<str1ngs>can someone test this please. install qttools and run qtpaths . and confirm they get this error please. libz.so.1: cannot open shared object file: No such file or directory <str1ngs>I'm on commit: a3e6cf987684cc7476b6ee544091c464378cbcd9 but it might accure on 1.0 as well <nckx>str1ngs: Nope. ~ λ guix environment --pure --ad-hoc qttools → λ qtpaths | Usage: qtpaths [options] [name] | … <nckx>Same in an impure environment. <str1ngs>thank you for testing, what version of guix please? <nckx>Hmm, let me calculate that… <nckx>the equivalent of upstream 26fad5b7dfffef5722b7efb0234a94a0c2b6679c. <nckx>Plus and minus some stuff but none of them even near the qt graph. <str1ngs>hmm I'm wonder if I should pull . or it's related to foreign distro <nckx>It ‘feels’ very much like a foreign distro problem. <str1ngs>when does patch elf get run? on install? <str1ngs>well pure environment works okay, so that's a start <reepca>Has anyone else had luck with getting emacs to actually load dynamic modules? I can get it to load one, run emacs_module_init and everything, but it seems like none of the module api functions actually do anything. Nothing gets defined, and calling elisp functions doesn't do anything <str1ngs>reepca: writing your own modules you mean? <reepca>I see that current emacs is built with --with-modules, so it should be working... <reepca>I did some print-statement debugging, and all of the module api functions are being run, but nothing appears to change inside emacs. No new functions become available. <str1ngs>reepca: first check with . C-h v system-configuration-features and you see MODULES <nckx>str1ngs: patchelf, the tool? Never? <str1ngs>reepca: it covers both C and Go lang <str1ngs>one common error is the so file needs to be the name of the module. and you need to require the so file as module name <reepca>str1ngs: followed the steps, at "Testing C module with Emacs" I get: "Symbol's function definition is void: cmodule-version" <donofrio>pkill9, yes, cowork thought it required a real kernel and I'm like nope just a demon <reepca>could someone else attempt to use emacs modules with guix's emacs? <str1ngs>reepca: did you try with emacs -Q -L ./lib -batch -l cmodule --eval "(message (cmodule-version))" <reepca>str1ngs: yes, I used that command only with -L ./lib changed to -L . (I put the .so into the same directory as the source) <crab2>How do I remove a package that I installed with guix package --install-from-file ? <str1ngs>I can't test with guix emacs right now <sirgazil>Should I expect the graphical installer to generate an configuration file that includes %base-services by default just as it should include %base-packages? Because my config.scm generated by the installer does not include %base-services. <crab2>Oh wait it's working normally now <dftxbs3e>uhm, getting "out of memory test failed" on Guile - it doesnt make much sense, a test that tries to run out of memory and fails to do so? <lisbeths>I have a percentage chance of having bad news <lisbeths>last nigth I installed guix on my xubuntu machine and when I logged in this morning a line was added to my xinit from a hacker <lisbeths>One is microsoft corporation and there is an 85% chance they are the culprit <lisbeths>another is that my grandma downloads random exe files and so it could be a russian hacker. it is very unlikely it was them though becuase it was a higher tier hack as well as they have no motive to add this line of code to my xinit <lisbeths>another possibility is that someone in the guix team was able to slip me the line exit 0 into my xinit <lisbeths>this is also unlikely becaues the guix team has "no motive" to slip me this line. There is nothign to financially gain from it and as far as I know there is no one in the guix project who knows me who would want to exact revenge on me leaving no motive <lisbeths>however there is still a small percent chance that some program from guix is respoinsble and so therefore it is my duity to inform you of the possibility tha tone of your coders is being a bit too silly and invovled with light grey hat hacking <lisbeths>in all likelihood it was microsoft corproation however <dftxbs3e>I am good yes :D - Well not much, it's been stuck on a stupid Guile test... that tests if it's able to get out of memory.. but apparently it can't!!! <samplet>dftxbs3e: That’s too bad. Do you know how much swap is configured? <dftxbs3e>It's so weird that there's a test that tests if it's able to run out of memory..? <dftxbs3e>I increased the RAM of the VM because some other stuff was failing to build and now something fails to run out of memory.. what! <samplet>I don’t know anything about it. I’ve never had to build Guile. <dftxbs3e>it would be such a mess to get bootstrap tarballs independently <dftxbs3e>samplet: I set up another more powerful machine for this and I found an option in guix-daemon to increase the number of parallel jobs, so I made it 4 instead of 1 <samplet>dftxbs3e: Do you have the list of changes we’ve made so far to “core-updates”? I am going to try building everything on a spare machine I have. <dftxbs3e>(note that dpaste doesnt have the newline at the end so; add it and then patching will work) <switchy>how would I best maintain my own updates on top of pre-existing packages? GUIX_PACKAGE_PATH runs into conflicts (understandably...). should I have a local clone of the standard channel and just manage merging in changes from upstream? <dftxbs3e>switchy: ./pre-inst-env and a git checkout <dftxbs3e>and yes, git pull upstream <your-branch> <dftxbs3e>with an upstream remote set to Guix's git <dftxbs3e>or if you don't submit your changes to a git server, simply leave it as origin <samplet>switchy: What kind of conflicts? Naming conflicts? <switchy>dftxbs3e, what's the ./pre-inst-env for? <dftxbs3e>switchy: it allows you to run guix compiled from a git checkout as if it was installed without installing it and potentially breaking your setup <dftxbs3e>it allows both guix and your custom guix to live together <dftxbs3e>pre-inst-env is created when you "make" the project from git <switchy>dftxbs3e, so for standard packages, I should just use regular guix, but if I'm modifying them I should use pre-inst-env? <dftxbs3e>switchy: yes, you should recompile guix from your git checkout for it to use your modification, and the ./pre-inst-env <switchy>dftxbs3e, why would I need to recompile guix just to change packages? <dftxbs3e>switchy: I think that's how it works, I may be wrong, I'm quite new. I think package definitions are compiled inside Guix <switchy>dftxbs3e, cool, I'll give that a shot. thanks! <dftxbs3e>samplet: building at full speed with all 32 cores used at all times now, with -M 4 on both the daemon and the guix build command <dftxbs3e>Before it was stuck at some steps for hours with only a single core occupied <apteryx>switchy: I think you could also use channels <apteryx>which is kind of the same idea (a git repo with your custom modifications), except that it's lighter (you don't need to duplicate the full guix project) and more convenient to share across machines/users. <apteryx>it's relatively new and I haven't used it yet, but it seems nice! It's documented in the manual. <samplet>dftxbs3e: Nice. I just started building “hello” for powerpc64le-linux-gnu. <dftxbs3e>I could cross compile it over here without an issue <samplet>dftxbs3e: Yes, and it will warm up all the cross-building tools. <apteryx>switchy: on top of pre-existing package to me means you'd add your channel that contains your custom packages, inheriting from the standard ones that you want to modify. <samplet>dftxbs3e: The Glibc build failed because I clobbered the CFLAGS. :S <dftxbs3e>samplet: well we're lucky, this machine is very fast, so it should be easier to test <dftxbs3e>samplet: if you're surprised it prints for several different builds at the same time, use -M 1 to force a single job at a time <samplet>dftxbs3e: It’s building again. Wow this machine is speedy! <samplet>Yes. It is not strictly necessary though, since Guile will check the timestamps on the files and interpret the source files if they are newer than the bytecode files. <dftxbs3e>samplet: isnt there some CFLAGS to modify though? Instead of adding things like this -g -O2 -mfloat128 <dftxbs3e>or this is passed to the Makefile that's going to manage the rest <samplet>dftxbs3e: I should have done it as a configure option, I believe. However, it doesn’t matter, since it failed in the same way anyway. <samplet>I’m going to try to build bash-static with “-mfloat128”. <samplet>The same as before about “_Float128”. <samplet>Yes. Glibc built, but building other things with it failed. <dftxbs3e>samplet: hmm alright, so we're going to have to set all CFLAGS to -mfloat128 <samplet>dftxbs3e: Probably. At least everything that uses static Glibc, I guess. <apteryx>how much ram is needed to build icecat? <apteryx>my OK. I have a 8 GB, 4 core machine and the RAM filled up to 90%. I killed it. <dftxbs3e>apteryx: lower the amount of cores used to build, guix has an option for that <apteryx>I'll keep in my --cores=2 for my next attempt. <dftxbs3e>samplet: why does it think the C compiler doesnt work? <samplet>dftxbs3e: Because it can’t link libc. <dftxbs3e>samplet: Note: "configure: WARNING: unrecognized options: --enable-fast-install, --disable-shared" <dftxbs3e>samplet: from the --help on the configure of bash, it's "--enable-static-link" rather <samplet>dftxbs3e: That may be worth fixing, but it is not PowerPC specific. <dftxbs3e>samplet: don't you think it could be the cause for us here? <samplet>dftxbs3e: Not really, since it would fail for all architectures then. <dftxbs3e>samplet: we're on core-updates, other architectures may fail <dftxbs3e>samplet: over on gentoo, they used gcc 7.3, we're on 7.2 right? <dftxbs3e>"aarch64-linux-ld: /tmp/guix-build-glibc-cross-aarch64-linux-2.28.drv-0/build/libc_pic.a: error adding symbols: archive has no index; run ranlib to add one" <dftxbs3e>should we run ranlib on the object first? Or I think there's an argument to pass to ar <dftxbs3e>samplet: "s Add an index to the archive, or update it if it already exists. Note this command is an exception to the rule that there can only be one command letter, as it is possible to use it as either a command or a modifier. In either case it does the same thing." <samplet>dftxbs3e: Not sure what to do. Would it make sense to try a different arch? It seems that aarch64 has its own problems. <dftxbs3e>samplet: x86_64 then - it looks like that archive problem is common to cross compiling <samplet>dftxbs3e: I’m trying to get a PowerPC cross compiler setup that I can play with to see if I can find a small example of the problem. <dftxbs3e>all these people having similar problems while cross compiling <dftxbs3e>it'd be nice to have a log about the changes that were made to gcc stuff on core-updates <dftxbs3e>samplet: well I used --target=x86_64-linux even though we were already on that platform, so it must be cross compiling that's broken entirely; I'm trying without --target now <apteryx>dftxbs3e: git log --grep=gcc in guix sources doesn't cut it? <samplet>dftxbs3e: Right. Using target always makes cross compilers. <samplet>The big change was moving to 7.4. But there was also a change in cross compiler environment variables. <dftxbs3e>samplet: are the people in charge of doing this work here? What about Ludovic, do they only interact through the mailing list? <dftxbs3e>I would think there's a set of maintainers that keep track of every change and would have a clear view of the state of each branch, no? <dftxbs3e>Or the changes that are left, why things don't work <samplet>dftxbs3e: It’s early days for “core-updates”. Some big changes, like GCC 7.4, have been committed and a lot of things are broken. In the next month or two, those problems will be addressed, and “core-updates” will be merged. <dftxbs3e>samplet: okay, does that mean we're better off waiting few months before getting it to powerpc64le? <samplet>dftxbs3e: It would be easier, sure. Getting started will be nicer, because you will be able to download a lot more binaries. Specific PowerPC problems will still be there, though. <dftxbs3e>samplet: yes, powerpc specific problems are fine, we've gotten over lots of problems that get similar over various projects <dftxbs3e>the community knows about all the problems specific to powerpc <dftxbs3e>I've met some, many others have met some <dftxbs3e>It's the system that Guix use that I don't know of, so if cross compiling is just broken, then we'll wait I guess.. I'll try one last time on version 1, to make sure it doesnt work there <dftxbs3e>samplet: compiling bootstrap-tarballs without target didnt work <dftxbs3e>so I guess it's far away for core-updates here <samplet>dftxbs3e: What was the error this time? <dftxbs3e>"guix build: error: corrupt input while restoring archive from #<closed: file 298fcb0>" <dftxbs3e>samplet: I'm running it again, it may be a temporal issue <dftxbs3e>samplet: so nope, we're meeting the floating point issue of glibc again, that requires gcc 6.2 + (on version 1) - can't we downgrade glibc? <dftxbs3e>glibc 2.24 or so didnt require such a thing <samplet>dftxbs3e: Huh? We are meeting that issue on x86_64? <dftxbs3e>samplet: nooo I just tried again building bootstrap-tarballs for powerpc64le on version 1 of guix (not core-updates) <dftxbs3e>samplet: for non-cross compiled bootstrap, it's still going on <samplet>dftxbs3e: Ah I see. I’m not so sure. It would be harder to get the changes merged into Guix, I would imagine. <samplet>It would be a lot of work for something that never gets merged. <dftxbs3e>samplet: well for now, if we can create the bootstrap binaries, it can still get merged right? As long as the process to create them is documented <dftxbs3e>or if it doesnt get merged, at least I can bootstrap guix 1.0 with bootstrap binaries created by say, 0.13.0 <samplet>dftxbs3e: I can’t say for sure. I guess you could create a branch on 0.13.0 (or whatever), get the bootstrap tarballs, and then keep that branch so that people could reproduce them in the future. <dftxbs3e>samplet: there's a chance it works out of the box on v0.13.0 :P <dftxbs3e>it looks like it's making progress, we'll see :D <dftxbs3e>bootstrapping x86_64-linux from x86_64-linux still ongoing on the server <dftxbs3e>samplet: do you think I should cancel it..? <samplet>dftxbs3e: Maybe. It might not be that interesting. <dftxbs3e>samplet: alright, I canceled it, I'll use the power for v0.13.0 + c89e21078862bd29b70a3bc0f33d160c59720364 - it probably works without cross-compiling, I'm thinking the cross compiling process as a whole has an issue <lisbeths>I think perhaps guix intall icecat is stuck in a loop <dftxbs3e>samplet: what are you testing out on the machine? <samplet>dftxbs3e: I was trying to build small example of the “_Float128” error, so I could iterate on it. <dftxbs3e>samplet: okay, I changed it at some point but I did git stash so I reverted it then, because I realized you would be using it <dftxbs3e>did you lose anything? I didnt make any data damaging operation <samplet>dftxbs3e: Heh. Okay. That makes sense! Everything’s okay. <dftxbs3e>try recompiling guix, it doesnt take so long <efraim>Would it make more sense to add it to the gcc flags instead of glibc cflags? <lisbeths>nevermind I am not in a loop just taking a long time to compile <dftxbs3e>lisbeths: it may compile things multiple times because the dependencies arent compiled the same way exactly so it has to make several copies compiled each with their own parameters so it's fully reproducible :P <samplet>efraim: Do you mean “-mfloat128”? Maybe. It doesn’t work at all yet. <samplet>Hey that’s neat. Right now my understanding is that it is only needed for code that uses the “_Float128” type, so it might not make sense to enable it everywhere. <dftxbs3e>samplet: hey I think what I'm trying is working!! <dftxbs3e>it's using the cross compiler right now, and successfully! <dftxbs3e>:D - If I'm able to get Guix bootstrapped today it's amazing <dftxbs3e>samplet: oh no :( - Didnt work, investigating.. <dftxbs3e>../../../../gcc-5.4.0/libsanitizer/asan/asan_linux.cc:222:20: error: ?SIGSEGV? was not declared in this scope <dftxbs3e> return signum == SIGSEGV && common_flags()->handle_segv; <samplet>dftxbs3e: I have to go now. I detached my screen session which is building one last (doubtful) test. I’d like to know the result, but if you have to bring down the server or anything – feel free. Hopefully the 0.13.0 project works! <lisbeths>I have icecat installed with guix and it works except that none of the text renders properly <decent-username>Is there some hidden pgp knowledge everyone except me has? I find it always really annoying that almost no website ever tells you how to import the key. <dftxbs3e>or the manual, that's probably much better <decent-username>user@computer$ gpg --keyserver pool.sks-keyservers.net --recv-keys 3CE464558A84FDC69DB40CFB090B11993D9AEBB5 <dftxbs3e>decent-username: try a few times until it works <civodul>decent-username: could it be that there's a firewall blocking the relevant port? <roelj>What's the state of using a proxy server with guix-daemon? Will it honor ftp_proxy, http_proxy, and https_proxy? <civodul>i'll need support for https_proxy real soon at work though <civodul>i hope someone can do it before i have to :-) <roelj>What would need to be done for https_proxy? I thought gnutls made HTTP/HTTPS transparent. <civodul>proxies are handled by (web client), not by GnuTLS itself <roelj>Aha, and there's only current-http-proxy there. <civodul>i'm not sure exactly how HTTPS proxies work <roelj>Yeah.. the proxy is a MITM, which it is supposed to protect against. <roelj>Heh, this needs more investigation indeed <ZombieChicken>Just did a quick install using the GUI installer and I have a few questions <ZombieChicken>1) Does it not provide a way to do complex layouts? Like having /boot on a seperate drive? <civodul>ZombieChicken: yes and no, respectively :-) <civodul>i'm biased, so you'd better judge by yourself ;-) <civodul>and report concrete issues that you find <ZombieChicken>I recall reporting issues before and people just ignoring them, saying thing were "good enough" <reepca>civodul: speaking of concrete issues, have we ever tested that module support in emacs works? I've been experimenting with it and it's been acting very strangely <civodul>reepca: i haven't tested it, but the person who submitted it (Pierre-Antoine) has been working on a Guile module for Emacs, i think <reepca>referring to the dynamic module (.so) loading support present in emacs since version 25-something? <ZombieChicken>what are the default commands for managing the network in guix? I'm not seeing ifconfig anywhere, nor ip... <civodul>reepca: yes, though i can't find their code, dunno if it's public <roptat>ZombieChicken, in guix system the best way to manage network is to add a networking service to your configuration <roptat>if you need ip, it's in the iproute2 package <roptat>but yes, you have NetworkManager or Connman <roptat>most of guix is managed by the services, and they don't need to add anything to your profile <ZombieChicken>okay, that just makes things worse in a sense; I'm trying to update some software and running (as root, since I'm just testing this in a VM) "guix pull && guix package -u" and guix package -u is complaining I need to run guix pull which, as you'll notice, I did literally seconds ago> What gives? <roptat>probably your shell is not calling the new guix <roptat>make sure ~/.config/guix/current/bin is first in your PATH and run "hash guix" <roptat>(hash guix has no output, it just invalidates bash's cache of binary's locations) <ZombieChicken>yeah, that's really weird. I would ahve thought that would have been done automatically, or a message printed telling the user "Hey, guix was updated and now you have to manually run 'hash guix' so things don't spit out useless error messages" <rekado_>ZombieChicken: it does that, but there’s a bug report about it not working … <roptat>it should have told you that on your first guix pull <rekado_>not sure about the “doesn’t” because I’ve only ever seen it display the message <rekado_>but apparently that’s what happens in some cases. <ZombieChicken>alright, that's fun. ran "guix package -i xorg-server xterm" followed by "guix package -i xinit", then 'startx -- vt1' (expecting a driver or permission error). Instead, I'm getting complaints that tty, hexdump, expr, uname, and hostname aren't available... <dftxbs3e>Guix is now bootstrapped to powerpc64le-linux! <dftxbs3e>civodul - ZombieChicken - rekado_: All it took was some careful git cherry-picking on top of guix v0.13.0 aha <dftxbs3e>newer glibc requires gcc 6.2 + to function properly on POWER9 <dftxbs3e>and gcc 7.4 support for bootstrapping Guix is immature as of now <ZombieChicken>I thought it was a mix of a few older and newer instruction sets <dftxbs3e>ZombieChicken: probably but I havent found how and this has been the easiest <dftxbs3e>I'm not sure you can actually disable the requirement from the glibc <rekado_>it’s the current default GCC that we use to build packages. <rekado_>you can get 9.1 by installing gcc-toolchain <rekado_>newer versions are usually stricter and we do these upgrades on the core-updates branch to make sure that most of the 10k packages can still be built with the new default. <rekado_>we can easily and do provide newer GCCs, but we don’t build all packages with the newest version. <rekado_>dftxbs3e: once you have the bootstrap binaries are you able to use Guix 1.0.0 at all? Or do you wait for core-updates with gcc6 as a default? <rekado_>speaking of core-updates… does anyone know its status? <rekado_>I guess we’ll need to merge master into core-updates and start building it. <ZombieChicken>okay, weird. Apparently I installed which while trying to install busybox, yet it's not showing up <civodul>i think apteryx worked on it recently <rekado_>civodul: I’d like to add a couple of views to Cuirass to more easily see what’s up. <rekado_>civodul: would you be okay with deploying development versions of Cuirass on berlin? <rekado_>I’d be working on a fork / branch and reconfigure berlin with the experimental Cuirass every once in a while. <rekado_>one thing that could easily be added, for example, is a search of builds by nix_name, so that we can easily see the history of certain package builds. <boogerlad>what is the difference between guix environment --container and guix system container? I can understand wanting to run an application in a container, but the os? what for? <efraim>With that I might try ppc 32bit from 0.13 <efraim>And I should write my bootstrap test I've been talking about for years now :/ <sirmacik>is there a way to modify settings of alsa-service-type loaded with gnome-desktop-service-type in config.scm? <rekado_>(hint: you can use “Hey Guix” as a more inclusive greeting here) <rekado_>sirmacik: yes, you can use modify-services to match on alsa-service-type and replace its configuration. <sirmacik>so it'll look like (modify-services alsa-service-type (alsa-configuration (etc))) ? <rekado_>sirmacik: no, modify-services takes a list of services and any number of match expressions. The manual has a few examples. <rekado_>in the info reader of your choice hit “i” and then type “modify-services” <sirmacik>one more trouble, I'm running gnome-desktop and gpg --list-secret-keys sees whole keyring, but when trying to decrypt password with pass or runing seahorse there are no gpg-keys to be found <dftxbs3e>where do you think I can upload bootstrap binaries? <g_bor[m]>Was wondering what is the guixy way to set up a simple static website. I would like to clone a git repo (something like a git-reference origin), and set up an nginx service to serve from that. <efraim>dftxbs3e: I'm pretty sure you can 'guix download' the file on your ppc64le machine and host them somewhere later, as long as you put their hash in commencement.scm <roptat>g_bor[m], maybe create a package for it, and use that package as the root of the website? <roptat>look at cat-avatar-generator-service-type for an example :) <dftxbs3e>efraim: well yes I have it on my ppc64le machine; I'm trying to edit the source to add the hashes etc <g_bor[m]>I was also thinking along similar lines, and just seen cat-avatar-generator :) <dftxbs3e>efraim: hmm I see now, well I don't think I have to edit this, bootstrap.scm rather <dutchie>hmm, how come my laptop can see CJK fonts (and other stuff) but my desktop can't <brendyyn>dutchie: you may require running fc-cache -f <dutchie>i tried running that inside `guix environment fontconfig` <dutchie>i guess that may do nothing and i should install it for real <dutchie>ok, well it worked after i installed properly <dutchie>and yeah, i see how --ad-hoc was necessary <dutchie>must have been running /usr/bin/fc-cache then <efraim>Ah, I wasn't in front of my computer <dftxbs3e>what to do when the backtrace doesnt say where it failed? It only mentions code in eval.scm <efraim>I'm still on my phone so I can't help much right now, but I'd love to see the full backtrace and maybe the patch too, to se if I can spot anything <dftxbs3e>efraim: it's not crashing in the patch code and the backtrace doesnt say anything useful <dftxbs3e>maybe eval.scm means it failed in code that was eval'd? <amz3>dftxbs3e: the error is: `match-error' with args `("match" "no matching pattern" "powerpc64le-linux")'. <amz3>dftxbs3e: maybe it is missing a dash like powerpc64-le-linux <efraim>i get caught up in typos too often <sirmacik>how can I run mcron if I just want to run scripts from users crontab? <sirmacik>In procedure copy-file: No such file or directory <sirmacik>I've got (service mcron-service-type) in config.scm <sirmacik>but that doesn't cut it, that pasted error is what I get after crontab -e <efraim>I don't know that we have a service for that yet <efraim>but you can run 'mcron -d' as a user and put your cron scripts in ~/.config/cron/ <arshin>in guix even cron can be configured in Scheme syntax, cool <sirmacik>how to setup mcron -d to be run at system start? <civodul>rekado_: i'd be ok, but it'd be nicer if it were integrated with the checked-in config <civodul>rekado_: i think you could define a Cuirass package in the OS config, that would use (git-checkout ...) as its source instead of a Git snapshot <sirmacik>wee, I've managed to solve gnupg/gnome-keyring problem with setting right path to pinentry (I need to get used to new paths), but still don't know how to automatically run cron daemon on system start :/ <pkill9>sirmacik: there is an mcron service <dutchie>is there some scheme function i can call to get the path to the current build directory? <pkill9>dutchie: i wanted that a whie ago, i don't think there is, you can get current directory with (getcwd) though <sirmacik>pkill9: I've got (service mcron-service-type) in config.scm <dutchie>starting to think this test is just actually broken <sirmacik>and after crontab -e I get /var/cron/tabs/sirmacik no such fil or directory <pkill9>sirmacik: what does `herd schedule mcron` say? <pkill9>have you reconfigured te system? <pkill9>maybe it needs an mcron-configuration passed to it to register? though i would have thought it wouldn't need that <efraim>sirmacik is looking for a crontab service, not an mcron service <pkill9>doesn't mcron do that? I dont actually use either mcron or crontab <sirmacik>efraim: I thought mcron provides crontab <rekado_>civodul: yes, that’s exactly what I wanted to do: define a package variant in the confi. <sirmacik>pkill9: I do not have any service I want to run in mcron-configuration :/ <civodul>i just want to make sure we don't add ad-hoc services <civodul>i have experience doing things in a half-baked fashion on hydra.gnu.org :-) *rekado_ has been upgrading this Fedora workstation for hours to test the SELinux policy… *rekado_ –> afk for yet another reboot <efraim>sirmacik: mcron does provide a crontab binary, but the mcron-service-type runs 'mcron job1 job2 ...' not 'crontab' <sirmacik>so what I need to do to get anything crontab like running? ~/.cron/job.vixie works for me <sirmacik>it doesn't have to be edited via crontab -e <efraim>after logging in you can run 'mcron -d' and it'll read the jobs <efraim>all the files in ~/.config/cron (or the deprecated ~/.cron) with .guile or .vixie extensions <sirmacik>I actually have few programs I want to run like that <sirmacik>to date I've done this via crontab and @reboot <sirmacik>but that's not an option since mcron doesn't support anything like @reboot <efraim>in the past I've relied on my DE's run-on-login for things like that if I didn't use @reboot <sirmacik>so I need to create desktop file to run mcron in gnome <efraim>I think you can put it in .xsession or something <efraim>i've turned most of mine into shepherd services and after logging in i run 'shepherd' and then 'herd start syncthing' and others <M4R10zM0113R>in the manual there's an example to "(define libressl-instead-of-openssl)", could I apply the same but for alsa-instead-of-pulseaudio? <sirmacik>but It prevents gnome-session from starting (xsession replaces regular start) <pkill9>sirmacik: herd doesn't see mcron for me either <dutchie>ugh these tests fail non-deterministically <sirmacik>pkill9: do you have any cron jobs set up? <pkill9>then again i don't have much use to <sirmacik>another funny error, cups not printing any text <sirmacik>and gnome-tweak-tool runs but doesnt show, without any error <pkill9>gnome-tweak-tool fails for me too, i think it's out of date <pkill9>it's version doens't match the version of gnome-shell packaged in guix <pkill9>i think gnome-tweak-tool started failing for me when i upgraded gnome-shell, but the gnome-shell upgrade branch has a newer gnome-tweaks (the new name for gnome-tweak-tool) <pkill9>no, only workaround is to change settings using dconf <pkill9>or write a package definition for gnome-tweaks version 3.28.x <pkill9>it's kinda awkward because they changed the name upstream, so you can't just build the current gnome-tweak-tool package definition with the newer source <pkill9>hmm i'll see if it is, an if not i'll try using gnome-tweaks 2.28 with that definition <sirmacik>oh, co what I need to do? can you share file? <sirmacik>when finally replaced wlan card with atheros on my x220 <pkill9>it doesn't actually need all those imports at the beginning, i just couldnt be bothered to sort out which ones are actually needed, lol <pkill9>so just dumped them all in there <sirmacik>in the meantime I've managed to start mcron -d at login along with syncthing <sirmacik>by placing custom desktop files in ~/.config/autostart <civodul>if you're using GMail, chances are you've been unsubscribed from the mailing list(s) <civodul>due to a misconfiguration issue with one of the sender's email servers <civodul>i still don't quite understand this failure mode, but that's the way it is <sirmacik>so apart from gnome-tweaks (which I can wait for) looks like everything works :3 *sirmacik has liberated his laptop <pkill9>and now after restarting the gnome-shell, the new gnome-tweaks is working again o_O <iyzsong>civodul: i looked 'https_proxy' for a while, i think we just need to change 'open-connection-for-uri': send a CONNECT request to the proxy, read a 2xx response, then return the result socket. <pkill9>and now it's decided to not work again, lol <pkill9>luckily i got the output when it was working so i cna compare <pkill9>sirmacik: ok my only guess as to what's going wrong is that the version of gnome-tweaks must match the version of gnome-shell, so here is an updated package definition that uses the matching version: https://paste.debian.net/plain/1082343 <pkill9>you can set them separately as well though <M4R10zM0113R>oh... I didn't see that in the manual... and I already have it opened... hahah... <pkill9>sirgazil: hmm it's the same for me <pkill9>it might be one of the extensions icecat comes with <sirgazil>I disabled most of them, except for HTTPS everywhere... <civodul>sneek_: later tell iyzsong looks like you're the right person to look into https_proxy support :-) <sirmacik>sneek_: same for me, with disabled adblocking, noscript and so on <sirgazil>civodul: I just sent a support request to zoho about the bounces. <efraim>dftxbs3e: I'm looking at commit 3b88f3767d9f3ad2cc64173525cd53d429bfe7e7 for building powerpc (32-bit) bootstrap binaries <civodul>nckx: nckx would be able to explain the nitty-gritty details i guess <dftxbs3e>efraim: I'm currently adding the stuff in Guix so it can use the bootstrap binaries I compiled <sirgazil>But I think it could be what Giovanni Biscuolo is trying to reproduce in "Installer: howto remotely debug/report on bare-metal". <civodul>sirgazil: that's when you attempt to restart the installer, right? <M4R10zM0113R>How do I gc the duplicates? There's like... 2 or more duplicates of the same package (different hash) <efraim>Take a look at the commit I mentioned above, it's the commit to add aarch64 to Guix <dftxbs3e>efraim: yup, I had that done already when you sent the commit <rekado_>sirgazil: I previously also had a support request open with the Zoho folks, but they eventually concluded that it’s not their problem. <rekado_>sirgazil: the GNU sysadmins also said that it’s not their fault. <rekado_>so I’m getting bounces and delays all the time <sirgazil>rekado_: I remember you mention that before. I also have delays, and my subscription to the mailing lists have been suspended twice. <rekado_>I think the subscription thing is actually a GNU problem. <sirgazil>And I wouldn't know what other email services I could use... <rekado_>I’d like Guix to have some services to *easily* set up a mail server.. <sirmacik>there is mailcow which uns in docker, that might be a start <sirmacik>setp guix server, generate guix based docker images for the task and so on <tao[m]>Yes NixOS makes it very easy to set up your own and to be honest it is also a big part of what is keeping me from jumping ship <sirmacik>hm, one more error related to printing apparently <sirmacik>while viewing pdf file in emacs I don't get whole file but some half baked pewview <sirmacik>but while viewing file in evince it looks okay, but prints the same way as view from emacs <sirmacik>so probably I'm missing something for ghostscript but no idea what <dftxbs3e>hmm, "/tmp/guix-build-file-boot0-5.33.drv-3/file-5.33/src/.libs/file: error while loading shared libraries: libz.so.1: cannot open shared object file: No such file or directory" - that's a bug <dftxbs3e>it fails to set libz.so.1 to its full /gnu/store path and instead uses system's libz.so.1 <joshuaBPMan>I'm starting to wonder, if me forcing my laptop to shutdown via the power button may be causing some filesystem corruption... <joshuaBPMan>and if mac laptops are really the best way to run GNU/Linux. I just seem to keep running into weird issues. <nixd>Hello guys, I am fairly new to guix and I have a little question regarding the package repo design of guix. Is there a reason why most packages are not in the repo with different versions, but only the last? Would this not hinder the complete reproducability? Say you have a full system configuration and require a specific version of a package, but <nixd>only the latest one is present in the repo. Would I have to write my own package definition for my old version? Why are previous version package definitions thrown out of the repo (If I am even correct in saying this)? <joshuaBPMan>I believe that you can install old versions of packages. I'm pretty sure that you can specify which version to install. <joshuaBPMan>when "old" versions of packages are "thrown out", it is probably because no packages depend on them. <philip>Trying to boot into Guix right now. I dd'ed the iso onto my USB stick, but when I reboot into the stick, it yields unknown filesystem errors and puts my into recovery grub. <vagrantc>you might have to use a guix build from a commit that still has the version <nixd>That would be great to hear, but last time I checked I couldn't find anything like this unless I didn't look hard enough. The version I last used was 0.15 <rekado_>nixd: it is true that we don’t keep around old versions, because maintaining historic versions of most packages would be too much of a burden <rekado_>(it’s hard enough to keep up-to-date software in a state where it can be built) <rekado_>nixd: instead of keeping old versions around we recommend using older versions of Guix to access these older package variants. <rekado_>you can use “guix pull” to get older versions of Guix, which come with these older package variants. <joshuaBPMan>philip: that sounds weird...I'm not certain what one can do with recovery grub. <nixd>Thank you for that answer, however, would keeping around the old package definitions really be a burden? I mean after all it would just be an old definition, right? <rekado_>in fact, you can access these older versions of Guix even programmatically via inferiors <nixd>And the build shouldn't be affected by newer package definitions <nixd>Did I mess something up in my head in thinking this? <joshuaBPMan>nixd: as rekado_: just said, guix inferiors allow you to use old packages really well. You might look into those. <rekado_>nixd: we try very hard to ensure that packages can actually be built and that they don’t represent security problems for Guix users. <rekado_>nixd: keeping old package definitions is not without cost <nixd>Could you elaborate? I am not an expert on any of this <rekado_>we are improving Guix internals continuously, and this can require changes to package definitions. <rekado_>to ensure that dozens of older variants of 10k packages are buildable is not easy. <rekado_>in some cases we do maintain package definitions for more than one version, but we only do this when there are good reasons for keeping more than one version <rekado_>(e.g. for GCC or for some scientific software libraries with more than one major version in wide use) <joshuaBPMan>rekado_: If you had to pick some various laptops that would work well with Guix, which ones would you pick? I personally would prefer to have a laptop that has 4-8GB. <nixd>But these would essentially be separate packages with separate build instructions, correct? <rekado_>yes, but old package definitions would still need to be built an ever updating toolchain. <rekado_>or we would *also* have to keep maintaining multiple variants of toolchains <rekado_>ultimately we would have to commit to maintaining all snapshots of the software universe as seen from all versions of Guix <nixd>I see, but what I don't seem to understand is how just appending new versions for toolchains/build systems and everything else instead of deleting the old ones would cause much stress. If everything would be kept around it should work out, no? <rekado_>but that’s ridiculous when a much simpler way exists: to use an older version of Guix to build old software. <rekado_>package definitions are not mere data; they are code <rekado_>keeping them around while Guix itself changes requires changes to the old definitions. <rekado_>I don’t know what else I can say to show that this is clearly not a sustainable thing to do. <nixd>I find using different versions of guix a little unintuitive for now but I think I'll get the hang of it <philip>okay, so I'm trying a different usb live disk creator tool. When I open the iso extracted from the xz file I downloaded, the tool warns that the image does not contain a partition table, and thus probably can't be booted. <joshuaBPMan>philip: hmmm. I am able to boot via that image. So it should be working. <joshuaBPMan>philip: Does your machine have a CD drive? Perhaps you could boot your computer that way? <philip>oof, it does, but that would be kind of a pain I would think. <joshuaBPMan>philip: True. But on my older laptop, I was not able to boot via usb. I had to use the cd. <joshuaBPMan>before you do that....I believe that image should have a partition table.... <philip>yeah, but that would explain why grub pukes. <philip>just a sanity check. I should be extracted the iso.xz file into an iso file when I download it, correct? <sirgazil>I installed Guile in the GNU system, and found that my .guile file is already set up to include readline and colorized. Nice :) <philip>I guess I never did the wget command. I downloaded straight from the website. <joshuaBPMan>straight from the website is essentially using graphical wget ^_- <philip>yeah, i thought so, but in these situations, I always second-guess everything, including how I breath. <philip>alright, I will be back, but I'm going to try to boot into this image now. *sirgazil secretly thinks the release of GNU Guix 1.0.0 was the release of the GNU Operating System 1.0 <sirmacik>is here someone who worked on ghostscript package? <philip_>Well, I'm back, and the boot worked, but I wasn't able to successfully install on a particular partition. So I've decided to do a backup of my important directory and then wipe away all the other meaningless distros on my laptop <arshin>philip_: might make everything else unbootable (or hard to boot) unless you're careful with how guix installs grub <philip_>oh, that's fine. I don't really care about all the other distros. <philip_>I'm just going to install it into my whole hard drive without doing any manual partitioning <arshin>philip_: suit yourself. it's usually a better educational experience (if you're interested in that) to figure such stuff out rather than "delete everything -> reinstall" <philip_>true, true. I've done a lot of that with other distros (NixOS, Arch, Void, Ubuntu) though, so I don't think I'm losing out by skipping the setup. <philip_>Skipping the hard (but probly better way) that is. Before, I was trying to have 5 50GB partitions, one for each distro, but given how large the Nix/Guix store can get, I would like a larger partition anyway. <philip_>I've only used NixOS actually, but I assume guix gc is the same as nix-garbage-collect. It does help, but the store can still easily get to 50GB, right? Plus, the store is never the only thing on a system. <arshin>philip_: I don't know, I'm only beginning to play with GuiX in VM. <pkill9>this is a little off topic, but since it was mentioned earlier, how feasible is it to run your own mail server? <sirmacik>pkill9: I do it with mailcow and it is like a breeze <sirmacik>it gives you every dns configurations you need to make <efraim>Unless you never clear your old profiles or try to install everything 30GB should be fine for the store. On my main dev machine I leave it around 80, but I have tons of old and test packages <pkill9>sirmacik: do you have any problems with other providers? E.g. your email being marked as spam <sirmacik>pkill9: last time it was with custom server of one of my clients, a year ago or so <sirmacik>damn, that ghostscript bug with missing text is really annoying <sirmacik>pkill9: there is an update for gnome-tweak-tool after guix pull :o <arshin>efraim: and how much for /? (say I'll create a separate partition for store) <crab1>Hey, has anyone setup jack2 as a service before? <crab1>I'd like to add it to my system config <pkill9>sirmacik: that one doesn't work for me either as gnome-shell is on version 2.28 <pkill9>a tool for browsing guix package definitions with the ranger interface would be neat <efraim>etc and var are fairly small, so other than gnu most of what is left is home <efraim>Guix doesn't have bin, sbin or usr <pkill9>crab1: I have a pam-limits-service to make jack run, but otherwise i don't think there is a system-wide service for it, it just runs as a daemon as the user i think <arshin>efraim: thanks. my /home is a separate partition too. <arshin>"Currently Guix System only supports ext4 and btrfs file systems." - is this only for root/store, or in general? <rekado_>crab1: jack2 is activated on demand via dbus. <rekado_>I don’t think you need any service for that. <rekado_>if you don’t need the dbus feature I suggest using jack1 instead. <michaelmcandrew>hello all, a noob question that I couldn't find an answer to via google: is there an obvious gui mail client choice? (I use thunderbird but saw that it is not available). Is it claws? Anything else available? <bavier>michaelmcandrew: I've used 'claws-mail' in the past <rekado_>there’s also evolution, but it seems people had difficulties using it. <rekado_>rain1: that’s not available in Guix, is it? <buenouanq>I came from thunderbird and claws-mail has been nice. <apteryx>michaelmcandrew: I use Gnus, which comes with Emacs, which could be seen as a GUI option. <apteryx>it is not exactly 'obvious' though, if I understand the meaning of obvious in this context to be 'simple to get a hold of'. <michaelmcandrew>gnus sounds like fun but I am not sure i could break out of html for email <michaelmcandrew>Is there an obvious choice for a calendar app? (I guess by obvious, I mean fairly well known, lots of people are using it (I use the thunderbird plugin lightning at the mo)) <sirmacik>is there a way to add keyfile to grub so I can type my password only once? <sirmacik>I now how to add keyfile to luks container but how to pass it later? <arshin>sirmacik: what's your encryption scheme? <sirmacik>arshin: encrypted root with esp partition <arshin>sirmacik: something like this on kernel cmdline: BOOT_IMAGE=/vmlinuz-linux-libre cryptdevice=UUID=11111111-2222-3333-4444-354354645645:crroot cryptkey=rootfs:/crypto_keyfile root=/dev/mapper/crroot ***alessandro is now known as Guest5754
<adamantinum>I'm a Guix enthusiast and I'm trying it on VM! (I'm new to IRC but this seems to be the only official Guix channel, so I hope not to do sth wrong here) <pkill9>rekado_: evolution works for me now that i have gnome-desktop-service installed in my config <rekado_>pkill9: that’s good to know. Can we do something to make it work without having to use GNOME? This seems like a big non-obvious dependency. <samplet>Evolution needs (at least) a D-Bus session bus with “evolution-data-server” available. <rekado_>so this package must be installed globally? <yrk>the graphical installer is telling me I have no network devices detected, but I can definitely ping the internet (via ethernet) <yrk>how do I tell the graphical installer to check again? or point it to the right network device? <samplet>rekado_: Yeah. With or without GNOME it needs “evolution-data-server”. Specifically, the service files from it have to be in the search paths for the session D-Bus.