IRC channel logs

2019-03-09.log

back to list of logs

<civodul>jackhill: hmm the gear used to be displayed at least no?
<civodul>i tested in a VM: the gear does appear if you have several WMs/DEs to choose from (not just GNOME)
<civodul>however, GDM now fails to start ratpoison for example
<mikadoZero>Using the `package -m` the manifest would duplicate what I have in my system config file. Is there an equivalent of .guix-profile for things that are specified with a system config file and put in place with `system configure`? I am trying to refer to ca-certificates.crt.
<mikadoZero>I have the file `locate ca-certificates.crt` shows many results.
<vagrantc>ok, i rebooted and suddenly the issue with 'guix pull ... openssl' went away.
<vagrantc>?
<vagrantc>what would cause: "guix pull: error: Git error: OpenSSL error: failed to load certificates: error:00000000:lib(0):func(0):reason(0)"
<vagrantc>that would be fixed by rebooting?
<mikadoZero>Upon further consideration I like the seperation of concerns for a system configuration and user manifests.
<mikadoZero>vagrantc: Thanks for your work on the c201.
<vagrantc>i've only done some basic testing, someone else did most of the work :)
*vagrantc is currently eyeing up a u-boot target to boot from depthcharge... https://github.com/archlinuxarm/PKGBUILDs/tree/master/alarm/uboot-veyron
<vagrantc>should basically be the same for c201
<mikadoZero>I think the naming is clever u-boot and depthcharge.
<mikadoZero>vagrantc: With u-boot is there full disk encryption like with grub?
<vagrantc>mikadoZero: no, there isn't full-disk encryption in grub
<vagrantc>i have had minimal success loading grub-efi from u-boot, which can then do full disk encryption
<vagrantc>but it's a little rough around the edges still
<mikadoZero>vagrantc: Are you using BIOS instead of UEFI then with the c201?
<mikadoZero>vagrantc: grub with full disk encryption. https://libreboot.org/docs/gnulinux/encrypted_debian.html
<vagrantc>mikadoZero: no such thing as BIOS on this architecture
<vagrantc>mikadoZero: u-boot has an EFI implementation
<mikadoZero>vagrantc: Thanks
<vagrantc>i'm not sure about coreboot/libreboot grub implementations
***catonano_ is now known as catonano
<vagrantc>you can do grub with full disk encryption without libreboot, although grub is still installed unencrypted
<mikadoZero>vagrantc: So grub would be on external media attached at boot time? While everything else is encrypted.
<vagrantc>it's possible, sure
<vagrantc>or grub on the disk itself, just like libreboot is on the flash rom
<mikadoZero>vagrantc: Thanks for your replies to my questions.
<vagrantc>the main x86 laptop i run guix on has grub installed on the disk, and a fully encrypted partition with guix on it
<vagrantc>it's supported "out of the box" on guixsd
<mikadoZero>vagrantc: Yes I am also using Guix SD setup that way.
<leungbk>how should i take someone else's github patches and integrate them properly into a package recipe? do i simply git checkout their branch and turn the relevant commits into .patch files?
<jackhill>sneek_ later tell civodul I was thiking of this https://lists.gnu.org/archive/html/guix-devel/2019-01/msg00131.html didn't know it had been fixed, and now I see that it mentioned the calendar thing too.
<pkill9>hi, is there a Danny Milosavljevic in the chat? I'm looking for help with this https://www.mail-archive.com/bug-guix@gnu.org/msg11188.html
<nly>Is anyone using simple-scan or SANE?
<nckx>nly: Many people, this week, it seems :-)
*nckx can now use their CanoScan through SANE, but not simple-scan or XSane. :-(
<nly>simple-scan is a GUI? I think it might be broken, i don't get any window on emacs x window manager
<nckx>nly: Oh. Yes. And it should start and work fine, it just freezes when I try to scan.
<nckx>nly: Do you get any interesting log messages when you start it from a terminal?
<nly>nope, mostly empty
<nly> http://nly.info.tm:9001/log/simple-scan.txt
<nckx>Still, a lot more than I get: http://ix.io/1CYH (that's running ‘$ simple scan’, then closing it from the top-left ‘icon menu’ or whatever GNOME calls it).
*nckx forget --debug, SMH.
<nckx> http://ix.io/1CYK
<nckx>nly: One difference is that I have hicolor-icon-theme in my profile and you don't. No excuse to freeze, but who knows.
<nly>Ah, it's hplip i am looking for(drivers), I see many commits by you, i presume, for hplip. :)
<nckx>nly: sane-backends should pull in hplip-minimal, though.
*nckx used to have an HP all-in-one scanamabob, but it's been a while and support might've gone stale.
<nckx>A missing driver also shouldn't freeze up a GUI…
<ArneBab>nly: I did not get my scanner working yet
<nckx>ArneBab: Does ‘scanimage > foo’ work?
<ArneBab>no
<ArneBab>dbus[9102]: arguments to dbus_connection_send() were incorrect, assertion "connection != NULL" failed in file dbus-connection.c line 3307.
<ArneBab>This is normally a bug in some application using the D-Bus library.
<ArneBab>scanimage: no SANE devices found
<nckx>…SANE uses DBus? Huh.
<nckx>That looks off.
<nckx>Ah, hplip does.
<nckx>ArneBab: I presume dbus-daemon is actually running?
<ArneBab>yes — three times :-/
<nckx>ArneBab: You win, I only have 2.
<ArneBab>:-)
<nckx>(One from dbus-service and one from dbus-launch i3, I guess.)
<nly>`scanimage > img.out` wow, thanks. Now i don't have to find the command i need from all these hp-* commands.
<nly>It works
<nckx>nly: It can also do jpg and even crop on the command line, but I just use the GIMP for that ;-)
<nckx>Glad I could (sorta) help!
<nly>thanks
<nckx>But looks like quite a few Guix users are just silently not scanning things, we should fix that…
<nckx>(I say we; I'm one of 'em; I have no clue what's wrong.)
<nckx>scanimage works; simple-scan scans only black pages; xsane freezes when I hit ‘Scan’. :-/
<nckx>…running simple-scan as root, of course, works. 🤦
<civodul>Hello Guix!
<nckx>Hullo civodul.
<nckx>It's possible that adding (usb-autosuspend? #f) to my tlp-configuration solved my simple-scan-as-regular-user woes, but who honestly knows at this point.
<phenoble>Hi #guix
<phenoble>Is anyone able to run the next browser properly?
<phenoble>When I try to have it access the internet instead of some offline html page, which is what that which is initially is shown looks like, the thing seems to lock up, giving me a white screen only.
<janneke>phenoble: no, it's very preliminary and needs more work
<janneke>it may just fit your use case, if you're lucky
<phenoble>Well, I can't seem to run it, anyway.
<phenoble>I wouldn't shy away from having to do some more-or-less serious lisp programming to get it do what i want, but it'd help if it ran before I did that ;)
<janneke>phenoble: oh yeah, that's bad -- it does show gnu.org for me; at least the version from begin this year
*janneke tries current master
<janneke>phenoble: works for me, as long as i enable my trackpad and navigate using touchy pointer-shoving
<janneke>that's master from 4 days ago: 8035819
<fosterite>Hello, if you run `guix challenge` on a brand new install it gives a division by zero error
<fosterite>I'm guessing 0/0 packages passed or something
<civodul>hello fosterite
<civodul>fosterite: how did you run it? did you pass any arguments?
<fosterite>just guix challenge
<fosterite>just installed in a vm
<fosterite>desktop.cfg, just changed timezone, bootloader, removed encryption settings
<civodul>could you paste the output?
<fosterite>yeah one sec
<fosterite>ah, my vm just crashed
<fosterite>I reopened it but I have 8 packages found by challenge (man, info, cert) so it isn't crashing
<fosterite>civodul: what if total was 0 : https://github.com/guix-mirror/guix/blob/f896abc71f7ed3de72df1c9af7477ee59737f768/guix/scripts/challenge.scm#L215
<fosterite>I think that's what happened
<civodul>fosterite: though when you run it without arguments, it takes all the live store items
<civodul>so it cannot be zero in this case
<fosterite>well, it was
<civodul>yeah that's i was asking for the backtrace :-)
<nckx>‘Cannot’; computers.
<fosterite>I assume there should be more than 8 items in the store, then? I am root
<fosterite>guix package -I gives no output
<fosterite>613 dirs in /gnu/store/ though
<rvgn>Any updates about Guix 1.0? XD Heard that tentative release is early March. O_o
<nckx>fosterite: ‘guix package -l’ lists only the contents of your (probably empty) user profile (root's in this case). All packages are probably ‘installed’ in the system profile.
<nckx>s|l|I|
<fosterite>nckx: looks like -l also lists something. How do I view the system guix profile?
<nckx>fosterite: Yeah, -l lists the generational changes over time.
<nckx>fosterite: That's what I was wondering too… ‘guix package -p /var/guix/profiles/system -I’ did not what I expect, which is ‘something’.
<rvgn>nckx What is the advantage of having all the required application packages in system profile vs in user profile? Like, everytime a user wants to install a package, one can either just do "guix package -i" or add the package in config.scm and do "guix system reconfigure".
<fosterite>specifically, it lists that ....root/profile does not exist
<nckx>fosterite: Makes sense, since you're running entirely from the system profile at that point (until you'd ‘guix package -i’ your first package as root).
<rvgn>nckx Of course the former method involve "per-user" advantage. But if there is only one user?
<nckx>rvgn: Advantage? Hm. I guess that all packages are available to new/unknown users immediately, so it makes sense for a ‘base set’, and it might save time/effort when deploying? I use manifests myself, so not an expert.
<nckx>rvgn: Disadvantage is that if, for example, you put libreoffice in your system configuration, ‘guix system reconfigure’ can take a long time. Not great if you're applying a security update. If libreoffice is broken, you can't reconfigure at all…
<rvgn>nckx I was thinking about deployment too. If there is only one user and if that user wants same set of packages in all of the devices, then mentioning all intretsed packages in config would be wise.
<rvgn>nckx Ah! That's an excellent thought.
<rvgn>Thanks!
<nckx>rvgn: I won't comment on wise, but it's certainly possible. I've done this. So nobody will stop you from doing that.
<nckx>There is no right and wrong, only taste and trade-offs. ☺
*nckx whispers ‘manifests are probably best in most cases though’.
<rvgn>So I think optimal would be having base packages in system profile and additional packages in user profile.
<OriansJ>nckx: there is impossible perfection and there is best practices; do not let the former prevent the development of the latter.
<rvgn>nckx You could be right about there is no right and wrong. But there is efficiency reasons.
<rvgn>*there could be efficiency reasons
<rvgn>and also could be logical reasons based on the guix's design model.
<rvgn>brb
<nckx>OriansJ: Not disputing that at all, but I can't well weigh the pros and cons of each approach if I'm not even familiar with one of them… ☺
<nckx>What makes a ’base system package’ is also a soft & subjective boundary.
*nckx has to run to catch a bus to see a man about a thing.
<mikadoZero>OriansJ: Thank again for pointing out fbterm. Looking into it I saw on the arch wiki a comment that it's development has stoped. fbterm's last commit looks to be 2008. So I looked into kmscon which I like. kmscon's last commit looks to be 2014.
<mikadoZero>I like the idea of getting the kernel console out of the kernel and into user space. Also aligns better with the idea of microkernel which I find more appealing than monolithic kernels.
<civodul>hmm we're getting many identical HTTP requests for Cuirass on berlin
<civodul>if anyone here is testing a script, consider giving it some rest :-)
<OriansJ>civodul: sudo iptables -I INPUT -s ${ip}/32 -j DROP with an at job to pull the ban after 30 minutes usually does the job
<OriansJ>mikadoZero: have you taken a look at http://genode.org/ yet?
<mikadoZero>OriansJ: I have looked at genode.org a few times. I have also been looking into other projects as well: sel4, helenos, minix 3 and others.
<mikadoZero>A combination that I would like is risc-v, a microkernel (preference for formal verification) and as Guix SD on top.
<OriansJ>mikadoZero: the problem is sadly no one has figured out how to formally prove correctness in a SMP kernel. Which really is essential
<OriansJ>then there is the problem of RISC-V being impossible to bootstrap with anything less than a C compiler (Or a big assembler)
<mikadoZero>So for microkernels with formal verification as a goal there is sel4, muen and maybe helenos (have to look into helenos further). Also I still have to check if muen is being ported to risc-v.
<mikadoZero>OriansJ: Still I think it is progress over monolithic kernels.
<civodul>OriansJ: yeah i could do that, we'll see if it's needed
<civodul>just checked the IP and apparently it's one of us :-)
<mikadoZero>civodul: What do you mean "I could do that" and IP?
<civodul>mikadoZero: i meant i could temporarily black-list the IP, but i think i won't need to do that
<mikadoZero>civodul: Okay
<mikadoZero>OriansJ: Is there any particular reason why you singled out genode?
***rvgn-net is now known as rvgn
<OriansJ>mikadoZero: it seemed the project you would most likely find interesting
<mikadoZero>OriansJ: Thanks
<fosterite>is there package search other than -A | grep ?
<OriansJ>fosterite: guix pakcage -s ?
<fosterite>OriansJ: thanks
<fosterite>where's the `clear` command? -s didn't find it
<OriansJ>fosterite: clear is a bash built-in
<OriansJ>fyi Ctrl-l does the same thing
<fosterite>not in default bash apparently
<fosterite>looks like it's in ncurses
<fosterite>in osx at least, should be in there somewhere
<fosterite>I didn't know about ^l, thanks
<fosterite>How do I switch widnows managers / desktop environments? I see that gnome and xfce are the default but I installed stumpwm and it did not show up in the list
<fosterite>does it need to be a system package?
<OriansJ>fosterite: assuming you installed slim F1
<fosterite>there is an f1 option but it only cycles between gnome and xfce (and gnome on xorg)
<kmicu>Hi fosterite: how did you install stumpwm? Could you add i3 to packages entry in config.scm, reconfigure and check (after Xorg restart) whether i3 is visible?
<dustyweb>hello hello!
<dustyweb>rekado: davexunit and I are doing a guix deploy hackathon, are looking at your berlin-deploy code
<fosterite>kmicu: as user. I will try what you said.
<OriansJ>fosterite: you can also in your ~/.xinitrc put exec stumpwm
<civodul>hi dustyweb, that sounds great!
<dustyweb>civodul: :)
<fosterite>OriansJ: will that mess with the login manager (which is slim by default, afaik)?
<fosterite>kmicu: are you a lojbanist?
<efraim>I was hoping to look at ocaml-findlib to figure out something similar for perl6, but I see it's the ocaml build system overlaid on the gnu build system instead
<OriansJ>fosterite: it will ignore the desktop you selected via F1 and open your desired desktop
<fosterite>OriansJ: that's what I mea by mess with it
<fosterite>It feels bad to layer a kludge like that on top of a heavy duty applied elegance like guix
<OriansJ>fosterite: true but it is required to work around lack of control of system configuration control and installing a custom desktop environment
<fosterite>OriansJ: how bad would just removing the problematic system config be? It looks like it's buried in the % variables
<OriansJ>fosterite: well if you would like you can leverage the most minimal guix example https://paste.debian.net/1072493/
<kmicu>fosterite: not realy.
<OriansJ>feel free to remove or alter anything in the package list you don't want
<fosterite>OriansJ: thanks this looks perfec
<OriansJ>fosterite: it also is fully encrypted (including /boot)
<fosterite>Is the guix program added by guix-services? I note it's not part of the packages
<fosterite>kmicu: ah, your name looks like it might the kind of name a lojbanist would pick so I thought I'd ask
<OriansJ>fosterite: here is an example procedure for using that config (hosted on a web server for simplicity) https://paste.debian.net/1072494/
<efraim>didn't solve my current issue, but I think I did figure out some other perl6 module issues
<fosterite>kmicu: it finished building. Yes, i3 is available
<efraim>tough choice time: propagate perl6 lib inputs or install them to %output
<efraim>i'm actually leaning the second one so we can have not conflicting versions like with python libraries
<quiliro>hello guix
<quiliro>where can i look for information about a message that comes up when starting emacs from the terminal window on gnome guixsd? I copy the message:
<quiliro>(process:1355): Gtk-WARNING **: 14:54:59.141: Locale not supported by C library.
<quiliro> Using the fallback 'C' locale.
<efraim>hmmm, looks like zef decided for me, propagated it is
<quiliro>please tell me the procedure to have help...only mailing help-guix?
<fosterite>usually asking here
<fosterite>* will work too
<quiliro>i just did
<fosterite>what locale did you set for your system?
<quiliro>es_EC
<quiliro>is it because i did not install glibc-locales-utf8?
<quiliro>but isn't the installer supposed to do that when the locale is es_EC?
<rekado>quiliro: glibc-locales should contain es_EC.utf8
<fosterite>irc is slow sometimes unfortunately
<quiliro>glibc-locales is not installed if the locale is es_EC?
<quiliro>automatically...
<quiliro>each user must install it?
<fosterite>apparently not. does es_ES.utf8 work? looks like that's a default locale (along with AC, CL, MX)
<quiliro>or must it explicitly be specified in desktop.scm??
<fosterite>you will need to add it looks like yes
<quiliro>(locale "es_EC.UTF-8")
<quiliro>is it wrong?
<fosterite>what does emacs say
<quiliro>that is the message emacs outputs
<quiliro>what i put first on the chat
<quiliro>should i repeat?
<quiliro>(process:1355): Gtk-WARNING **: 14:54:59.141: Locale not supported
<quiliro> by C library. Using the fallback 'C' locale.
<quiliro>that is the exact message
<quiliro>there is nothing more
<fosterite>try running it again with your locale set to es_ES.utf8, I'd like to see if it's finding that one.
<quiliro>good idea!
<fosterite>Also, what are the contents of /run/current-system/locale/ ?
<fosterite>for me, it's 2.28/ (a bunch of stuff, but es_ES.utf8/ )
<quiliro>es_EC.UTF-8/
<quiliro>but not es_EC
<quiliro>should i change it on the system reconfigure locale?
<fosterite>I think so yes
<quiliro>by the way, how do i execute the emacs with another locale? just run locale es_EC.UTF-8
<quiliro>and then run emacs?
<quiliro>locale command not found
<quiliro>when i run locale es_EC.UTF-8
<fosterite>I am not sure, but changing it in the system configuration and rebooting will definitely work
<rekado>*reconfiguring
<fosterite>I just reboot every time ¯\_(ツ)_/¯
<rekado>rebooting won’t change your system.
<rekado>only reconfiguring does.
<fosterite>I call guix system ... and then reboot
<fosterite>I see what you mean though, that's what you actually have to do
<quiliro>i installed glibc-locales. but locale is not found
<quiliro>that command is not installed
<quiliro>why is that
<quiliro>?
<fosterite>Did you change your locale in config.scm?
<quiliro>no...i do not want to restart...i just installed on the user
<quiliro>is it not possible to change locale on the user?
<fosterite>It probably is, I don't know how though
<rekado>“locale” is part of the glibc package, which is generally not installed in any profile.
<dustyweb>we need to serialize the operating-system definition and pass it across the wire
<dustyweb>I think I don't see such a procedure
<dustyweb>I'm guessing quasiquoting an s-expression of it is probably the best way for now
<lnikkila>Hello, is there a system-wide way in GuixSD to define channels so I could reference packages in /etc/config.scm?
<rekado>dustyweb: serializing the operating system isn’t possible yet.
<rekado>dustyweb: you may also need to ensure that the environment in which it is evaluated is the same on both sides.
<rekado>I guess for a first pass you can assume that the configuration only depends on Guix itself, no channels, no extra files.
<dustyweb>rekado: yeah, we were just discovering that... good to get a confirmation :)
<dustyweb>rekado: we are wondering if the final reconfigure step can just be done remotely
<dustyweb>and if that would just solve the whole thing
<dustyweb>untie the gordian knot and all that
<rekado>in my script it is done on the target, but that requires that the configuration be available on the target, so that’s not great.
<rekado>reconfiguring at a distance would be neat.
<rekado>if you already have the system on the target then you should be able to switch to that system remotely
<rekado>you would need to compute the operating-system value locally and then cause the effects on the remote, in the knowledge that all required files already exist there.
<rekado>I don’t think reconfigure does much more than switching symlinks, installing GRUB, etc (compared to “guix system build”).
<rekado>doing a remote reconfiguration should be possible
<OriansJ>rekado: it would probably be something along the lines of guix push ssh://user@host
<quiliro>installed glibc
<quiliro>executed locale but got error:
<quiliro>locale es_EC.UTF-8
<quiliro>locale: nombre desconocido "es_EC.UTF-8"
<quiliro>
<quiliro>unknown name
<OriansJ>followed by guix reconfigure script.scm on the remote end; where the push simply uploads the binaries but the configuration is entirely local
<quiliro>nombre desconocido = unknown name
<OriansJ>rekado: as pushing of binaries to a store has different security implications than changing of default binaries.
<rekado>we already have “guix copy” and my simple deploy script pushes the built system to the target.
<fosterite>quiliro: what's your locale set to in config? it found some locale, the 'C' locale is english
<OriansJ>and a simple ssh root@example 'guix system reconfigure example.scm' doesn't work?
<OriansJ>or were you hoping for more of an ipxe system image deploy sort of thing?
<dustyweb>rekado: we are more or less getting that sense so I'm glad you're feeling the same thing
<quiliro>fosterite: LC_ALL=
<quiliro>
<quiliro>it is blank
<quiliro>when i type locale
<fosterite>quiliro: dod you set a locale in any guix configurations?
<quiliro>all other variables are like LANG
<quiliro>LANG=es_EC.UTF-8
<quiliro>or LC_CTYPE="es_EC.UTF-8"
<fosterite>ok, and the "nombre desconocido" error is from emacs?
<quiliro>yes...unknown name
<quiliro>fosterite: i do not get that error on running epiphany
<fosterite>quiliro: what does "locale en_US" say
<fosterite>quiliro: or "locale C"
<fosterite>hey installing icecat just failed
<fosterite>specifically at rustc
<fosterite>it's probably a RAM issue though
<dustyweb>rekado: ping