IRC channel logs

2022-11-14.log

back to list of logs

<a12l>I get an error message when I try to build my guix home: https://paste.debian.net/1260605/
<a12l>Is there an easy way to see what it tries to do?
<lechner>a12l / Hi, the message "Cleaning up symlinks from previous home" seems inconsistent with "guix home: warning: cannot determine provenance for current system"
<stevenroose>Is there a way to view the guix documentation all in one page so I can grep it? Not a single web search engine actually returns decent results from these pages.
<stevenroose>And then to my actual question: How can I blacklist kernel modules?
<stevenroose>Akin to adding a `blacklist` line to /etc/modprobe.d/XXX.conf
<crono>I tried to submit that bug about that emacs-guix issue just now and I cant see if the email or the debbug issue went through?
<crono>Does submitting a bug take a while to register?
<iyzsong>yes, i think first email need manual approve, may take a day.
<crono>Makes sense, I didnt know if it was automated or not, or even how the debbugs works.
<user>night guix
<a12l>lechner: The problem seemed to be that I'd created the symlinks `~/.guix-home` and `~/.guix-profile` manually. After deleting them and rerunning Guix everything worked fine. Maybe the Guix builder uses some user without permissions to delete the files? My `$HOME` does have 0700 permissions.
<a12l>nckx: I solved the issue by beginning to save all state directly under `$HOME`. Not the solution I wanted, but I want to test out Guix. Maybe I'll have to write my own scripts to handle ephemeral roots in the future :0
<raghavgururajan>jorge[m]: Quickest way is to use `guix import`. It'll spit package definition.
<apteryx>would you prefer the kernel Linux documentation as A) a "doc" output or B) as its own package?
<apteryx>(it's part of the kernel sources)
<unmatched-paren>apteryx: probably as its own package
<zachariah>Agreed
<apteryx>thanks for your feeback :-)
<apteryx>*feedback
<user>How can I make guile fall in interactive mode if a exception occurs?
<civodul>Hello Guix!
<sneek>civodul, you have 1 message!
<sneek>civodul, apteryx says: yes it sounds like the same bug! thanks for sharing the workaround, even crude
<raghavgururajan>apteryx: "doc" output. It's schemey.
<cbaines>this is an interesting bug, some random looking string (I see 7fbd0a171c30) is finding it's way in to the builder for protobuf-2
<cbaines>I bet this is related to the switch to gexp's a couple of weeks ago apteryx
<cbaines>I found this out from wondering why the Guix Data Service is flagging this Python package addition as changing protobuf-2 https://qa.guix.gnu.org/issue/59190
<cbaines>and it is, because the derivation/builder is changing constantly
<cbaines>but luckilly this is easy to reproduce as the package fails to build
<yarl>Hello guix!
<yarl>I run guix system, my /etc/config.scm does not contain "guile" in (packages ...). Still, /etc/profile sources /run/current-system/profile/etc/profile that in turn add /run/current-system/profile/share/guile/site/3.0 to GUILE_LOAD_PATH. Why is that?
<yarl>I am asking because If I understood correctly, /run/current-system/... is system related only. But my guile is ~/.guix-home/profile/bin/guile. Its load path should contain ~/.guix-home/profile/share/guile/3.0/ but not /run/current-system/profile/share/guile/site/3.0. Right?
<yarl>And currently, it is the opposite. Is it a bug or a misconfiguration
<yarl>The problem is, a shell login reads /etc/profile anyway. But according to the guix manual : "One should never have to touch files in ???/etc??? or to run commands that modify the system state".
<a12l>Gaah, more error messages when I try to rebuild my guix home component: https://paste.debian.net/1260645/
<a12l>Is it possible to get a more detailed view of what guix tries to do? It currently is very sparse with information, and it's very hard to debug with such little info
<rekado>apteryx There???s going to be SAN downtime for us in a few hours
<rekado>we???re only affected because we have no multipath.
<rekado>it???s a sequential reboot of the two controllers
<mekeor[m]>hello. is it possible to have auto-completion for guix-related files (e.g. system.scm) in emacs (e.g. based on eglot)?
<iska>mekeor[m] wdym? and where?
<sneek>Welcome back iska, you have 1 message!
<sneek>iska, unmatched-paren says: Hmm, yes, I can reproduce this...
<stevenroose>Does Guix look into /etc/modprobe.d/xx.conf on startup? I'm trying to blacklist a kernel module
<iska>steve456[m] there are blocklists in the kernel options of system config file
<iska>changing /etc is very un-guixy
<rekado>apteryx: phew, friday, not today
<mekeor[m]>iska: in emacs, e.g. when editing my guix-system.scm, i'd love to have auto-completion-suggestions, e.g. it could suggest me the properties/fields of the operating-system record
<iska>mekeor[m] geiser
<iska>just start a repl and load the needed modules
<iska>a must for any guile user :)
<mekeor[m]>but geiser does not suggest completions, does it?, iska
<iska>mekeor[m] it does!
<iska>you can also use readline inside the repl, it has tab completion
<mekeor[m]>iska: sorry, i meant suggestions inside the code buffer, e.g. the buffer for the system.scm -- not the repl-buffer
<iska>mekeor[m] I meant that too
<stevenroose>iska: hmm, was grepping for blacklist, is blocklist also used? didn't find much on how to do that, any snippet?
<iska>stefanc_diff look into system configuration in the guix manual
<stevenroose>iska: I have the manual open and I'm trying to grep things I could use, but not finding. It actually has an example of putting a /etc/modprobe.d file related to the kernel-module-loader-service-type service, but I'm not using that one. I tried anyway, but it seems to not work (lsmod still shows the module)
<iska>stefanc_diff ah, it's not explicitly told, but you can use kernel-arguments to blacklist modules
<stevenroose>Are any special extra packages needed to run LibreOffice? When I install it and try run it, it throws 'terminate called after throwing an instance of 'com::sun::star::lang::WrappedTargetRuntimeException''
<iska>stevenroose can you try launching it inside `guix shell -C libreoffice`?
<abrenon>hi guix
<apteryx>rekado: OK! that gives us some incentive
<apteryx>raghavgururajan: OK! I'll let guix-patches (rather, reviewers :-)) decide then, when I get to send it
<apteryx>can our ungoogled-chromium work with a yubikey (security key) ?
<apteryx>hmm, looks like that old yubikey nano might be blocked by big G:
<apteryx>m=n73qwf,ws9Tlc,e5qFLc,UUJqVe,IZT63,O1Gjze,xUdipf,OTA3Ae,COQbmf,fKUV3e,aurFic,U0aPgd,ZwDk9d,V3dDOb,rCcCxc,m9oV,mI3LFb,bXpTS,WCG2fe,U4Hp0d,O6y8ed,mzzZzc,ZKtqpf,MpJwZc,PrPYRd,NwH0H,OmgaI,lazG7b,XVMNvd,L1AAkb,KUM7Z,Mlhmy,WpP9Yc,s39S4,SeYxhb,duFQFc,lwddkf,gychg,w9hDv,EEDORb,RMhBfe,SdcwHb,aW3pY,pw70Gc,EFQ78c,Ulmmrd,ZfAoz,mdR7q,Z5uLle,I6YDgd,xQtZb,gJzDyc,JNoxi,kWgXee,MI6k7c,kjKdXe,BVgquf,QIhFr,lfpdyf,ovKuL
<apteryx>d,hKSk3e,yDVVkb,hc6Ubd,SpsfSb,KG2eXe,MdUzUe,VwDzFe,tRBqBe,pcKEle,zbML3c,zr1jrb,YTxL4,Ref59c,A7fCU,Uas9Hd,zy0vNb,icmqKf,OgOVNe,LGJfp,qfTGrb,qmdT9,IaeJtb,PrUyhf,zk0ux,spK8ic,MywJR,tDEcYc,pjICDe,J9VQ8d,IoZ2X:4372 The 'googleLegacyAppidSupport' extension is ignored for requests with an 'rp.id' not equal to '9oo91e.qjz9zk'
<apteryx>oof, apologies for that long string
<apteryx>that appears in the console of ungoogled-chromium
<civodul>apteryx: hi! i think YubiKeys present themselves as USB input devices, no?
<civodul>thus they can work with any application, even your IRC client :-)
<gnucode>morning guix!
<apteryx>ACTION didn't know these things expired (yubikeys)
<apteryx>civodul: yes!
<apteryx>it seems it's "working" but rejected by the web application due to being a legacy device
<apteryx>sounds like a sound business model. sell a tiny chip that must costs 1$ to manufacture 50 USD every couple of years
<apteryx>mine is the YubiKey NEO-n, for which their page says "Years in operation: 2014-2016", https://www.yubico.com/products/identifying-your-yubikey/
<stevenroose>iska: it says failed to open display
<mirai_>how do I "recompile" guix? I'm getting an ABI record mismatch when I try to run tests (https://paste.centos.org/view/b1333fa2)
<yarl>stevenroose: try "guix shell -C --share=/tmp/.X11-unix --preserve='^DISPLAY$' libreoffice -- libreoffice"
<civodul>mirai: try "rm gnu/services/*go && make"
<civodul>that's the meaning of "recompilation needed"
<yarl>Hmmm. I'm still trying to figure out how to congigure my environment to have emacs + geiser working with guix sources..
<tricon>yarl: what's your latest issue?
<yarl>tricon: I managed to have geiser launches a guile repl at startup where %load-path is exactly ("~/.config/guix/current/share/guile/site/3.0/" ".guix-home/profile/share/guile/3.0/") and %load-compiled-path is exactly (".guix-home/profile/lib/guile/3.0/ccache/") . (I created a ~/.guile-geiser file and set the appropriate option in geiser-guile customize group). The problem is that if I ",use (guix import texlive)". A
<yarl>bunch of "source file newer than compiled /run/current-system/blablabla" messages are printed. If I ",m (guix import texlive) then %load-path and %compiled-load-path have messed up values from my old (mis/no-)configuration.
<apteryx>civodul: hi! I'm planning to merge #58812 today in its v3 form, if that's fine with you
<apteryx>(guix: shell: Add '--symlink' option.)
<stevenroose>yarl: it gives Authorization required, but no authorization protocol specified
<stevenroose>and still shows Failed to open display too
<bdju>is anyone else having absolutely horrendous wl-clipboard breakage lately? two mornings in a row it's just broken for me and it's especially stubborn this time
<jorge[m]><raghavgururajan> "jorge: Quickest way is to use `..." <- invalid importer
<apteryx>is someone else using Zimbra as a webmail client? It's confused about my timezone in Icecat, yet I have resistFingerprinting=False in about:config
<apteryx>It thinks I'm UTC or somethnig
<apteryx>and even if I configure my correct timezone in its settings, it's still confused
<civodul>apteryx: yes, known problem i've been dealing with for a few years! :-)
<civodul>of course i occasionally have troubles because of that
<civodul>so much fun!
<civodul>one day, someone explained how to tell IceCat not to hide your timezone (it's a privacy feature)
<civodul>but i quickly forgot
<civodul>it might be on help-guix
<jorge[m]>Hola,algun servicio en la nube decentralizada que pueda usar en mi sistema Guix ?
<mekeor[m]>iska: i set up whats called "the perfect setup" in the manual. it works fine when editing a scheme file in my local guix repository clone. but when editing my os declaration or some package declaration from my local custom channel, it does not offer semantic completion suggestions. any idea, why? i use corfu.el as completion-ui
<jorge[m]>Hello, any free software decentralized cloud services that I can use in my Guix system?
<civodul>jorge[m]: ??hola! ??que tipo de servicio necesitas?
<apteryx>civodul: I think I was the one telling you to turn privacy.resistFingerprinting to false in about:config :-)
<apteryx>thing is, it stopped working for me, at leat for Zimbra
<civodul>apteryx: d'oh!
<apteryx>perhaps mozilla has added something new
<apteryx>hopefully more specific
<civodul>resistFingerprintingViaTimezone?
<apteryx>that'd be nice!
<civodul>no no, i'm just guessing what it could be
<civodul>i haven't checked
<apteryx>so your the date on your messages in Zimbra in our current IceCat still displays accordingy to your local time with privacy.resistFingerprinting=false ?
<apteryx>oh, privacy.resistFingerprinting.exemptedDomains would be a better option
<florhizome[m]>how would I recursively insert pairs of directory???s between which I want to copy
<florhizome[m]>so (copy???recursively x y) and I want to give an alist or other list of lists and map source and target files
<yarl>mekeor[m]: I am strugling for some days to get my emacs configured properly (with geiser) to play with guix code. I can't get things working properly, to the point that I am back to use directly guile and trying different GUILE_LOAD_PATH combinations. I am not winning so far...
<apteryx>civodul: time is also wrong in mattermost... grr
<apteryx>I've ruled out the privacy plugins and resistFingerprinting option, so it's something else
<yarl>By the way, I sent a mail concerning a precise question on GUILE_LOAD_PATH on guix, on guix-help. My mail doesn't seem to have arrived yet
<civodul>apteryx: ah, i didn't notice that in mattermost, there's no clear mention of time
<apteryx>was it your first mail to the community? if so, if'll await moderation
<podiki[m]>maybe not related, but I've seen timezone errors in things that use sandboxing, usually exempting the typical timezone location in FHS distros
<podiki[m]>e.g. flatpak doesn't pickup guix timezone by default (you can share with flatpak via an option though)
<apteryx>OK; that's IceCat running on my Guix System, not in a container
<civodul>podiki[m]: you'd have to share /etc/localtime with the container
<yarl>apteryx: on guix-help, yes.
<podiki[m]>I just mean that icecat does use some sandboxing no? e.g. my recent vaapi adventures
<podiki[m]>so they may be sandboxing some processes within the program
<civodul>IceCat has its own thing to not tell its timezone to servers
<apteryx>yarl: in case it helps, when visiting a guix project file the first time in Emacs, it should prompt you to accept some .dir-locals buffer local variables... if you say yes it should magically set things up for you, I believe.
<podiki[m]>but icecat does know the proper time? I saw someone saying this about firefox recently, I forget how they checked (console command within browser?) and the workaround I had was seting TZ explicitly
<apteryx>podiki[m]: yes, if I type some javascript in the console, it returns the date fine
<apteryx>and other sites are OK... not sure what Zimbra/Mattermost do to get my timezone
<apteryx>oh, that's no longer true, I just tried
<rpana>Hello
<apteryx>let currentDate = new Date(); let time = currentDate.getHours() + ":" + currentDate.getMinutes() + ":" + currentDate.getSeconds(); time -> 16:32:41, should have been 11:32:41
<apteryx>so my timezone appears to be globally wrong now in IceCat :-)
<rpana>I have a very basic question I'd like to ask you guys. I am planing on switching from NixOS to Guix, but there is something the bothers me a bit. The thing is I have read (maybe I didn't understand) that in Guix it is not recommended to install all packages declarative on the system config, but rather on the command line?
<unmatched-paren>rpana: not in the system config, but either with the command line, a declarative manifest, or Guix Home :)
<unmatched-paren>either one of the three will install packages in your user profile, so you don't need to have root
<unmatched-paren>(Guix Home is the best of the three, imo ;P)
<rpana>Is the declarative manifest also written in scheme?
<unmatched-paren>yeah
<rpana>Cool!
<unmatched-paren>(specifications->packages (list "foo@version" "bar" "baz"))
<unmatched-paren>so is the guix home config
<rpana>The thing is I really hate to forget the stuff I have installed
<rpana>Thanks for the answer :)
<rpana>Would you have an example config with Guix Home?
<unmatched-paren>here's my home.scm: https://git.sr.ht/~unmatched-paren/conf/tree/root/item/home.scm
<apteryx>yarl: the local buffer variables magic is defined in the guix checkout in the .dir-locals.el file
<unmatched-paren>it lets you manage things like dotfiles and services that run under the user along with packages
<rpana>Top, many thanks
<rpana>This is what I wanted
<jorge[m]><civodul> "jorge: ??hola! ??que tipo de..." <- Decentralizado y que pueda guardar hojas de calculo,documentos,imagenes,ect.
<yarl>apteryx: I know, thanks but this doesn't work so well for me :)
<podiki[m]>apteryx: might try launching browser with TZ set? I think that was what I was seeing in firefox
<mekeor[m]>yarl: do u want to edit the source code of guix?
<yarl>mekeor[m]: Yes:
<yarl>!
<yarl>mekeor[m]: and possibly evaluate it directly in emacs.
<yarl>At the beginning there was this latex package (mathdots) I cannot import (guix import texlive mathdots) : says "failed to download description...". So I tried to enter (in emacs) the module (guix import texlive). and try some functions there directly.
<yarl>But I'm confused.
<yarl>I could with some configuration eval, for example, guix-name. Sometimes, I could not.
<yarl>Less simpler functions is another story.
<yarl>It would be so nice to have somewhere a working configation of someone that uses emacs + geiser to hack on guix.
<yarl>Along with an example of use cases.
<apteryx>podiki[m]: still nope!
<podiki[m]>yarl: the "perfect set up" in the manual should get you started, but admittedly I haven't used gesier much
<yarl>podiki[m]: Thank you but, of course I already read it ;)
<podiki[m]>apteryx: huh. monster strace to see where it fails to get time zone info from? (doesn't sound too fun, but that's all I got)
<podiki[m]>yarl: ah. well just in case thought I'd mention it! would be good to send a report maybe of what is not working/what you want, maybe to help or devel mailing list
<apteryx>hey, that or looking at the 3.5G source? hmm
<yarl>podiki[m]: Right. Already sent a (first) email to guix-help today. I will pursue.
<podiki[m]>yarl: thanks for sending, I'll also be curious (been meaning to get a proper geiser setup going)
<podiki[m]>apteryx: yeah, pick your poison...
<podiki[m]>ACTION away for now
<apteryx>the problem only exists in IceCat not in the original (tried using nix)
<minima>hi, i don't see either the qmk or the menelaus (microscheme) firmware projects packaged under guix - is there any particular alternative that i could/should look at?
<apteryx>podiki[m]: ltrace -e getenv to the rescue! libnspr4.so->getenv("TZ") = "America/Montreal" but libicuuc.so.71->getenv("TZ") = "UTC"
<apteryx>so apparently, something's wrapping TZ to UTC by the time libicuuc.so tries to read it
<apteryx>the only getenv("TZ") in icecat sources is at js/src/vm/DateTime.cpp, in js::DateTimeInfo::internalResyncICUDefaultTimeZone()
<abhicherath[m]>hello! I've been having some trouble with trying to get wayland working on guix system
<abhicherath[m]>I understand from the docs (here https://guix.gnu.org/manual/en/html_node/Desktop-Services.html) that I should configure SDDM
<unmatched-paren>abhicherath[m]: why sddm?
<gnucode>abhicherath[m]: I believe that GDM now supports running wayland.
<unmatched-paren>yup
<abhicherath[m]>unmatched-paren: oh because it says so in the docs
<gnucode>abhicherath[m]: I vote you try out sway. It's super stable!
<unmatched-paren>you're reading manual, you need to read manual/devel
<abhicherath[m]>O
<unmatched-paren>gdm-configuration now has a (wayland? ...) field :)
<unmatched-paren>(wayland? #t) enables running gdm under wayland and adds support for wayland sessions
<abhicherath[m]>cool! uh how do i access that field in the operating system config?
<unmatched-paren>so, you need to use (modify-services ...)
<unmatched-paren>(modify-services %desktop-services ...)
<unmatched-paren>in there, add (gdm-service-type config => ...)
<unmatched-paren>and in there, (gdm-configuration (inherit config) (wayland? #t))
<abhicherath[m]>ahhh I see
<unmatched-paren>abhicherath[m]: a bit like this: https://git.sr.ht/~unmatched-paren/conf/tree/root/item/system.scm#L132
<abhicherath[m]>oh sweet I think that worked
<abhicherath[m]>or atleast its doing the reconfiguring
<abhicherath[m]>is there any way to get nice autocomplete and such for guile and guix outside of emacs?
<abhicherath[m]><gnucode> "abhicherath: I vote you try..." <- one new thing at a time ????
<unmatched-paren>abhicherath[m]: there's a scheme lsp server being developed
<unmatched-paren>but you'd need to manually set it up with neovim lsp, as lspconfig doesn't support it
<unmatched-paren> https://codeberg.org/rgherdt/scheme-lsp-server
<abhicherath[m]>unmatched-paren: thank you!!
<abhicherath[m]>and yes i'll do that
<abhicherath[m]>what's the `=>` syntax do?
<unmatched-paren>abhicherath[m]: it's just a fancy keyword
<unmatched-paren>(SERVICE-TYPE OLD-VALUE => NEW-VALUE
<unmatched-paren>)
<civodul>overdrive1 and kreuzberg are back to work, but pankow and grunewald are offline; any idea what's up with them, rekado?
<abhicherath[m]>cool
<abhicherath[m]>unmatched-paren: ahh ic so service-type has a way of extracting that
<abhicherath[m]>thanks!
<unmatched-paren>abhicherath[m]: nah, it's a macro
<unmatched-paren>the modify-services macro looks like (modify-services SERVICES CLAUSES...)
<abhicherath[m]>hmm i'll have to look into that later, (here's where that LSP will be helpful)
<unmatched-paren>and each CLAUSE is either (delete SERVICE-TYPE) or (SERVICE-TYPE OLD-VALUE => NEW-VALUE)
<abhicherath[m]>ahhh no more frame-tearing thank god :)
<unmatched-paren>abhicherath[m]: have a look at gnu/services.scm (define-macro modify-services ...) if you want to see the definition
<unmatched-paren> https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/services.scm#n313
<abhicherath[m]>huh that's very sensible. So if i wanted sddm instead i'd use modify-services, have (delete gdm-service-type) and (sddm-service-type '() => sddm-configuration ...))
<unmatched-paren>nope, the OLD-VALUE is a symbol which will be bound to the old configuration value in the NEW-VALUE
<unmatched-paren>you'd need to cons (service sddm-service-type (sddm-configuration ...))
<unmatched-paren>onto the modify-services value
<unmatched-paren>but delete gdm-service-type is correct :)
<unmatched-paren>(the configuration forms are in parens)
<podiki[m]>apteryx: nice work! TIL about ltrace, that will be handy (too often I'm grepping through strace logs)
<acrow>Good to be back with guix.
<unmatched-paren>acrow: Welcome back :)
<acrow>unmatched-paren: Thanks.
<lechner>Hi, in a nutshell what are the items removed by 'gc' (under the standard daemon config) that may have to be downloaded again in the next 'system reconfigure', please?
<unmatched-paren>lechner: i think there's a few packages used in the builder for origins
<unmatched-paren>those aren't technically in use unless you're downloading an origin, so they're deleted by guix gc
<lechner>unmatched-paren / forgive my innocence, please. what is an origin?
<unmatched-paren>lechner: ah, have you not made packages before? sorry for assuming :)
<lechner>i have
<lechner>but somehow my download seem more comprehensive
<unmatched-paren>it's the thing that goes in a (source ...)
<lechner>downlloads
<unmatched-paren>usually
<unmatched-paren>(origin (method git-fetch) ...) etc
<unmatched-paren>it has a gexp-compiler, so it's possible to turn it into a derivation (and thus a store output)
<lechner>unmatched-paren / i currently keep derivations and outputs in my daemon config. that stopped the downloaded but may come at a storage price
<lechner>downloads
<lechner>i am thinking about undoing those options
<nckx>sneek: later ask yarl: Has your mail arrived yet? You wrote 'guix-help' twice, instead of help-guix; that might be the reason.
<sneek>Got it.
<rekado>sneek: later tell civodul no idea what???s up with the two other aarch64 nodes; I???m still sick but I hope I can visit the data centre some time later this week.
<sneek>Okay.
<cbaines>are the emails to guix-commits for the main Guix repository slow or broken at the moment?
<unmatched-paren>hmm, i see that too
<unmatched-paren>my "gnu: moreutils: Fix corrupted manual pages." email still doesn't have an ack reply
<minima>how would one go at packaging this https://git.sr.ht/~technomancy/menelaus since the project doesn't seem to have a separate executable - it's purely makefile-driven
<minima>would it need a thin interface - e.g. a standalone command that wraps around 'make upload ...'?
<unmatched-paren>minima: i think you could just install the makefile
<minima>unmatched-paren: do you mean installing as in keeping the makefile in a folder and then using it from that particular folder?
<minima>or as in having something that can be in your path and hence called from anywhere?
<unmatched-paren>use copy-build-system to copy over Makefile, gui.rkt, keyboardio-dvorak.scm, keycodes.scm, menelus.scm, multidvorak.scm, qwerty.scm, test.rkt, usb_keyboard.c, and usb_keyboard.h to #$output/share/menelaus, maybe
<minima>hm i see - but then one has to cd into #$output/share/menelaus to use it i guess?
<unmatched-paren>make -C (guix build menelaus)/share/menelaus
<unmatched-paren>s/(/$(/
<unmatched-paren>make -C $(guix build menelaus)/share/menelaus
<minima>hm, i see - that's very helpful, thanks unmatched-paren
<unmatched-paren>(that looks like a really cool keyboard btw ;P)
<minima>it is!! :)
<minima>well, i suppose the firmware can be used with a family of keyboards - but atreus is probably the main target?
<unmatched-paren>i wouldn't know, but i'd guess that keyboard firmware is necessarily specialised to a single model
<minima>i wonder if there's any other small project that's packaged in guix and that has a similar minimalist makefile-only structure
<unmatched-paren>i don't think so
<unmatched-paren>minima: actually, perhaps you should run ``make build'' before installing those files
<minima>it might be worth to speak to the author and see whether they'd be interested in a patch that adds a thin wrapper - that said i'm very grateful for your tips of course - i might use your tips as a quick solution and then contact the author
<unmatched-paren>oh, looks like you can just use gnu-build-system and replace #:test-target and 'install
<unmatched-paren>and delete configure
<minima>cool
<unmatched-paren>so then the artefacts will all be prebuilt and uploading will be all that's necessary
<minima>i'll give it a go yay!
<unmatched-paren>you'll need to add the avr gcc toolchain, racket, and microscheme to the inputs...
<unmatched-paren>and whatever provides avrdude, too
<unmatched-paren>and then substitute the avrdude call with (search-input-file inputs "bin/avrdude")
<mirai>Can someone help me with figuring out why this service definition hangs Shepherd? (https://paste.centos.org/view/bb614564)
<unmatched-paren>so that someone can just ``make upload -C $(guix build menelaus)'' without installing avrdude
<unmatched-paren>mirai: maybe you could add a #:log-file
<unmatched-paren>and check that
<unmatched-paren>does ``herd restart mympd'' actually hang or just fail?
<mirai>unmatched-paren: I'm trying it out with tests first
<mirai>and it just never reaches the login tty
<apteryx>do other reproduce the problem with latest icecat where it doesn't honor your /etc/timezone?
<unmatched-paren>mirai: sorry, i can't help :(
<apteryx>a site (which sadly uses javascript...) to test is everytimezone.com
<mirai>unmatched-paren: :(
<mirai>I'm really not sure how to debug this thing, as shepherd doesn't talk much when things go south
<unmatched-paren>i can relate
<mirai>the code doesn't seem wrong to me? unless I'm misusing something
<unmatched-paren>i don't see any particular issue either
<apteryx>mirai: quickly, I don't see what's wrong
<apteryx>in these case, I find the built program by guix in the store using 'guix gc -R profile-containing-the-service' | grep shepherd, then open it in emacs with emacs-guix so that I can M-x guix-scheme-mode, then try to run it manually
<apteryx>in guix repl or similar
<apteryx>it's very manual but when it gets to that...
<mirai>Here's the whole thing https://paste.centos.org/view/98c8b498
<podiki[m]>apteryx: can confirm the timezone problem with icecat (on a foreign distro)
<podiki[m]>it shows UTC
<minima>hey unmatched-paren thanks for the additional advice! sorry, i went afk for a bit - i'll post my results here
<nckx>podiki[m]: With resistFingerprinting (or however it's capitalised) off?
<podiki[m]>that was the default (first run) icecat, let me check
<podiki[m]>it was defaulted to true; turning it off showed the right timezone (for that timezone website at least)
<nckx>Huh. That's not the case on Guix today. Used to be!
<podiki[m]>I don't think I've run icecat on this machine or changed settings, resistfingerprinting did show it was true (default); I'll try on my guix system later
<nckx>Probably due to newer IceCat, not Guix per se.
<podiki[m]>resistfingerprinting defaulting to true you mean?
<nckx>That has always been the case, but disabling it would report your timezone to the feds^Website, and now it does not. No matter what the value, even after a browser restart.
<podiki[m]>huh
<vagrantc>maybe it's a daylight savings time glitch
<podiki[m]>I'll compare on a guix box later (or maybe in a container here), perhaps something from being on a foreign distro then?
<nckx>Now, IIRC, resistFingerprinting was not gnupstream's documented method or anything. It was ???discovered??? by a user after being told to ???try toggling some stuff I guess??? by Mark.
<nckx>* quotes may not be ???accurate??? :)
<nckx>So maybe the magic combo has simply changed now.
<nckx>podiki[m]: I had missed the implication that this is Guix's IceCat too. 102.4.0?
<nckx>podiki[m], apteryx: IceCat assumes that /etc/localtime is a symlink to <zoneinfo>/share/zoneinfo/Your/TZ, and bugs out if it isn't. Some GNOME software is similarly broken.
<nckx>Is there a technical reason it is not a symlink on Guix System?
<nckx>Anyway, there's a heavy-weight work-around for those who need it.
<podiki[m]>nckx: yes, guix's icecat for me (on top of Arch on this computer)
<nckx>Yeah, any other distro will have a symlink.
<podiki[m]>indeed does
<podiki[m]>isn't there some TZ_DIR or similar variable?
<nckx>Hm, not sure how that's relevant?
<nckx>There's TZDIR but see previous message.
<podiki[m]>it is TZDIR; well some software will hardcode that to /usr/share... to look timezones, like I found out for flatpak: https://github.com/flatpak/flatpak/issues/4715
<podiki[m]>but yes, different mechanism I guess
<podiki[m]>(add TZ stuff to my list of weird but needed every day stuff I don't understand)
<bdju>I am once again unable to boot, dropped into a guile repl after the second drive couldn't decrypt
<nckx>You'll still have to ???manually??? specify the time zone, so it doesn't win you much as a proper fix.
<nckx>podiki[m]: Heh. I've been there. Things got dark.
<nckx>Anyway, manually invoking TZ=Europe/Brussels icecat also ???works???, it's just not something we can expect of users (and in my usecase it's more tedious than flipping /etc/localtime, but that'll differ).
<nckx>It clearly wants a pretty name, not just the data file :(
<podiki[m]>I think I still have TZ set on my guix machine due to a bug guix had (briefly) where something (libc?) wouldn't read time zone otherwise
<podiki[m]>(and actually is probably why it is set on my Arch machine too, same basic dotfiles and all)
<rekado>apteryx: I???ll probably go to the data centre tomorrow, primarily to kick the aarch64 nodes. Want me to also unplug some SSDs while I???m there?
<bdju>please someone tell me how I skip decrypting this second drive, I am in hell
<bdju>I commented out lines for this drive last month but guix is still trying to mount it for some reason
<rekado>commented annd reconfigured?
<bdju>yes, several times since then
<bdju> https://gitlab.com/bdju/dotfiles/-/commit/160a0ef228a6a0f71130c2c87040f12c2b4e48fd maybe these lines weren't enough
<tricon>bdju: you need to remove them from the device mapper.
<bdju> https://file0.s3kr.it/ff44db321c2b.jpg
<bdju>ah... I see it
<bdju>I don't know how to get ah my config now, though. for some reason the old gen that booted for me last time fails now
<sharlatan>Hi all! Does anyone use geiser in Emacs?
<tricon>sharlatan: i do, though not via the shell. i use the REPL and such features.
<mekeor[m]><minima> "hi, i don't see either the qmk..." <- you can compile and flash your qmk firmware on guix using docker. thats what i do, at least
<bdju>so what do I do? can I edit my system config from a liveusb?
<tricon>bdju: that's what i would do, yes.
<bdju>oh, I tried my oldest generation and I think I'm in
<tricon>bdju: leave it to the Boomer generation to pull through.
<bdju>haha
<mekeor[m]>acrow: welcome back to guix :)
<mekeor[m]>rekado: get well soon! :)
<oriansj>why did the guix signing key change?
<apteryx>rekado: hmm. perhaps I can try to migrate to the two hdds tonight; it'd be nice to make it a one time thing
<apteryx>(all the ssds at once)
<nckx>oriansj: Which?
<oriansj>nckx: for the .sig files
<oriansj>specifically: "https://ftp.gnu.org/gnu/guix/guix-binary-${VERSION}.tar.xz.sig"
<nckx>Oh, I thought you meant recently. ????
<nckx>Releases aren't signed by the same person in perpetuity. Occasionally the person responsible will rotate.
<nckx>The last one was <https://git.savannah.gnu.org/cgit/guix.git/commit/?id=185709221d4cd2f3a06fa3c4777e9e71d15ed09d>.
<oriansj>ok, so how exactly am I supposed to know the keys that are to be trusted to know I am downloading a trusted tarball?
<oriansj>just routinely checkout etc/guix-install.sh from the guix repo?
<nckx>For example.
<oriansj>I guess the old excuse for not having a Guix build server signing key is now gone
<oriansj>hopefully someone is extending guix challenge to check the tarballs on website as well
<nckx>No.