=/etc + make
<snape>doesn't work
<snape>on the latest guix git repo
<snape>I can't understand why I'm alone not being able to build guix
<swedebugia>@snape, there was a fix related to zh_CN translation .texi file not being found comitted recently. Try again.
<snape>swedebugia: I just ran 'git pull'
<snape>but indeed the error is ./doc/guix.zh_CN.texi:15: @include: could not find version.zh_CN.texi
<snape>swedebugia: do you have a link to a related discussion or bug?
<roptat>Eh, that should be version-zh_CN.texi
<roptat>I'm not on my computer right now, but could you check it's the file name in po/doc/guix-manual.zh_CN.po?
<civodul>Hello Guix!
<snape>roptat: ok
<snape>hey civodul
<civodul>hey snape!
<snape>roptat: it works now, I'll push the fix
<roptat>Thanks
<brendyyn>Hi civodul . I have a new email address and I'm trying to submit patches but the email is not going through and i get no responese
<brendyyn>i was able to subscribe to the list but it still doesnt work
<civodul>brendyyn: the first post is delayed a bit, be patient
<civodul>roptat: i've reconfigured bayfront
<civodul>starting knot doesn't work yet, but i suppose that's because we need the FSF sysadmins to DTRT
<roptat>civodul, that should not be the case
<roptat>does knot fail to start or can you simply not resolve?
<roptat>also make sure people can request bayfront on UDP port 53
<rekado_>I wanted to write system tests for the new NFS service, but this won’t work because the VMs use overlay file systems. Can this be avoided in the marionette?
<civodul>roptat: knot starts and runs fine, but it says: error: [guix.gnu.org.] zone event 'journal flush' failed (failed)
<civodul>rekado_: you could avoid the overlay by running on a full file system like we do for the install tests
<civodul>but that'll take space and time
<rekado_>I see. I’ll give that a try. I’d rather take up space and time than not include tests for this :)
<roptat>civodul, that's not supposed to happen
<roptat>I think I had this once... maybe try to increment the serial
<civodul>roptat: i'd happily "increment the serial", but how do i do that? :-)
<roptat>in (sysadmin dns) I think, there is knot-zone-file-serial
<civodul>ok
<civodul>rekado_: is it OK if i reconfigure berlin with the knot service?
<civodul>will UDP/53 inbound be allowed?
<civodul>roptat: i incremented the serial but AFAICS that number doesn't appear in knot.conf
<raingloom>hey, so, I specified --keep-failed but there is no build directory in /tmp? even though the log even says "note: keeping build directory `/tmp/guix-build-mininet-2.2.2.drv-9'"? what do?
<rekado_>civodul: I don’t know. I guess it’s blocked.
<rekado_>civodul: you may reconfigure, of course.
<civodul>rekado_: i wonder what to do if we cannot have it unblocked on time
<roptat>civodul, it should appear in the zone file, so that's OK
<civodul>roptat: ↑
<roptat>does it work now?
<civodul>nope
<roptat>ah, try removing the dnssec stuff
<rekado_>civodul: instead of using the ticketing system to ask them to unblock it I could call them to ask for having it done right now.
<civodul>roptat: there's also: "warning: [guix.gnu.org.] with automatic DNSSEC signing and outgoing transfers enabled, 'zonefile-load: difference' should be set to
<civodul>avoid malformed IXFR after manual zone file update"
<rekado_>I’m not in the office this week, but I can ask my colleagues to call on my behalf :)
<rekado_>(they are also better at pestering IT than I am.)
<civodul>heh, ok :-)
<civodul>sudo unblock that port
<roptat>civodul, I need to read knot's documentation
<rekado_>zactly :)
<rekado_>civodul: if you reconfigure now, please don’t reboot as I’ll be out for a few hours.
<civodul>alright!
<civodul>i'll wait for instructions from roptat anyway
<civodul>and i need to go to the bakery too
<civodul>important things :-)
<civodul>raingloom: did you enable offloading? -K doesn't work when offloading
*rekado_ goes to the garden to water plant babies
<roptat>it seems there are new possible fields in the config
<roptat>I'll do a quick update to the guix service then
<raingloom>civodul: no offloading afaik. i'm not even sure how i'd enable it.
<civodul>raingloom: hmm did you run guix-daemon with a TMPDIR variable?
<civodul>otherwise it should definitely be there
<civodul>roptat: could you add yourself to bayfront.scm, along with your SSH public key?
<civodul>then i'll reconfigure and you can test there directly
<brendyyn>My patches turned up on debbugs, but not the guix-patches archive
<raingloom>civodul: not that i know of, but i'll check...
<thomassgn>this is strange, I run 'guix build -c1 -M1 youtube-dl' and it produces '/gnu/store/xnj83g9qd872cm130h370lygxabd9vlk-youtube-dl-2019.04.24'. I then run 'guix package -i youtube-dl' it says ' youtube-dl 2019.02.18 → 2019.04.24 /gnu/store/xnj83g9qd872cm130h370lygxabd9vlk-youtube-dl-2019.04.24' so I'm thinking it will only update my profile with the store item produced just now, instead it starts 'building
<thomassgn>/gnu/store/3wbks17jgqsys7az05w11iqh9sib5dpk-ghc-8.0.2.drv...'
<raingloom>civodul: i can't see it being set in its /proc/$pid/environ file, so i guess that's a no
<thomassgn>same for libcgroups, guix build gives me a store item, but guix package rebuilds the same store item instead of updating the profile...
<janas>I've been trying to hibernate to an encrypted swap file, I think I need to add a "resume=" arg to the kernel. Does anyone know if there's a recommended way to do this?
<janas>Otherwise I'll keep looking through the initrd files, but I don't think I understand them yet!
<janas>Nevermind, I think I figured it out!
<roptat>thomassgn, could it be that it actually builds profile hooks, or there's grafting going on?
<thomassgn>I don't know. It previously used a long time building ghc-10 and then it starts building ghc-8. could be it's needed for grafting or some such.
<roptat>but I don't think it's rebuilding the same store item it already has
<rendaw>Hello all, how can I install a secure file as part of the build? Ex: nginx private key or ddclient secrets file. (local-file) will copy it into the store where it's world readable
<thomassgn>I'll wait it out, see what happens. :)
<roptat>rendaw, we don't have any support for secrets, although that's something many users want
<roptat>for now, you can only have that file managed out of band and include it in a configuration
<roptat>so for nginx for instance, instead of (local-file) you would use a simple string with the absolute path to the secret key
<rendaw>roptat: Ack, so in a disk-image build where root is read only it's not possible to distribute a ssl cert?
<roptat>a cert is public so that's ok, but for the private key, I don't know how to do it
<rendaw>right, sorry, meant the private key for the cert
<rendaw>:( ... maybe I could mount the disk image and install the file post-build? Would that mess up etc-service?
<roptat>I don't think it would mess up anything
<rendaw>I'll give it a try. Is there a ticket or anything tracking secret support?
<roptat>I don't know
<rendaw>Okay thanks
<roptat>it's something we discussed physically and on IRC at least
<thomassgn>rendaw: you could also look at --export and --share flags for 'guix system vm-image' maybe you could expose the keys through FS that way
<rendaw>thomassgn: looking!
<rendaw>thomassgn: Since the vm-image output is qcow it isn't straightforward to flash that to a drive, and there may be other build differences that make it unsuitable I think
<rendaw>my current plan is to do something with losetup
<thomassgn>:)
<roptat>civodul, done
<g_bor[m]>hello guix!
<civodul>hey g_bor[m]!
<civodul>roptat: done, you should be able to log in now!
<g_bor[m]>I am having a look at static networking stuff right now.
<civodul>muahaha, got ya!
<civodul>:-)
<civodul>g_bor[m]: nice
<roptat>civodul, I also sent an updated patch for (guix self)
<g_bor[m]>I wonder if we have any standard way to report an error if a config field is malformed. An example would be handy...
<roptat>civodul, "Could not chdir to home directory /home/roptat: No such file or directory"
<civodul>roptat: awesome, i'll look at the patch shortly
<civodul>re /home, that's because user-homes didn't restart on that machine for obscure reasons
<civodul>let me create it manually
<civodul>roptat: should be fine now
<roptat>thanks
<civodul>g_bor[m]: there's 'leave' and 'report-error' in (guix ui), but note that we carefully separate UI code from non-UI code
<roptat>how do I test a new config?
<civodul>for instance (guix build syscalls) is non-UI code
<civodul>roptat: you can do "sudo su -"
<g_bor[m]>civodul: ok, thanks.
<civodul>roptat: there's a checkout of guix-maintenance in ~root/maintenance, and an up-to-date guix
<civodul>root's guix is at 435551f0faccb4f91cadc709ffb9819d850d7e69, says 'guix describe'
<g_bor[m]>I belive that for now I will just come up with a rough outline, then you can comment if anything is off. For now I intend to do the call out to ip approach, and later fix up the netlink things.
<civodul>sounds good!
<civodul>we probably won't have enough time to integrate it in 1.0.0, but that's okay
<civodul>we can always continue this afterwards and we'll happily have it in 1.0.1
<roptat>civodul, ah but I don't have a password ^^
<civodul>good point
<civodul>roptat: BTW, when you reconfigure and such, it's often more efficient to pass --no-build-hook
<roptat>ok
<civodul>otherwise too many little things get offloaded and it can actually be slower
<brendyyn>Sigh. I make typos in messages sent to fix my previous typos
<OriansJ>brendyyn: it is expected when your hands move slower than your mind.
<brendyyn>OriansJ: Isn't the opposite? my hands move faster than my mind can pay attention to correctness
<OriansJ>brendyyn: generally no; since one's hands get the feel of common corrections and patterns. What tends to cause typos is your brain is sending write this word and that word faster than the hands can process.
<brendyyn>I think my problem is that my table is too high so my hands are always in an awkwark position
<brendyyn>I can't type reliably regardless
<brendyyn>I have a new mechanical keyboard too which are clunky
<OriansJ>well put your hands down
<brendyyn>The chair doesn't go up further :P
<OriansJ>brendyyn: I didn't say put chair up further, I said put hands down. Nothing should be shopping you from moving keyboard off desk and onto your lap
<brendyyn>sounds crazy
<OriansJ>one needs not see the keyboard to type well
<kmicu>brendyyn: make tyopos. Own ti!
<brendyyn>I really gotta learn to not use the mouse at atll
<kmicu>You could also don’t use keyboard. Problem solved.
*brendyyn wonders how kmicu uses a computer
<shcv>given our character oriented computers, and the current state of direct brain interfaces, I find keyboard only to be much more effective
<shcv>emacs for everything!
<OriansJ>shcv: needs a working hyper key on that keyboard though
<shcv>hmm, I only use C, M, and s... never seen any
<shcv>maybe I could make my "normal" left control key a hyper, since I'm using capslock as a control
<shcv>I use the windows key as a super, mostly for controlling EXWM
<OriansJ>shcv: doesn't that make H-C-M-s t a 3 hand combo?
<shcv>just two hands
<shcv>I would have to move my right over to hit the "t a 3" though
<shcv>trying to do it with one hand would technically be possible, but very awkward
<OriansJ>shcv: personally think someone needs to start making new space cadet keyboards
<shcv>sweet, it has roman numeral keys
<shcv>doesn't go up to V though
<kmicu>brendyyn: just talk to computer (at 32min https://www.invidio.us/watch?v=YRyYIIFKsdU )
<shcv>hey, thumbs up and down buttons
<shcv>way before its time - Facebook hadn't even been invented yet
<shcv>or reddit
<shcv>yeah, that's a pretty cool keyboard
<shcv>if I ever figure out how, I want to build a custom keyboard that looks a bit like this one: https://deskthority.net/viewtopic.php?f=7&t=11734&start=
<OriansJ>shcv: the how is simple, you experiment and make several iterations.
<raingloom>so, i'm setting BINDIR to %output/bin, and it seems like i'm doing the exact same thing the other packages do, but the build complains there is no such directory, what am i missing?
<raingloom>also, where is the %outputs variable documented? the online docs are not always super easy to search...
<buffet>hey, im trying to install guixsd on my laptop. mostly going with the barebone "template" config, everything seems to go well, but then i install i get a bunch of "failed to start service" and then nothing.
<buffet>here is what happens (sorry that it's ony a photo): https://0x0.st/zbVW.jpg
<dalz>Out of curiosity, what do y'all use as web browsers? I used to use qutebrowser, but the guix package is outdated (for ethical reasons I think?), I'd rather not spend 5 minutes every time I vist a new page with icecat just to end up whitelisting everything, next seems to be too unstable and eww is rather extreme.
<roptat>civodul, knot service is working on hydra \o/
<Formbi>dalz: normal Firefox
<roptat>dalz, icecat
<roptat>buffet, could it be that it failed to mount you filesystems?
<roptat>how did you partition your system, and can you share your config?
<dalz>roptat: i admire the dedication
<roptat>:)
<dalz>Formbi: spare some build script? :)
<Formbi>I use Arch with Guix so I don't need one
<dalz>Or package definition
<dalz>Or whatever they are called
<dalz>Formbi: i see, well another one on the list of things to package
<Formbi>I think a modified icecat definition would be ok
<roptat>dalz, https://framagit.org/tyreunom/guix-more/blob/master/more/packages/gnuzilla.scm
<roptat>it won't make it into guix proper though
<roptat>and it's a bit outdated because I only needed it before the last icecat update
<roptat>but it's still newer firefox than icecat
<roptat>there might be security issues too
<OriansJ>you may wish to make a pull request to https://github.com/guix-users/guix-nonfree if you make a firefox package definition
<dalz><roptat "dalz, https://framagit.org/tyreu"> roptat: <3
<civodul>OriansJ: please this is not the channel to discuss this
<civodul>roptat: so knot is running on bayfront, right?
<roptat>yes it is!
<roptat>you may wish to adjust the zone file in (sysadmin dns)
<civodul>yay!
<civodul>what for?
<roptat>otherwise, it's only waiting for the delegation from the gnu sysadmins
<civodul>ok
<roptat>in case it's not what you want ;)
<civodul>i guess it doesn't hurt to leave it as is, right?
<roptat>sure
<civodul>cool
<buffet>roptat: the esp on sda1 (mounted to /boot/efi), ext4 on sda2 (mounted to /) and a swapfile in /swap. config is at https://0x0.st/zbVg.txt
<civodul>roptat: do you want to email sysadmin@fsf.org Cc: me asking them to do whatever needs to be done?
<civodul>DNS-wise i mean
<roptat>buffet, looks good... are you sure sda2 has a label?
<civodul>BTW Firefox _is_ free software and as such it'd be welcome in Guix; there'd be changes similar to what IceCat does to make follow the FSDG, but it's doable
<civodul>just sayin' ;-)
<buffet>roptat: yes
<jayspeer>is it really? as far as I know you cannot redistribute the binary
<roptat>if you build it as aurora, disable drm, change the main addons page, it should be ok, right?
<OriansJ> https://lwn.net/Articles/676799/ for reference
<jayspeer>I have a noob question: how do I install old package version? sed-4.5 fails to build and is blocking all the other stuff
<ng0>jayspeer: that goes for the branding
<ng0>there is a non-trademarked branding
<ng0>and you can get permission to redistributed the branded one, as did Nix
<civodul>roptat: something like that, which is what IceCat does (it does additional things, too, but these are about settings that improve privacy)
<roptat>I'll see if I can update my package next week then
<roptat>I tried the regexp-unquote thing, but it unquotes parts of the regexp I add to the strings ("[ \n]+"), so that doesn't work
<roptat>should I use it on msgid only?
<roptat>civodul, berlin doesn't seem to resolve guix.gnu.org yet
<buffet>roptat: berlin as in the city? works fine here
<buffet>roptat: even not using the labels, but specifying the devices direclty, i get the same errors on boot
<roptat>buffet, I think we're missing the first error messages, maybe you can go up with Shift+PageUp?
<buffet>sure, one sec
<roptat>civodul, there are still errors in /var/log/messages though...
<buffet>roptat: first screen: https://0x0.st/zbV5.jpg second screen: https://0x0.st/zbVR.jpg third screen is what i sent before
<roptat>what's the file-system for sda1?
<buffet>fat32
<civodul>roptat: regexp-quote should just for the msgid i guess, perhaps i gave incorrect advice
<roptat>civodul, yes, it worked
<roptat>so I'll push in a second
<civodul>yay!
<buffet>acpi
<buffet>whoops
<civodul>roptat: re guix.gnu.org, it currently resolves to berlin + bayfront (see "getent hosts guix.gnu.org")
<jayspeer>sorry for a little of spam, but I'm not sure where to take it. I am unable to install anything, since sed fails to install with: inplace-selinux.sh: set-up failure: CONFIG_HEADER not defined
<jayspeer>ERROR: testsuite/inplace-selinux.sh
<roptat>civodul, hey, so it works already?
<kmicu>jayspeer: distributing blessed Firefox binaries is only necessary when we want to keep using Firefox™ name.
<roptat>ah it's not a delegation, it's only an A record
<buffet>if nobody else got an idea, i guess its time to give up for now
<kmicu>dalz: I’m using IceCat and w3m but I block js by default. I enjoy unblocking it for some sites; that ceremony makes me aware which pages I should avoid and generally forces me to stick to js-free internet.
<kmicu>buffet: could you tell us more about your filesystem layout and changes to bare-bones.scm?
<roptat>kmicu, https://0x0.st/zbVg.txt
<buffet>kmicu: the esp on sda1 (mounted to /boot/efi), ext4 on sda2 (mounted to /) and a swapfile in /swap. config is at https://0x0.st/zbVg.txt
<roptat>it's minor, but: (home-directory "/home/alice") :)
<buffet>oh tahnks
<roptat>buffet, try "vfat" instead of "fat32"
<kmicu>Could you check sda2 is labled 'root'?
<brendyyn>Ok so a couple hours later my emails go through, but are displayed in some random order. fantastic
<kmicu>brendyyn: no need to worry, that’s understanable.
<brendyyn>Because we understand it, it's fine?
<kmicu>Cuz a new mail setup is never perfect from the beginning.
<brendyyn>What's wrong with my mail? many other people's are backwards too
<buffet>roptat, kmicu: works. thanks a lot!
<roptat>cool :D
<kmicu>buffet: could you apply fat32 to vfat and /home/alice to /home/buffet changes and let us know?
<kmicu>buffet: glad if works. roptat good catch.
<brendyyn>and i just sent with git send-email
<kmicu>brendyyn: are they in random order on the bugtracker ML or in your MUA?
<brendyyn>webpage
<janas>Has anyone else tried resuming from an encrypted swap partition or swapfile in guix?
<janas>I can't quite seem to get everything working
<civodul>janas: i think someone played with hibernation support long ago if you search through the guix-devel archive
<apteryx>I think resuming from swapfile is not possible, so I heard.
<civodul>but yeah, it's currently not supported on master
<apteryx>so you probably want a swap partition
<janas>civodul: apteryx: thanks for the info, I was hoping to avoid a repartition but I guess it's inevitable now :/
<janas>I had been reading through the arch wiki which made it seem doable but I'm guessing that guix's init ramdisk probably doesn't have the support yet
<jayspeer>I'll respond to myself for future generation reading log of this channel: if you encounter build error try running 'guix gc' and trying again. It seems to be working so far
<nckx>jayspeer: What kind of build error? The SELinux one above?
<nckx>Hullo #guixen.
*civodul sends a bomb to guix-patches! 💣
<janas>Also, when I set up guix with an encrypted root partition I noticed that suspend-to-ram didn't work, is that normal? And is there a known workaround?
<civodul>janas: i have an encrypted root and suspend to RAM works
<civodul>perhaps there's something else causing the issue?
<nckx>civodul: Encrypted swap tho'?
<civodul>ah, oh, dunno
<civodul>gentlefolks, feedback welcome! https://issues.guix.info/issue/35456
<civodul>we're agile, we can add a new CLI a couple of days before releasing!
<buenouanq>new guixsd install on an old pentium desktop
<buenouanq>install succeeds, boots grub, begins booting
<buenouanq>then drops me into guile
<buenouanq>but the text is huge and I can see back far enough to have any idea what's going on
<buenouanq>how do I uh.. figure this out?
<nckx>buenouanq: Debugging Guix boot failures is *horrible*, no sugar-coating that.
<civodul>buenouanq: the likely issue is that the root device was not found
<nckx>‘no such file or directory’ [Guile prompt]
<civodul>that could be because the UUID specified in the config is wrong
<janas>civodul: I turned swap off to test, still failed. Probably something else is wrong
<nckx>Enjoy.
<civodul>buenouanq: or because the /dev name is wrong, etc.
<buenouanq>well, this is the first thing that made me add an initrd line to my config
<nckx>civodul: Is there any place where more info could be found in that case? I have a very similar problem with a custom kernel.
<buenouanq>something about sata-sil
<buenouanq>including it allows install to happen successfully
<civodul>nckx: normally you should at least see a couple of informative line before the prompt
<buenouanq>but might this be related civodul?
<civodul>what "something"?
<civodul>:-)
<nckx>civodul: No ☹ Just the glorious UNIX error above.
<buenouanq>(initrd-modules (cons '("sata-sill" %base-initrd-modules)))
<buenouanq>something like that, I forget exactly
<buenouanq>s/sill/sil/
<civodul>nckx: and ,bt ?
<civodul>buenouanq: it might be that sata-sil.ko is not found in the initrd
<civodul>which could happen if there's an obscure dash vs. underscore issue
<civodul>what if you write it as "sata_sil" in your config?
<civodul>that seems to be its real name
<nckx>civodul: Hm, didn't see anything relevant but had very few lines (boot logo) and no scrollback. Will try again with logo.nologo & smaller font if you say it should contain more info. Thanks.
<civodul>and ",bt" doesn't show anything either?
<nckx>I was talking about ,bt.
<civodul>oh sorry
<civodul>ok
<nckx>No, sorry for assuming it was useless and making me take a closer look ;-) No Guile built-in pager by any chance?
<nckx>(No idea why scrollback's buggered but it was.)
<nckx>civodul: & an enthousiastic ‘sure, whatever, I guess’ from me on your patch set ;-)
<nckx>Does it work with manifests? Not saying it should, might be best to keep it as a tiny subset.
<civodul>nckx: thanks!
<civodul>it doesn't work with manifests
<buenouanq>RE:aliases here's what I've been using for a long time https://rectilinear.xyz/p/7181077289+
<buenouanq>be warned, some of them are super hack and gross
<civodul>nckx: it's really a stripped-down version of the 'guix package' equivalent
<civodul>buenouanq: fun, thanks for sharing
<nckx>civodul: OK, so I'll never use it so anything goes!
<civodul>:-)
<civodul>i might use it occasionally, we'll see!
*nckx uses C-r ‘x pac’ to recall ‘~/guix/pre-inst-env guix package -i …’, which is even shorter, when feeling declarative.
<raingloom>oh, the boot thingy reminds me, is there a place where I can put kernel parameters? I was experimenting with a shared BTRFS partition dual boot and that needs that kinda stuff.
<nckx>raingloom: KERNEL-ARGUMENTS.
<nckx>A list of strings.
<nckx>A top-level field op OPERATING-SYSTEM.
<raingloom>nckx: thanks!
<nckx>If I seem grumpy it's because TIL my ISP['s modem-router] silently drops DNS responses for 192.168… records and that's a very confused half hour of my life I won't see again.
<nckx>Why is everything suck.
<nckx>(Except Guix.)
<jayspeer>and TempleOS
<nckx>Is… wow. Is that still a thing?
<nckx>We must port Guix to it post haste.
<buenouanq>I wonder how Terry would have felt about that...
<jayspeer>Propably sth about Glows Not Unix...
<civodul>nckx: damn, you use 'pre-inst-env'? heresy!
*civodul goes cooking
*nckx didn't know Terry had died. RIP.
<nckx>civodul: No good?
<quiliro>bonan tagon
<quiliro>saluton
<buffet>hey, anybody using sway on guixsd?
<kmicu>buffet: I saw some folks using it here but they are not around here right now.
<janas>not sure why, but it turns out that tlp-service was breaking suspend-to-ram
<jayspeer>janas: That is kinda strange, given whole point of tlp is battery saving
<rekado_>oof, the R upgrade broke a number of packages in non-obvious ways. That’s a first.
<nckx>Unfortunately, not that strange… PC power management on anything that isn't Windows is usually hit & miss (and certainly unsupported by the OEM.)
<roptat>civodul, re translation on monday, I won't be at home, so you'll have to take care of it yourself :/
<jayspeer>damn it, sed error is back :(
<ArneBab>dalz: I installed Firefox from outside the packet sources — needed quite a few hacks to get it working — aside from that i use ungoogled-chromium and icecat.
<dalz>ArneBab: I tried installing ungoogled-chromium three times, but it hangs the system every time when building, around 80%. It seems memory related, as SysRq+F fixes it. I remeber it did the same last week when i was installing next (i think it was compiling webkit or smth), but with enough perseverance it eventually succeeded. I don't have as much patience now :)
<ArneBab>dalz: I had my system almost-lock-up two times today until I restricted guix to use only two cores -> guix pull -c 2 -M 2 && guix package -c 2 -M 2 -u
<dalz>I think I'll stick with icecat for now, and maybe try to package firefox as time allows
<ArneBab>I have 32 GiB of memory on this machine, but it’s still not enough to build chromium efficiently …
<dalz>ArneBab: I'll make sure to try that, thanks
<ArneBab>(though it does work, it just cannot exploit all my cores)
<dalz>ArneBab: looks like a bug then?
<nckx>There is no way to build Chromium efficiently ;-)
<dalz>I only have 8 GB, and assumed that was the issue
<ArneBab>dalz: no, it’s chromium. It also regularly almost killed my Gentoo box when I was still on Gentoo
<dalz>I'm sure it happend with next too
<dalz>And it doesn't use chromium afaik
<ArneBab>let’s see whether the firefox build from guix-more works
<ArneBab>Firefox is also not harmless, and KDE also has quite long build times, but Chromium is the worst I know.
<ArneBab>dalz: to try guix-more you can simply use channels
<dalz>I remember, months ago, waiting unpatiently for polybar to compile on arch...
<dalz>Times have changed
<ArneBab>see `info guix` and search (via CTRL-s) for channels
<dalz>ArneBab: nice, time to warm my box :)
<ArneBab>(cons*
<ArneBab> (channel
<ArneBab> (name 'guix-more)
<ArneBab> (url "https://framagit.org/tyreunom/guix-more.git"))
<ArneBab> %default-channels)
<ArneBab>just put that into ~/.config/guix/channels.scm
<ArneBab>then guix pull && guix package -i firefox
<wingo>good evening :)
<wingo>does anyone remember what needs to be installed to be able to compile -m32 on an x86-64 guix system?
***cbaines_ is now known as cbaines
<nckx>civodul: I've now added file:///home/nckx/guix to my channels.scm. Better? 😛
<wingo>for the archives, apparently the answer is that our gcc is not compiled with multilib support, you have to "guix environment -s i686-linux --ad-hoc gcc-toolchain"
<ArneBab>what would be needed to get firefox into guix? Would it be enough to replace the trademark stuff (logo + name)?
<roptat>ArneBab, hey I never used that as a channel, so it works?
<ArneBab>roptat: I’ll see now — it already pulled successfully, but firefox just failed to build
<roptat>ArneBab, I don't maintain that package anymore, so that was kind of expected
<roptat>maybe you can fix that?
<roptat>otherwise I'll give it a try on tuesday
<roptat>currently it builds as aurora so there's no trademark issue
<roptat>but there's still a drm software in there (I think it's downloaded at runtime) and the addon main page should be changed to something that doesn't recommend non free software
<roptat>then there's probably some more privacy-related changes that could be done, but they are not necessary for firefox to get into guix I think
<roptat>they would be nice though
<ArneBab>roptat: ah, that sounds good
<ArneBab>roptat: I thought there was a configure flag to disable the DRM … at least there’s a config flag and it’s off by default.
<ArneBab>so DRM is off by default (but might ask)
<roptat>we should check that it's not downloaded at runtime
<roptat>but maybe I already took care of it
<roptat>it would be best to even remove that part of the source code
<kmicu>ArneBab: basically we need to create a half-baked IceCat.
<ArneBab>roptat: there should really be a configure flag … (as in "mozilla should have that", not "I’m sure it is there")
<ArneBab>kmicu: basically I just want something which is up to date, because Firefox is evolving really quickly.
<ArneBab>it would be even better to have firefox beta
<kmicu>That’s why IceCat is based on LTS. Maintaining (fixing) non-LTS Firefox version is a lot of work.
<kmicu>(Cuz Mozilla still loves pushing questionable experiments and features on users (especially in dev and beta versions))
<ArneBab>I really don’t mind those experiments, so icecat isn’t the right tradeoff for me
<ArneBab>I know that some of it is questionable, but part of this is just that I want to test it quickly
<buenouanq>hopefully someday soon will have a working server browser paradigm over gnunet and can leave that whole mess behind once and for all
<ArneBab>won’t be that easy since we still need to be present where our friends are
<ArneBab>though I’m trying to get people into Freenet
<rekado_>nckx, civodul: I’m also still using pre-inst-env, but that’s because (for historical reasons) I’m using a different localstatedir.
<ArneBab>roptat: it now tries to build firefox-66.0.3
<ArneBab>rekado_: why shouldn’t we use pre-inst-env anymore?
<rekado_>I didn’t say that it shouldn’t be used.
<rekado_>pre-inst-env is crude, though.
<roptat>ArneBab: I thought I had 63...
<rekado_>it doesn’t ensure that Guix runs in a suitable environment containing matching Guile dependencies.
<rekado_>“guix pull” achieves that.
<ArneBab>roptat: that failed to build so I first updated to 66 :-)
<ArneBab>rekado_: ah, ok — thanks for the info!
<roptat>I see
<roptat>Good luck :)
<ArneBab>thank you!
<kmicu>ArneBab: My point was more about a need to constatnly monitor new commits in Firefox and remove any recommending nonfree software. Mozilla doesn’t make that job easy.
<ArneBab>That’s something I don’t really need, but I understand that that means it cannot get into guix proper
<ArneBab>guix-more sounds good for that …
<ArneBab>I basically just need a simple way to update the package to the newest version
<kmicu>And that’s why we should not talk about guix-more here 😺
<ArneBab>😺
<civodul>rekado_: i think 'guix pull' is supposed to preserve localstatedir and similar settings
<civodul>you could check with "guix pull -p test", but i'm pretty sure this is the case
<ngz>Hello. I have a question about outputs in a package. AFAIU, a output is simply a new directory in the store where you split stuff from the main package. However, is it possible to have, e.g., different configure flags per output. For example, what if I want to have "--with-gui" for "gui" output, but "--without-gui" for "out"?
<nckx>ngz: Sure.
<ngz>nckx: Thank you, but the ratio (length answer)/(length question) is not great ;) Do you have an example in mind?
<nckx>(add-after 'install 'reconfigure-with-foo) (add-after 'reconfigure-with-foo 'rebuild-with-foo) … whether this is a good idea or not depends on the package. Might make more sense to have a bar and bar-minimal package for example, but I trust you know what you're doing.
<nckx>ngz: How's that for length? ;-)
<ngz>Much better :)
*nckx is human and needs time to type.
<nckx>Which is exactly what a bot would say.
<ngz>Ok, time for a Turing test ...
<nckx>sleep $RANDOM; echo $next_char
<nckx>Yes, I'm a bot written in bash. Explains SO MANY THINGS.
<raingloom>hey, can somebody help me with this? I can't figure out what to set the output to >_<
<raingloom> http://okturing.com/src/6142/body
<nckx>Back to your question: there are other approaches (configure two build directories in one phase, build both in the next, etc.). There are no restrictions, really. Just readability and common sense.
<nckx>‘okturing’… is this the test?
<ngz>nckx: are there package definitions to study?
<raingloom>nckx: what test? ,:|
<nckx>ngz: No idea, sorry. It all so depends on what exactly you're trying to do, and how well the upstream build scripts are written.
<nckx>raingloom: Could you elaborate?
<raingloom>nckx: it fails with the followin: install: cannot create regular file '/gnu/store/nxah9qfgq026ana848p11i5141rcnac4-mininet-2.2.2/bin': No such file or directory
<raingloom>(well, there is more in the log but that's the gist of it)
<nckx>(And the reason ‘build’ runs tests is probably because ‘make all’ does... you can (replace 'build) with something that doesn't.)
<ngz>nckx: Ideally, I would split "xcas" output from "giac".
<nckx>raingloom: (mkdir-p out)?
<ngz>And rename "giac-xcas" to "giac", which is the true upstream name.
<nckx>Having a stroke == the cool new package naming trend.
<ngz>Well, when I added "giac-xcas", I thought it was its real name.
<nckx>I've never heard of any of those things, I'm afraid.
<roptat>raingloom, maybe try to add a phase to mkdir that directory?
<ngz>Nix seems to do something smarter than me with outputs <https://github.com/NixOS/nixpkgs/blob/cf3e277dd0bd710af0df667e9364f4bd80c72713/pkgs/applications/science/math/giac/default.nix>
*nckx sometimes suspects their messages aren't getting through.
<jayspeer>Where should I report that package is unable to build?
<raingloom>roptat: i kinda thought these are created implicitly... dam.
<nckx>ngz: So xcas is just a GUI for gias?
<ngz>Yes
<nckx>ngz: But you can't just build gias --with-xcas and move xcas out of the way?
<nckx>I just want to avoid the :xcas output having a full copy of :out with some slight differences, basically.
<nckx>jayspeer: Which package?
<ngz>I'm not sure about what "--with-xcas" implies, besides the executable.
<jayspeer>sed-4.5 :/
<rekado_>jayspeer: where do we have that package?
<nckx>jayspeer: Builds fine for me.
<jayspeer>it fails at testing stage for me
<rekado_>the default “sed” package (version 4.5) builds just fine.
<rekado_>how does it fail?
<nckx>So if we're talking about a non-deterministic build failure I'd call that a bug and bug-guix@ is the place.
<jayspeer>ERROR: testsuite/inplace-selinux
<nckx>With full build output or at least anything that might be relevant.
<jayspeer>================================
<jayspeer>
<jayspeer>inplace-selinux.sh: set-up failure: CONFIG_HEADER not defined
<jayspeer>ERROR testsuite/inplace-selinux.sh (exit status: 99)
<nckx>jayspeer: Whoa there.
<nckx>jayspeer: Oh, I don't think SELinux is well-supported (or even supported?) at all.
<nckx>rekado_ would probably know
<jayspeer>I'm not trying to install it
<nckx>jayspeer: ?
<nckx>sed?
<jayspeer>selinux that is
<jayspeer>only sed
<jayspeer>There is not a trace of selinux on any of my systems
<nckx>jayspeer: Oh.
<nckx>jayspeer: What does ‘guix describe’ return?
<jayspeer>guix describe: error: failed to determine origin
*nckx is at 5 successful rounds of building sed and counting.
<nckx>Hum.
<nckx>jayspeer: guix --version?
<nckx>This is the latest guix-pulled version of guix, rite?
<jayspeer>I'm trying to pull right now.
<jayspeer>Most packages depend at some point at sed, so I have quite a pickle there
<nckx>…indeed.
<jayspeer>here is whole build log for sed http://paste.debian.net/1080282/
<jayspeer>it fails at testing stage, but my gentoo system is running its own sed 4.5 just fine
<nckx>jayspeer: This isn't a fix for your problem, but: are you intentionally not using substitutes and building the world from source?
<jayspeer>nckx: guix (GNU Guix) 0.16.0
<civodul>roptat: with the update of the 'guix' package, you can run the installer in Castellano or German, which is super cool
<nckx>jayspeer: Whoa. That's super old. You need to guix pull ☺
<jayspeer>I'm trying :(, fails at sed.
<nckx>Heh.
<rekado_>if you’re building from source anyway you could fetch the latest Guix from git and build that.
<jayspeer>I might as well try if it's not working anyway
<rekado_>Though I’m not sure if you’d be able to build sed then anyway. I don’t think the package has changed since 0.16.0.
*rekado_ looks at the logs
<rekado_>I wonder what CONFIG_HEADER is supposed to be.
<rekado_>building sed on a system where “ls -Zd .” matches “?” causes sed to look for the variable CONFIG_HEADER, which isn’t set.
<rekado_>I suppose that’s expected to point at config.h, where the autotools record whether SELinux support is available or not.
<rekado_>could you show us what “ls -Zd .” returns in your temp file system?
<rekado_>(the build farm nodes run the Guix system and there’s no SELinux support.)
<jayspeer>like this?: /tmp $ ls -Zd . output: ? .
<jayspeer>rekado_: I'm not using selinux
<nckx>jayspeer: ‘.’ and not ‘./’? On Guix System I get ‘./’.
<nckx>Maybe the comparison is that naive?
<jayspeer>ls -Zd ./ gives: ? ./
<nckx>jayspeer: Hmm, never mind, it's ls being p(r)etty by default when run by a human.
<nckx>env -i `which ls` -Zd . gives ‘.’ here too.
<rekado_>(when I run this in my shells I see “? .”, not “? ./”)
<jayspeer>I'm using eshell, let me try bash
<rekado_>should be fine.
<nckx>jayspeer: Nah, it's a red herring. Sorry for the noise.
<rekado_>I mean, this only tells us that SELinux is, in fact, not available.
<rekado_>and that should be okay.
<rekado_>this is the same as on the build farm.
<rekado_>so … why is CONFIG_HEADER not set?
<jayspeer>I cannot even find it anywhere :C
<rekado_>another thing that’s weird: your test suite log says that there are 60 tests in total, but I see a total of 161.
<rekado_>I suggest getting the latest version of Guix from git and try using that.
<rekado_>maybe one of the changes since 0.16.0 affect the behaviour of the “configure” script
<nckx>jayspeer: Did you see my question about substitutes?
<nckx>That could be a work-around and save you lots of time in the process.
<nckx>But building from source out of principle is fine, too.
<nckx>(And supported.)
<jayspeer>nckx: I tried adding default substitute server but it just hanged on me. I'll give a show
<nckx>jayspeer: Be sure to --authorize it too, i.e. sudo cat /etc/guix/acl returns something.
<nckx>(Guix can't just trust any --substitute-url from any user since it could be used to poison other users' packages.)
<jayspeer>I added ci.guix.info substitute server, and package is here
<jayspeer>still, it is not a satisfying solution
<jayspeer>I'll go with pull, then try turning substitute off
<nckx>jayspeer: Good compromise.
<nckx>I guess there could theoretically be a Trusting-Trust attack in your pulled guix but you're already trusting us with the rest anyway ;-)
<nckx>‘Come on, one pull won't hurt.’
<jayspeer>Thank you very much everyone. As we can see this 'bootstrap' was for from true bootstrap, but can we do ;)
***janneke_ is now known as janneke`
***janneke` is now known as janneke
*nckx has 4 keys in their ACL file and doesn't recognise 3 of them 🤷 this is fine
<jayspeer>if I simply erase entry in /etc/guix/acl the substitute will be remove? I don't see command line option to deauthorize sth
<nckx>jayspeer: Yes.
<nckx>If you don't want to use substitutes in the future you can just remove it (or truncate it if that causes errors).
<jayspeer>Oh s***. Here we go again: building /gnu/store/2iy8bjj9lqj0kx3djjyxwrvc5vin3sp9-sed-4.5.drv...
<nckx>jayspeer: We still have ~250 MiB of bootstrap binaries so your bootstrap wasn't that pure to begin with. Some day…
<nckx>Hum.
<nckx> /gnu/store/2iy8bjj9lqj0kx3djjyxwrvc5vin3sp9-sed-4.5.drv is the latest sed by the way.
<nckx>Hm, no, I lied, or pulled since then, but it's a very recent one.
<jayspeer>and now I am unable to authorize any server - it gives error: invalid access-control list ()
<jayspeer>you know what? I'll just nuke it and try again with substitutes
<nckx>👍
<rekado_>no
<rekado_>don’t
<rekado_>the acl file is incorrect
<rekado_>that’s what it tells you.
<rekado_>did you edit it?
<jayspeer>as nckx said
<nckx>jayspeer: But not by hand?
<rekado_>it’s fine to delete the file and then run “sudo -E guix archive --authorize < /path/to/the/key”
<jayspeer>by hand. I just echo'ed empty string to it
<rekado_>yeah, that’s not going to work.
<jayspeer>how so?
<rekado_>because the file exists, so it’s expected to contain a valid expression.
<nckx><nckx> ‘…or truncate it **if** [just deleting it] causes errors’ 😛 I remembered that one didn't work, just not which one.
<jayspeer>what's funny I'm running guix pull on my laptop (using source method) and it's running just fine
*nckx is having their own guix pull troubles on aarch64… OOM-killed with 8 GiB of RAM. :-/
<rekado_>nckx: what’s it building? Python?
<jayspeer>maybe chromium?
<rekado_>not for ’guix pull’!
<nckx>rekado_: .guix-real gets killed.
<rekado_>wha…?
<nckx>I made the mistake of retrying with swap ­— tip: don't retry with swap — and now have to wait 4 days until I'm home to reset the thing.
<nckx>Today was day 2.