<podiki[m]>I don't think refresh takes that into account, just by package dependencies in general? (e.g. if you update the package version, not the specific rebuild that is done) <civodul>vivien: try "./pre-inst-env guix build libreoffice -n", and see whether that would rebuild the world *vagrantc had to look up -n *vagrantc HEART --long-options <KiranShila[m]>Is there some trick to getting the prometheus-node-exporter service working? I can curl the /metrics endpoint locally but I'm seemingly not able to hit it remotely <clever>KiranShila[m]: check netstat and see what ip its binding to, check iptables rules <KiranShila[m]>Netstat says its bound to `*:9100`, no drop rules in iptables <KiranShila[m]>This is with the default prometheus-node-exporter-config, btw <KiranShila[m]>Yeah something weird was happening with 9100, changing to another port worked fine <gnucode>nckx: I think I found a way to let users easily set up filters without having to use include files. <johnjaye>how does one create an image to burn to an actual sd card for the raspi? <johnjaye>the file gnu/packages/raspberry-pi.scm just has 2 functions that say they create the bootloaders <johnjaye>does a different function use the arm-chainloader thing? <gnucode>nckx: What do you think of opensmtpd-filter-configuration->mixed-text-file ? <yewscion>Hello all, I have a weird question: What is the most canonical way to package a .h file that lives in a Github Gist? <clever>yewscion: you can run `git clone` on a gist's http url <clever>yewscion: so anything that clones from git could be used as a starting point <yewscion>clever: Ah, okay. That makes sense. I was unaware You could use git to clone it. Thanks! <clever>yewscion: you can even `git pull` after cloning, to get changes a user has made, and you get full git history for the changes <clever>the only real difference vs a normal github repo, is that it doesnt allow subdirs or branches <yewscion>Interesting choice by the company, and good to know that under the hood it is just a normal git repo. <johnjaye>clever: ah i didn't realize you were here. i thought i had to go to #raspberrypi <johnjaye>i guess to expand on my other question, are there similarities between nix and guix so that porting your vc4 thing over would be possible? <vivien>OK so now I understand that I won’t repair bash cross-compilation, there are too many things missing in bash. ***puff` is now known as puff
<clever>johnjaye: the biggest question, is how guix handles cross-compilers <clever>johnjaye: once you figure that out, the rest should be fairly simple <yewscion>Is there an equivalent to search-input-file for the store location of an input? All I can find are search-input-file and search-input-directory. <johnjaye>clever: because the x86 to arm64 cross compiler is the main step? <clever>johnjaye: the scm file you linked earlier already has one of those, the problem is the vc4 cross compiler <podiki[m]>yewscion: what do you mean, to get the store path for that input? <podiki[m]>that would just be e.g. #$(this-package-inputs "packagename") <podiki[m]>(just like how output gives the store path for the package output, or other similar gexps) <yewscion>podiki: Does that need to be within a gexp, since it uses #$? I am trying to patch a makefile's hardcoded LDFLAGS. <podiki[m]>yes, as in you have arguments in package def, using #~(... and then use #$ <yewscion>podiki: Thanks. I used to use (let ((out (assoc-ref outputs "out"))) …), still learning how to use gexps. The manual is nice, but I'm struggling for some reason. <podiki[m]>I always look for examples, you'll find plenty now in the code <podiki[m]>once you get the hang of it it is pretty nice, much easier to type out :) <yewscion>I'm sure! It seems very efficient, but also abstract. Kind of like Emacs. I'm excited to add them to my toolbox. <podiki[m]>welcome! I still get stuck on trickier ones, but getting used to the stuff you do a lot like that type of patching for input paths is not bad <podiki[m]>the manual explains it pretty well, but for practical purposes I just think of them as you mentioned: when you need store paths <podiki[m]>sneek: later tell apteryx I tried buildroot and I think it mostly works but fails at the end of a make due to broken urls (for the random configs I tried at least) <sneek>apteryx, you have 1 message! <sneek>apteryx, podiki[m] says: I tried buildroot and I think it mostly works but fails at the end of a make due to broken urls (for the random configs I tried at least) <cnx>hi, how do i disable gdm but keep other %desktop-services <apteryx>rekado_: uh, 'guix build r' produces an empty output <apteryx>du -sh -> 0 /gnu/store/z9x27i7nh1yj63y34vszb4sk5z981xyl-r-4.2.1 ***scisssssssors is now known as unwox
***unwox is now known as scisssssssors
***scisssssssors is now known as unwox
***unwox is now known as scisssssssors
***scisssssssors is now known as unwox
***daviid` is now known as daviid
***sneek_ is now known as sneek
***rgherdt_ is now known as rgherdt
<rekado_>apteryx: r just propagates other packages. r-minimal is one of the propagated packages. <yarl>guix shell -C --preserve='^DISPLAY$' emacs -- emacs <yarl>Display :0 unavailable, simulating -nw <yarl>In the guix manual, 8.1, there is an example with DISPLAY preservation. <yarl>My command should work, right? <GNUtoo>My patches in android-make-stub were merged, so I'll now try to update it in Guix <GNUtoo>The big question is if we need to wait for a new release or not <PotentialUser-65>i'm new here and i'm getting new hardware (frame.work). So i'm planning to give guix a try. How is the hardware support for the frame.work laptop? <GNUtoo>Ah it's the same person who maintain android-make-stub that packaged it to Guix <GNUtoo>So maybe it would be possible to get a release <GNUtoo>PotentialUser-65: do you have a link with the specs of the frame.work laptop ? <GNUtoo>ahh maybe I've a better solution, there is a build-inputs <GNUtoo>hmm that's not accessible from the package apparently <unwox>PotentialUser-65: usually the biggest concern is wifi card driver. but without exact specs it's hard to say if it's going to be a problem <GNUtoo>Sometimes sound doesn't work some recent computers too <GNUtoo>PotentialUser-65: a way to know if it works with FSDG compatible distributoins would be to find a way to ask the makers of that laptop <GNUtoo>It's often possible to do that with computers being crowdfunded <GNUtoo>Though we probably need to suggest them something easy to do, like boot a Trisquel live system <unwox>also try asking on framework forum <GNUtoo>The issue is that to know you really need to use an FSDG compliant distribution to do tests <GNUtoo>Or debian without nonfree (assuming they don't ship sound-open-firmware) <GNUtoo>Another option would be to make a Guix image that can go on an usb key with a browser and h-client on it to help people submit entries to h-node <GNUtoo>(and it would also contain test files for sound, some games that use 3D acceleration but don't use a lot of RAM (like extreme tux racer) etc ***unwox is now known as scisssssssors
***scisssssssors is now known as unwox
<GNUtoo>Else for recent laptops there is Puri.sm laptops <GNUtoo>While (as all recent intel computers) they contain nonfree software in the boot software, the distribution is FSDG compliant so it should work in Guix, and if it doesn't that can probably be fixed in Guix <GNUtoo>Is there a way to know where a dependency is being added in a package? <GNUtoo>There are some implicit dependencies, that for instance come with the build system being used <GNUtoo>"guix graph -t bag <package-name>" lists the dependencies I'm interested in but it doesn't tell if they are in native-inputs, inputs, propagated-inputs, etc <jpoiret>build system dependencies are implicit and are neither native-inputs, inputs or propagated-inputs <GNUtoo>oh ok, how can I change them then? <jpoiret>you'd need to modify the build system i think <GNUtoo>Basically I had issues in my guix.scm and the issue came from android-make-stub that had too much limitations <GNUtoo>So I fixed some of the limitations of android-make-stub and the code was merged but there is no new release yet <GNUtoo>So I'm a bit lost on what to do here, if there was a new release I could send a patch to update android-make-stub in Guix <GNUtoo>But here I'm in a situation where it's way easier to workaround in my guix.scm <GNUtoo>Like can I justify upgrading to a git version of android-make-stub to fix stuff outside of Guix? <GNUtoo>But maybe as you pointed out I could try to modify the build system in my guix.scm somehow but I'm not sure if it's possible or not <GNUtoo>The background issue is that I maintain a library that is supposed to work both on Android and GNU/Linux, and it uses autools and Android.mk, and I use a guix.scm for build tests before pushing commits <GNUtoo>This is why I need to use android-make-stub: to test building with the Android.mk as well, and it worked well so far but it was limited to 1 target only (I could build 1 library or 1 program, not 2 things) <GNUtoo>And I fixed that limitation in android-make-stub but now I need to find how to use that in a guix.scm somehow <oriansj>sneek: later tell rekado that the bootstrappable.org cert expired <yarl>Sorry for asking again : `guix shell -C --preserve='^DISPLAY$' emacs -- emacs` should work, right? <yarl>It returns "Display :0 unavailable, simulating -nw" <yarl>In the guix manual, "8.1 Invoking guix shell", there is an example with DISPLAY preservation and --container. <yarl>Is this supposed to work only on guix system installation, not on foreign distro? <GNUtoo>There might be examples in other places, I need to look <GNUtoo>Note that I've never used commands like the one you're mentioning, but I probably saw one in a blog post <GNUtoo>Ah the blog post used guix environment <GNUtoo>guix environment -C -N --share=$XAUTHORITY --share=/tmp/.X11-unix --share=/dev/shm --expose=/etc/machine-id --expose=~ --expose=$PROFILE/lib=/lib --expose=$PROFILE/lib=/lib64 --share=/dev/kvm -m some-manifest.scm -- env XAUTHORITY=$XAUTHORITY DISPLAY=$DISPLAY bash <GNUtoo>The blog post was to install some software that was probably non-free on top of Guix, though the instructions probably work for 100% free software too <GNUtoo>You just need to not download the software the blog post is talking about <yarl>Hmm, thank you GNUtoo I'll try this! <GNUtoo>For free toolchains for building Android software there is the Android rebuild project, even if it contains some binary toolchain, at least everything is free software <GNUtoo>AOSP might contain firmware though but that probably applies only to building images (which bring in nonfree software for some targets) <GNUtoo>We probably need to make SDK again in Replicant at some point <GNUtoo>PureOS might also have a toolchain since it's based on Debian <ardon>Hi Guix! I'm encountering this issue when trying to build this https://bpa.st/NEXA Python package https://bpa.st/LZRQ . Any hints? From what I gathered on the web, I need setuptools and wheel, which I already have as inputs <GNUtoo>PotentialUser-65: the usual workaround is to change the WiFi card or add an USB one <GNUtoo>If you change the internal WiFi card you need to check that the BIOS/UEFI doesn't prevent you from doing that <GNUtoo>some computers refuse to boot when you do that <yarl>The answer is (for who is interested) `guix shell -C --share=/tmp/.X11-unix --preserve='^DISPLAY$' emacs -- emacs` <yarl>Thanks you GNUtoo, I think you answered me as a side effect. <GNUtoo>Thinkpenguin even has M2 cards, though you also need to see which type of M2 are compatible <GNUtoo>The thing to really check carefully is the sound (open sournd firmware has signed Intel frmwares for instance), and the GPU (if you've got a black screen, the laptop is not very useful) <GNUtoo>ardon: python install fails, here it's a problem with the arguments <fnstudio>hi, i'm trying to replace my texlive package with a selection based on texlive-base plus a minimum set of deps <fnstudio>after the switch i've started receiving this error 'Cannot find mktex.opt; check your installation.' <fnstudio>that's if i run 'mktexmf --all', i should have said <fnstudio>any idea if there's any obvious dependency that i want to have, other than texlive-base, in order to compile a XeLaTeX file? <fnstudio>the mktexmf error might be a red herring... <GNUtoo>ardon: error: "option --single-version-externally-managed not recognized" is the real error <fnstudio>(re my xelatex error, it might be due to the fact that $GUIX_TEXMF is not defined in my system?) <efraim>oops, tried to allocate too much memory while building sbcl on powerpc ***wielaard is now known as mjw
<GNUtoo>Is there still time to submit a lightning talk? <GNUtoo>And if so, can lightning talks be really short? <civodul>GNUtoo: yes! you can submit a lightning or non-lightning talk, or propose a brainstorming or hacking session <civodul>we're very much interested in proposals! <efraim>What's the new way to refer to %build-inputs for configure-flags? <efraim>no, it looks like I was just messing up that part of the code <civodul>efraim: you can use #$(this-package-input ...) <munksgaard>civodul: Will any of the talks be livestreamed/viewable afterwards? <civodul>munksgaard: we're working to make sure we can record them, and perhaps we'll livestream too if we can manage it <efraim>I think I'm doing it wrong. I'm trying (if (not ,(this-package-input "valgrind")) `("foo") '()) but (this-package...) doesn't return #t, it returns a package <munksgaard>civodul: Sounds great! I would've loved to attend, but won't be able to this time around. <GNUtoo>civodul: thanks, some people wanted me to explain how Replicant used Guix, but since it's very short it a blog post would be too long, but a very short lightning talk would be a good idea <apteryx>interesting; I just learned about the systemd-related coredumpctl command <apteryx>fnstudio: try installing texlive-bin in your user profile <efraim>I made it there in the end, valgrind is always a native-input, so I needed this-package-native-input <GNUtoo>efraim: there are several valgrind I think, let me look <efraim>GNUtoo: there's 3. I'm checking in the configure-flags for the existance of a valgrind input since I already limited adding it to native-inputs based on being a supported system <GNUtoo>And that worked, so I also somehow also used native-inputs but I still have the old syntax <GNUtoo>In my case valgrind is used during tests (make check) <GNUtoo>I also still have the workaround for the stripped ld.so to be compatible with older Guix version, I'll probably get rid of it at some point <gnucode>can someone tell cbaines thanks for me? I have been using his us based substitute server, and it's pretty awesome! <nckx>They're connected (possibly through a bouncer) so I guess they'll see your message. You could also ask sneek ('sneek: later tell rms: this doesn't look very unixy') to relay messages. *nckx takes lookings at opensmtpd-filter-configuration->mixed-text-file <apteryx>which package provides the 'gio' command? <GNUtoo>apteryx: in Parabola gio is part of glib2 <reproducible>we're very much interested in proposals! // civodul: only from people who are able to come, right? <apteryx>rekado_: I barely understand R; do you think you'd be able to assess bug#50841: rstan needs gfortran 9+ ? Seems an actual problem and there's a suggested fix <nckx>gnucode: I guess I'm a bit lost. What I'd naturally write, I think, would be the (mixed-text-file "dkimsign" ...) snippet. I don't fully understand what the surrounding complexity achieves. *nckx also wonders why lines numbers don't show up in that paste. <civodul>reproducible: yup, it's an in-person event! (perhaps with livestream if we manage) <efraim>I can't find what valgrind is gating <efraim>I know libunwind is gating gstreamer <efraim>valgrind is gating 758 packages and I can't figure out what the big one is <gnucode>nckx: I hadn't thought of allowing opensmtpd-filter-configuration-proc to accept a value of mixed-text-file directly. That's genious! <nckx>Genious was the last thing I was attempting but I'll take it. *nckx hacks up their third lung of the day & throws it to the cat. Nap time. Bye-o. Thanks for working on this! <dirtcastle>(correct me if I'm wrong) In guixOS, generations will be made and every time we reconfigure. if the packages break and we can't boot, the bootloader will have options to boot from previous generations. can I have that feature when guix is installed on top of foreign distro. <civodul>(nitpick: it's called "Guix System" :-)) <cbaines>gnucode, that's good ;) You might be the only one using it though, in the last week it's only served ~20 nars <dirtcastle>civodul: do I have that feature on foreign distro is my question. <GNUtoo>dirtcastle: if you use guix to install packages on top of a foreign distribution, you can learn more about guix package --list-generations <GNUtoo>I'm not sure how it works as I didn't really use that feature of rolling back packages, though the manual probably has infos on that if you look for generations or Generations inside <GNUtoo>Also note that on a foreign distributoin Guix can also do a lot more things than just installing packages, for instance you can manage Guix servers/vm, create images for various targets (VMs, real hardware), manage your home (with guix home), create tarball releases of software etc *GNUtoo only uses a subset of all the features <gnucode>well I am coding on opensmtpd-records. <apteryx>it actually works, albeit poorly compared to ungoogled-chromium <dirtcastle>GNUtoo: I'm aware of these stuff. I read the manual. Thanks! <reproducible>apteryx: it gives me the error "It looks like you're using a browser the is not supported" <fnstudio>hey apteryx let me try with texlive-bin, thanks! <fiesh>reproducible: it actually works, you just need to make icecat say it's firefox <fiesh>I can't recall the plugin's name on the top of my head, and I don't have my laptop here... <fnstudio>hm it seems that it didn't help (texlive-bin) <gnucode>reproducible: i closed out of the chat room ... I was using ungoogled-chromium to log in actually. <fnstudio>apteryx: anything else you think i should try? <gnucode>I'll open up the chat room again. guix install ungoogled-chromium; maybe I'll see you there. <GNUtoo>dirtcastle: I was pointing it because it's not necessarily obvious as different part have different support for rolling back <GNUtoo>For instance I probably use it a lot more to do build tests than installing actual packages <GNUtoo>And I only install packages when they are either broken in Parabola i686 or they are not available there <podiki[m]>cbaines: oh, didn't know you had setup a US mirror; I see it slightly faster for me than bordeaux regular: 11.3 vs 11.2MB/s <GNUtoo>Some people also probably seem to not use guix package but guix home instead, so it really depends <podiki[m]>cbaines: from east coast US, both rates seems close to what I'm maxing out on this wifi right now, but will try on a wired connection later <cbaines>podiki[m], cool, and yeah, it would be interesting to hear the results when you've got a wired connection <gnucode>reproducible: I am having problems with the jitsi room now as well. Maybe we'll try again another day eh? ***civodul` is now known as civodul
<gnucode>reproducible: "inshallah"? what does that mean? <fnstudio>hi anyone knows what GUIX_TEXMF should be set to? context: guix on a foreign distro, i'm trying to replace texlive with texlive-base <fnstudio>i'm running into an error that i think might be related to GUIX_TEXMF not being set ***unwox is now known as scisssssssors
<apteryx>reproducible: weird, works for me. perhaps you have to turn some webrtc stuff on in about:config <apteryx>fnstudio: did installing texlive-bin not cause GUIX_TEXMF to be set? <fnstudio>apteryx: should i mention that i'm on a foreign distro? <reproducible>gnucode: its a very common arabic word that basically means "if its meant to happen it will happen" <fnstudio>apteryx: sorry, i had tried opening a new shell, but now i tried with a new login shell and the var is there <apteryx>fnstudio: glad you could figure it out <reproducible>on a fresh install, i'm trying to create a pdf from a tex file with pdflatex, but i get an error that it doesn't find pdflatex.fmt. i have texlive-tiny, texlive-pdftex, and texlive-pdfx installed. anyone know what i'm missing? <fnstudio>hey reproducible i've had a similar problem today <fnstudio>in my case i had to install texlive-bin, then i had to make sure the var GUIX_TEXMF was defined in my env <fnstudio>and finally run this command: fmtutil-sys --all <fnstudio>reproducible: to see whether this may apply to you, i'd start checking if GUIX_TEXMF shows up in your env? <reproducible>hmm but i ran fmtutil-sys --all which warned "you seem to have not formats definted in fmtutil.cnf files! <fnstudio>the env where you run fmutil-sys is the one where the var shows up? <two[m]><reproducible> "on a fresh install, i'm trying..." <- i can create a pdf with just texlive-base <reproducible>am i suppose to generate a pdflatex.fmt file somehow? perhaps thats something I did in the past but am forgetting now <fnstudio>reproducible: do you have texlive-bin installed as well? <podiki[m]>apteryx: I forgot to mention in the email about fhs containers I just set that for your buildroot project I think I needed to add /lib/nss to ld.so.conf since we put our nss libraries in there for some reason <apteryx>hm, our gtk4 package is borked, at least from pkg-config's point of view: Package 'gtk4' requires 'pango >= 1.50.0' but version of pango is 1.48.10 <apteryx>re gtk@4, actually it propagates a recent version of pango enough, via pango-next. perhaps there are propagation conflicts <yarl>How do you people configure geiser for using with guix? specifically geiser-guile-binary? <apteryx>some geiser load path related config is automated in the .dir-locals.el file of Guix <apteryx>my geiser-guile-binary is set to ("guile" "--no-auto-compile") <yarl>apteryx: How do you tell that to emacs in init.el please? <yarl>I ended up with an outdated guile then set it to "~/.guix-extra-profile/dev/dev/bin/guile" something like that. <yarl>The problem is if I try to run emacs in a guix shell, using the same emacs config. The thing is messed up. <yarl>I don't know how geiser got it's default value. <gnucode>nckx: that's a link to a video update of why opensmtpd-filter-configuration-proc should be a string or list of strings and/or gexps instead of just being a mixed-text-file. What do you think? <nckx>sneek: later tell gnucode: I didn't actually disagree, but 'a [simple] list of strings/gexps' wasn't what I took from your earlier paste. <peterpolidoro>hi. I am trying to fix a guix package that references other guix packages using environment variables. is the best practice to set those environment variables to the absolute paths in the store in the main package definition? or should I be copying files or references to those packages into the main package directory structure and set the <peterpolidoro>environment variables to local paths within the main package directory structure? <podiki[m]>cbaines: on my wired connection I see a factor of 2 increase :) 36.5 vs 18.9 MB/s in favor of the US mirror for me <peterpolidoro>the package I am trying to fix uses native-search-paths set to directories relative to the main guix package <peterpolidoro>but the files are in other packages outside of the main guix package <efraim>I can't adjust openmpi and still keep the hash on unaffected systems so I'm going to have to just move stuff around to make more sense after the adjustments <peterpolidoro>are native-search-paths always relative to the package being defined? <peterpolidoro>if so, what should I use instead to reference directories in other packages? <sneek>rekado_, you have 1 message! <sneek>rekado_, oriansj says: that the bootstrappable.org cert expired <rekado_>oriansj: fixed, thanks for letting me know ***Dynom_ is now known as Guest9265
***mark_ is now known as mjw
<podiki[m]>yup just did, though I don't think my webclient could put the correct reply message id <podiki[m]>but should have gotten an email from me to you and list, with that subject line <cbaines>no worries, if it's on the mailing list, it'll be a little harder to lose :) <podiki[m]>cbaines: also made me realize how much slower the main ci.guix.gnu.org server is for me, I should prioritize bordeaux <dirtcastle>GNUtoo: noted. I'll look into the details. thanks for your time. <fnstudio>hm, i'm extremely close to having a minimal texlive env to compile a tex file of mine - i think there's only one more challenge... i was planning to use the latex package pdfpages <fnstudio>i can see pdfpages packaged under guix already <fnstudio>however if i use it, latex will then fail saying '! LaTeX Error: File `pdflscape.sty' not found.' <fnstudio>and pdflscape doesn't seem to be available under guix <fnstudio>i tried with 'guix import texlive pdflscape' but that also failed <fnstudio>somehow, this pdflscape seems to be a pdfpages dependency (as complaints re pdflscape only appear when pdfpages is added) <fnstudio>this might be a bug in texlive-latex-pdfpages, if i may dare, as pdflscape (as well as a few other packages) are listed as required in the pdfpages official docs <fnstudio>but these deps don't seem to be mentioned in 'guix edit texlive-latex-pdfpages' <mmorkoz12[m]>Hello everyone. Just wondering if someone can direct me on a service script for guix for sysvinit if already available. Thanks <dirtcastle>GNUtoo: I was thinking maybe guix home doesn't need to show list of generations in bootloader. because if my host system is working, I can login into my user account, even if an update broke xorg, I'll still have tty and I can use time machine or switch generation. but if my host system breaks, then that is bad. that is where guix's power is required. <weidtn>Hey. is there any way to softlink python3 to python in the /gnu/store ? alias is not enough i think <weidtn>the answer is python-wrapper. just 1 more minute of searching. sorry <jab>well my HDD and RAM just came in the mail. Guess I am a little closer to having a guix server. <jab>Though I could just have bought an olimex device and only spent $40. <rekado_>fnstudio: all the TeX live packages with three components in their name are old, so it’s plausible that they are terribly incomplete <rekado_>mmorkoz12[m]: do you mean an init script for running the guix-daemon? <jab>I bought a cheap old and used Dell Optiplex 7020. 4TB HDD, 16 Gigs RAM <rekado_>mmorkoz12[m]: if so, that’s included in the Guix source repo under etc/. <rekado_>jab: I see. I imagined some aarch64 device :) <jab>rekado_: Have you ever used the olimex stuff? <jab>apteryx: right? Do you want me to ship one your way? <jab>also any advice on how to HARDEN my guix system server would be a great help! <jab>rekado_: what did you think of those devices? And if you want to use guix system, then you probably want 64 bit arm right? <apteryx>but no, I'm well stocked at the moment, thank you :-) <apteryx>should we make our 'inkscape' variable fixed, keep inkscape-next for the updates? 2k packages depend on it, through dblatex <rekado_>jab: I bought the rockpi S to use it as a MIDI piano, but it just didn’t have enough RAM to play the better sounding pianos. <rekado_>I’m using the rockpro64 as a little home server to replace my aging i686 netbook. <rekado_>it’s better suited for the job of running mpd and serving media files; the optical audio interface is also nice (integrated audio circuits in small devices are terribly noisy). <rekado_>haven’t reserved enough time to fully set up the rockpro64, but it’s currently running Guix System (off a USB stick…) and doing just fine. <jab>apteryx: I'm resourceful. :) I could probably ship out 10+ of these devices this month. also I live near a big university. They get rid of old computers all the time. <jab>rekado_: rockpro64 from olimex? it has what...2GB of RAM? <jab>also how "free/open" of a device is the rockpro64 as sold by olimex? Does it require some fireware blobs <vagrantc>jab: 4gb of ram ... ethernet, usb, pcie, microsd, serial console all definitely work without binary blobs ... pretty sure hdmi too, but i haven't tested that as much *vagrantc forgets if there's wifi, but if there is, probably requires non-free firmware <jab>vagrantc: ok. cool cool. would it pass RYF certification? That sounds like an ideal play with this simple computer type deal. <vagrantc>jab: i'm no expert of RYF process ... but seems like a plausible maybe? :) <dirtcastle>since the guix upgrade was too slow, I tried guix package --no-substitutes. but it downloads tar.gz files which is exactly the same size of binaries? I expected it to download the source code and compile locally <vagrantc>jab: the boot firmware and bootloader is buildable in guix, the linux-libre kernel in guix worked last i tried, though it's been a while <mmorkoz12[m]>rekado_: thanks will have a look. Wondering why the installer script under slack do not rekon sysvinit <jab>vagrantc: ok you probably convinced me! <roptat>I'm translating some more tonight and I'm having a bit of trouble figuring out how to translate "honor" as in "honor an environment variable" <jab>roptat: what are you translating english to? <roptat>is "honor" used in the same sense as "honor a contract" or is there a slight difference in meaning? <vagrantc>jab: i managed to even get sata booting working, though rekado had issues with that <jab>also Facebook just released an AI that does translating for you...I think they open souced it...maybe <jab>vagrantc: sata lets you use a big and faster HHD or SSD? <vagrantc>heh. rekado added features to support rockpro64 in 2013, years before the rockpro64 existed. <roptat>oh maybe "respect" is synonymous? <roptat>I feel like "honorer" in French isn't the right term <vagrantc>jab: looks like i last tested it in april <jpoiret>i feel like "honor an environment variable" without specifying what kind of "contract" it is honoring would be vague anyways in the original context <vagrantc>jab: yup, i've got an SSD plugged into mine <roptat>when explaining the search paths: "the @code{ASPELL_DICT_DIR} variable (honored by the GNU Aspell spell checker)" <vagrantc>jab: might also be able to get an NVMe for even faster access <jab>what!!? That sounds crazy awesome! <vagrantc>jab: untested, but there's support in u-boot for booting off of NVMe <jpoiret>i feel like "prise en compte par" would sound the most natural <vagrantc>or rather, i haven't tested ... presumably someone did <GNUtoo>I finally managed to fix it: (1) my code was merged in android-make-stub, (2) I used package transformations to use the latest commit <GNUtoo>(define with-fixed-android-make-stub (options->transformation '((with-commit . "android-make-stub=4bc0e068c32fe525741da00af064ddc079de7e4b")))) <GNUtoo>(with-fixed-android-make-stub libsamsung-ipc-gcc-android) <GNUtoo>Both are sufficient to override the android-make-stub from the build system <GNUtoo>(libsamsung-ipc-gcc-android is a package) <jpoiret>GNUtoo: Guix malleability strikes again! a victory for hackers around the globe <GNUtoo>It's a bit like git, it removes conflicts <GNUtoo>(here there was a conflict of interest between Replicant's interest and Guix's interest and being able to override made that conflict disapear) <fnstudio>rekado_: oh i see re 3-word texlive packages, good to know <mlan>the beginning I installed using VMware guix.iso and I set up guix follow the book directly. is my way right? <roptat>GNUtoo, any interest from replicant to move to a newer android version? <roptat>is it too complex, or is there some freedom issues? *GNUtoo is working on that <GNUtoo>Especially because the kernel is based on upstream linux instead of vendor forks of linux <roptat>ok. as you might know I have a wip soong-build-system for newer android versions that you may want to try :) <roptat>it won't help that much I think, because right now it builds the SDK tools (adb, fastboot, aapt, and friends) <roptat>I'm trying to make our clang easier to work with in the context of cross-builds, which should help build some android system libraries, probably <roptat>and I'm a bit stuck with Kotlin (required for gradle and all the android apps eventually) <GNUtoo>Android is indeed moving to soong <GNUtoo>Though Android.mk still work in Replicant 11 at least <weidtn>Is there any way to make the python-ipykernel use "python" instead of "python3"? <lilyp>weidtn: we dropped python2 some time ago, you might want to go back with the time machine to find the old packages <weidtn>Ah no. I mean python v3. Just the executable with the name python not python3. (python-wrapper) <mlan>the guix team can help me and I have a mistake in the guix builder. <mlan>roptat, I'm zero in on the guix builder. I started using VMware guix.iso, next, I installed the guix package, but there was a problem with the error failed to load, what does that mean? <roptat>I don't know, what's the exact command you ran, and what's the exact error message? <lilyp>a cookie to the person who can tell me why gtk4 needs rest (the package) <lilyp>use sneek, because /me → zzz <mlan>roptat, errors written guix package: warning: failed to load '(cuirass base)' <roptat>do you maybe have cuirass installed in your profile? <mlan>roptat, what is the meaning?..sorry i am weak english <roptat>mlan, which language(s) do you speak better than English? <roptat>maybe you modified GUILE_LOAD_PATH and/or GUILE_LOAD_COMPILED_PATH? <mlan>roptat, I used to speak Malay. do I need to set up on emacs comments? to type GUILE_LOAD_PATH <roptat>how is emacs related to your problem? <mlan>roptat,when I type guix package --export-manifest, the error "guix package: warning: failed to load '(cuirass base)'" appears <roptat>did you install cuirass or change GUILE_LOAD_PATH? <mlan>roptat, before or after guix package -i cuirass still exists "guix package: warning: failed to load '(cuirass base)'" ***dthompso` is now known as dthompson
<mlan>roptat, or I made a mistake installing guix startup? effect have error massage <yewscion>Hello Guix! Can someone explain how to use home-bash-extension in my guix-home profile? I'm trying to add something after the environment variables in my .bash_profile, but every way I've tried to add that field to the home-bash-service-type causes a pattern matching error. <apteryx>rekado_: for your info, /gnu/store/9zwmhcwrac997rh0vb9cbygd2wfv18jk-lv2-mda-piano-0.0.2.drv', `/gnu/store/ccn2y93626vg84prkqfqb0fx6ivxj0pg-ams-lv2-1.2.2.drv', `/gnu/store/wfig2irafyd08jj58svwxc7bg5hi6pdb-lv2-mda-epiano-0.0.2.drv' failed (master) <apteryx>I'm guessing these must be yours :-) <civodul>yewscion: we need to add examples to the manual! <yewscion>Follow up question: Do You have an example of combining home-bash-configuration and home-bash-extension in the same service? Or are they meant to be their own services. <yewscion>I currently am defining it like (service home-bash-service-type (home-bash-configuration …) (home-bash-extension …)) <civodul>yewscion: a service always takes a single "configuration" value <civodul>so it's (service home-bash-service-type (home-bash-configuration ...)) <civodul>but then, you can have a separate service that extends home-bash-service-type with a home-bash-extension value <civodul>that's what the 'simple-service' snippet above does <yewscion>civodul: Ahhh, okay. That's the bit I was missing. <yewscion>The reconfigure worked now that I've separated the two configuration values into their own services, but unfortunately (even though the manual currently says extensions are added "to the end" of the files) the extensions seem to be added before the declared environment variables. <maximed>civodul: cannot connect to web.fdn.fr (in Firefox), 'this website possibly doesn't support the TLS 1.2-protocol, the minimal version supported by Firefox' <weidtn>I'm new to git over mail. Did I submit it in the wrong place? <yewscion>Is there another mechanism I am missing for adding something to .bash_profile after the environment variables declared in the home configuration file? <maximed>weidtn: right place, but the patch / commiters & reviewers with time ratio is rather high. <maximed>FWIW, you can review other patches yourself to help out. <weidtn>I need more experience in this. I just learned how to derive my package from another one with inheritance <maximed>And after sufficient time (to build trust), become one of the committers (who can push reviewed patches). <civodul>hmm we have substitutes for git 2.37.0, even though it's not in master yet <maximed>jpoiret: maybe: 'honor an environment variable' -> 'preserve an environment variable' (in the original English)? <jpoiret>hmmm, i understood that more in the sense of "makes use of that environment variable" <maximed>peterpolidoro: 'native-search-paths' and 'search-paths' don't care about which package <maximed>By definition, the environment value (set by the search path) consists of all (subdirectories of the store items of the) packages in the profile, matching the criteria. <maximed>peterpolidoro: I don't understand what you are trying to do, but there's 'substitute*', 'search-input-file', 'search-input-directory' and 'dirname', which could perhaps be used to eliminate the environment variable, depending on the circumstances. <maximed>jpoiret: Is this about "guix shell --preserve='^DISPLAY$'"? <maximed>'interpreted by OpenSSL'? 'used by OpenSSL'? 'supported by OpenSSL'?, 'requested by OpenSSL'? <maximed>FWIW, 'prise en compte' sounds good to me in my inexpert opinion. <jpoiret>right, and in the english case, i think i've seen '{env var} honored by X' for it to be almost idiomatic (to me) <maximed>I find it an awkward construction myself. <maximed>Looking up 'honoured' in a dictionary: Respected, having received honour <rekado_>apteryx: thanks for letting me know. Looks like lvtk no longer provides ttl2c. <maximed>What does honour, respect, etc., have to do with environment variables? <maximed>It's just a computer program, so value judgements don't seem appropriate here to me. <podiki[m]>"honor" something in this context makes sense to me (in English) <podiki[m]>it is not a value judgment so much as a statement of the rules something follows, in this case <podiki[m]>in any event, something I've seen frequently in this type of documentation and sounds fine <yarl>"utilisée par" or "lue par" sounds good to. <yarl>Do you need some help to translate things to french? <yarl>I think I might do that.