<dstolfa>lhp22: you would do it in your system configuration (by default, /etc/config.scm) as a part of your operating system specification <lhp22>Indeed, I've found that the 'date' command has a `--set=` option <dstolfa>i'm not 100% sure if that's persistent, as it's likely to be wiped on the next `guix system reconfigure` unless you change it in your OS configuration <lhp22>Yeah, for that I read what you've sent <lhp22>(and btw never bad reasons for making guile) <lhp22>Strange, I get the good timezone <lhp22>I'm in VirtualBox from a Windows Machine, that could come from it ? <lhp22>"You can run the tzselect command to find out which timezone string corresponds to your region. " <--- comes from the link above. The command `tzselect` doesn't exist <lhp22>Hm. I'vn't done 'guix pull' from my root account, maybe for that, I do <dstolfa>i've never done a `guix pull` from my root account <lhp22>Ah. But so should I from my user acount, should I? <dstolfa>well, all `guix pull` does is update the git repositories for your channels (roughly speaking, it does a bit more but that's the jist of it) <dstolfa>i don't see how a `guix pull` would change your timezone <lhp22>Visibly, it's not useless (4'149 commits). But doesn't explain why 'tzselect' doens't exist <lhp22>No, in documentation it's said that 'tzselect' is a command for being sure about the string for my timezone <lhp22>But this commadn doesn't seem to exist <lhp22>But. In all ways, my timezone is correct in /etc/config.scm <lhp22>So it should be 00:10, but it was two hours later before I changed it by 'date --set=' <lhp22>But i know that in dual boot Windows/linux (not my case), can happen hour problem. Maybe same problem for virtualbox <lhp22>Eh ! But if it's it, I'm going on telling my OS i'm not here >.> <dstolfa>has this happened to you with any other distribution? <lhp22>And I've just removed all my old virtual machines x'D <lhp22>But, never midn, it's not really a problem <lhp22>So, I've just discovered that PAris and Moscow are in the same timezone. Well. <lhp22>Ah, no. Chocked, should go bed <dp0>Quick question for anyone that's around, how exactly are package updates handled when it depends on an updated version of some other package? Do you have to update both packages? <dp0>Example Scenario: Say there's project A@1.0v and it depends on project B@1.0v. If there's a new version of A@2.0v and it now requires B@2.0v, is it expected that along with the update in the package to A@2.0v that B is also updated to 2.0v? <apteryx>dp0: yes, if updating B to 2.0v doesn't break any of its dependents <dp0>ok. thanks. I guess that makes sense. Just wasn't 100% sure since I didn't find an example of that in the documentation. <apteryx>it if does, and if it not reasonable to have all of them updated to work with B@2.0v, then the B package can have both version 1.0v and 2.0v in Guix, and packages depending on B@1.0v can continue using it <dp0>Bonus question, what if things are optionally dependent? <dp0>It's compilation flags let you use your own installed version of a library or it'll come prepackaged with one. <dp0>I'm assuming it's more preferable to have the dependencies separate so that guix can manage both individually? <dp0>Maybe prepackaged isn't the right term, more like statically compiled with it's own version... I guess <bdju>iyzsong: got around to that reboot and I'm in the input group now, controller still acting the same, unfortunately... <bdju>iyzsong: I tried my wii u gamecube controller adapter and dolphin-emu said "Error Opening Adapter: Access denied (insufficient permissions)", so it probably is a permissions issue in both cases. Just not sure what to do to fix it. ***califax- is now known as califax
<bdju>how does creating new udev rules on guix system work? ***cadmium.libera.chat sets mode: +o ChanServ
<bricewge>You can also use udev rules from packages instead of writing it yourself, ther should be examples of it in the documentation <bdju>bricewge: thanks for the example! I'm not sure if what I've added is right or not. I reloaded my rules and it didn't seem to fix my issue <bdju>it seemed like you had extra quotes in certain spots and also put escapes after a ==, so I tried to copy those things and then it reconfigured without error <bdju>and then I did sudo udevadm control --reload-rules <bricewge>You need to escape each " you want to appear in you rule file, you missed some of then <bdju>I think that's a lot closer to working now. it actually printed lines about udev, and I'm managed to find the rule file and it looks like it should at a glance. <bdju>hmmm I still get an access denied error in dolphin for some reason. at least I can cross udev rules off the list of possible issues for one of the controllers now <bricewge>Provably unrelated but pour dolphin-emu package dosen't contain any udev rules (contrary to arch's zone for exemple) <bricewge>Arf, wrond Android's keyboard language... <bricewge>s/pPovably/Probably/ s/pour/our/ s/zone/one/ <bricewge>bdju: Does your adaptator device have the corect right with your udev rule now? <bdju> bdju: it still doesn't work, but I think I'm gonna give up on that controller for now. a dualshock 4 seems to work fine. switch pro controller still doesn't work. not sure if that also needs udev stuff. there's apparently a hid-nintendo kernel driver but it might not be merged yet <bdju>oops idk why I pinged myself <Soheil[m]><raghavgururajan "Soheil: key_mgmt=WPA-PSK"> Thanks <Soheil[m]>But there is a problem with my connection that I will never understand! <Soheil[m]>Does anyone really know where the problem comes from? <raghavgururajan>Also, where does it show "Activation of network connection failed." ? <jab>Hey raghavgururajan! Are you still wanting to merge your libreboot guide to the guix manual? I'm happy to do that for you. <jab>Though you may need to point me to your guide again... <raghavgururajan>Soheil[m]: Wait, if you are in GNOME, you could directly use network-manager? ***Kimapr2 is now known as Kimapr
<raghavgururajan>Soheil[m]: May be restart and try `rfkill unblock all` before starting the network setup procedure. <Soheil[m]>Activation failed: The Wi-Fi network could not be found *raghavgururajan runs out of ideas <Soheil[m]>Sorry, I'm not going to promote nonguix; But I use Linux and RTL8192EU because of the Wi-Fi connection. <Soheil[m]>If anyone might check this out, there may be something wrong with setting it up. <raghavgururajan>Soheil[m]: I am afraid I can't help you there. Regarding kernel, I had some networking issues with latest versions, so started using `(kernel linux-libre-lts)` under operating-system declaration of config.scm. <jab>raghavgururajan: anytime! I'm glad to be able to help. Any suggestions about where you want me to put your libreboot in the guix manual? <jab>Should I put it in the System installation chapter? <wirez>i usually use the mit license but if i wanna start supporting gnu any probs with dual licensing anyting new i do mit & gnu? <MysteriousSilver>but doesn't dual licensing gpl and mit destroy the whole purpose of copyleft? <wirez>like maybe some ppl wouldnt wanna contribute code under non gpl license? <lispmacs>wirez: dual licensing GPL and MIT does not make any sense. Since MIT licensed code can be used in a GPL project, there is no benefit to anyone wanting to support copyleft, and even some cons <wirez>so copyleft can protect us from stuff like closed source vaccine passport code maybe being forced into mainline linux like systemd was? <lispmacs>um, I really don't have any idea what you are talking about. but the idea of copyleft is to protect free software projects from becoming proprietary projects <wirez>so i guess like some say poettering can deliver linux into microsoft (ibm) hands <lispmacs>when you say linux, are you referring to the linux kernel, or to the Gnu/Linux operating system? <wirez>and i guess gnu and copyleft and libre-linux kernel are all guards against proprietary/unfree licensed code? <wirez>do you think gnu libre-linux guix are viable (have enough dev resources) because the licensing philosophy? <lispmacs>yes, it is. I release like 99% of my code under GPL-3.0-or-later <lispmacs>it doesn't really matter. supporting free software and copyleft is the right thing to do, viable or not <lispmacs>but, all my computer system (two desktops and a server) are running linux-libre <wirez>ya but i mean do you think a lot of ppl choose to work on copyleft libre linux because of the license? and not necessarily the code being really documented or smth else <lispmacs>well, nobody really develops libre-linux, as you normally think of that. It is just a project that cleans out the proprietary garbage from what is, by its own license, supposed to be a 100% free-software, copyleft kernel project <wirez>and basically copyleft philosophy is free to look at and use source however you want (like mit) but if you make changes you have to share them? <lispmacs>a more accurate way to put it: why you share copylefted software, you MUST give all the software freedoms to those with whom you are sharing the software <lispmacs>one of the freedoms is to be able to inspect the source code, so you must give them the source code <lispmacs>others are to be able to modify the code and to share it with others, so your copyleft license gives them those privileges as well <wirez>ok so proprietary is fine, but if you distribute (like sell) you gotta do so with the copyleft requirement in tact <wirez>so it guarantees that if there's virility in licenses for software sharing and distribution which there is, it makes sure copyleft terms are what's viral and not non-copyleft <wirez>which doesn't help mit kinda license ppl which already lets you do whatever you want, but it does protect against less permissive licensing <lispmacs>if you distribute GPL code (for free or for money) you must hand off all the freedoms to the next guy <wirez>and distributing gpl code is even if it's a tiny piece of your own large code? <wirez>so mit kinda licenses hand off all freedoms, but then copyleft licenses hand off all freedoms EXCEPT that you must pass on same set of freedoms <lispmacs>with MIT you are giving the freedoms to the next guy as well, but the next guy is under no obligation to hand off the freedoms to the next guy after him <wirez>so it's restrictive in 1 way but only to set a minimum set of freedom <lispmacs>yes, copyleft is, in a legal sense, more restrictive that non-copyleft free software license <lispmacs>in a practical sense, though, it preserves freedom better <wirez>seeing how bad stewards of the mit/bsd side big tech has been, and how aggressive patent trolls have been, and how negligent bureacrats have been, seems almost necessary <wirez>does gpl enforcement have teeth practically or is it just a moral stance? <lispmacs>yes, there are several organizations that have prosecuted GPL violation and have won significant cases <wirez>if oracle fleas can slash and burn i hope gpl is respected <lispmacs>A trick part though is you have a lot of the chinese companies violating GPL and getting away with it <lispmacs>they add some proprietary junk to a custom released kernel for some Android tablet, and it is difficult to prosecute them <lispmacs>so, GPL enforcement isn't easy in all cases, but it happens <lispmacs>I think Software Freedom Conservancy is a non-profit to support if you want to support GPL enforcement <wirez>it's prolly sjw so i won't do that but ill start using gpl on code <lispmacs>GPLv3 specifically also protects against some other tricks, like "TiVoization", where manufacturer uses free software code on your device, but locks down the hardware so you can't make any changes <lispmacs>GPLv3 also includes some language (helping) to protect against tricks involving software patents <wirez>what's the agpl thing about? <lispmacs>wirez: AGPL has additional requirement that if you run code providing a service over a network, you must make your software changes available to those accessing the service <lispmacs>I wouldn't personally want to use that on just any code you release <lispmacs>make sure you understand section 13 of the AGPL. Can get a bit complicated <lispmacs>you can end up with a project with some parts AGPL and some parts GPL *lispmacs goes back to bed <leoprikler>Well, sure, but can't you also end up with parts GPL and parts LGPL in the same manner? <tonos>Hello! I have question about submitting patches. Should I add my name to the copyright header at the top of the file? <leoprikler>Unless it's a trivial change, adding it is typically wiser than not to. <leoprikler>Some committers will do it for you, but not all of them – I am very prone to forgetting that myself. <tonos>Does adding package definitions count as trivial? <leoprikler>I think that's to be evaluated on a case-by-case basis <leoprikler>if you're piping the output of `guix import' into a file, an arbitrary line count won't matter <tonos>I wrote around 50 lines, but most of them are just the package fields, so I guess its trivial. I'll submit the patches without the copyright line. <Rooks>I don't think I quite understand how to add a service managed by the herd command, I am attempting to execute things like cpupower and thinkfan on boot? <iyzsong>Rooks: to add a shepherd service, you need a 'service-type' that extend 'shepherd-root-service-type', 'sysctl-service-type' is an example for 'one-shot' shepherd service. <dstolfa>Rooks: shepherd is configured in guile, it doesn't actually have any syntax file it parses. instead, it implements an EDSL (which in this case is just a set of record types and functions) that you can use to configure your services. in guix, it is standard to do what iyzsong said <roptat>if we manage to finally merge it, it'll be part of the main channel :) <maxwell_TGAP>core-updates.... is there a way for me to install it as my defult python <roptat>it's a different branch of the guix channel <roptat>you can use an inferior to get just python 3.9 <roptat>or you can specify the branch in your ~/.config/guix/channels.scm to use it, though I don't recommend it, it might not be very stable ;) <mekeor[m]>hmm mayve maxwell_TGAP could add core-updates as an additional channel and then install python@3.8 in parallel to @3.9? <clemens3>mekeor[m]: thanks, so far enjoying it, packages downloading.. tried to replicate the java bootstrap in normal linux, but figured it is easier to have everyting installed first..:) <jorge[m]123>Hola,tengo errores al ejecutar guix pull y en la instalación de cualquier paquete. <apapsch>guix never finishes building a docker image from gnu/system/examples/docker-image.tmpl <apapsch>there is a qemu process constantly consuming cpu time <apapsch>this happens in two guix binary environments, a gitlab shell runner (ubuntu) and current master locally (arch linux) <apapsch>is this a known issue? how can I debug it? <clemens3>mekeor[m]: some feedback, i did choose openbox and no other DE, when i was in x, no xterm.. did install again with xfce only, then got a terminal.. <apapsch>guix system docker-container --debug 9999 leads to another error: "guix system: error: | | | bind mounting `/dev/full' to `/gnu/store/m6b5pvx50zwrnqbdswl89kr93p6zasxr-guix-docker-image.tar.gz.drv.chroot/dev/full" <ggoes>how would i go about installing a package with a different input, *and* have that input pull from the master branch? e.g. i've tried doing guix package -i emacs-exwm --with-input=emacs=emacs-next --with-branch=emacs-next=master (and the equivalent in scheme) with no luck <ggoes>i'm not above defining a new emacs-exwm package for myself, but it seems like that shouldn't be necessary <efraim>you might also need --with-input=emacs-minimal=emacs-next <ggoes>oh good call, i didn't think to change emacs-minimal as well. hmm. <maxwell_TGAP>is it possible to copy the package deffinition for python3.8 from core-updates and add it to my packages path <civodul>apapsch: believe it or not, --debug is buggy (what you're seeing here), but more importantly: it's rarely useful, because it shows debugging output from the daemon <sneek>civodul, you have 1 message! <sneek>civodul, oriansj says: that live-bootstrap have guile and gcc 4.7.4 (both C and C++ backends) all properly bootstrapped (no pregenerated files unlike the current root) and it might make for a stunning release. <apapsch>civodul: thanks. I see qemu gets started with console=ttyS0. Tried connecting there with screen/minicom/cutecom to see what's up, but no success (I must be doing something wrong) <Soheil[m]>Please describe this section, the Wi-Fi connection problem may be fixed. *dstolfa now has a swanctl network to test out with Guix System and will look to write an EDSL to configure with swanctl! <maximed>sneek: later tell civodul: Maybe the numbers.test failures have something to do with the wackyness of floating-point on i386 (see ‘excess precision’à <maximed>sneek: later tell civodul: maybe "-mfpmath=sse -msse2" will help <maximed>sneek: later tell civodul: I'll try adding that to CFLAGS <Soheil[m]>Please describe this section, the Wi-Fi connection problem may be fixed. <leoprikler>This section lists the service-types and configations, that make up (gnu services networking). <leoprikler>Jokes aside, what is your Wi-Fi connection problem? <maximed>sneek: later tell civodul: Guile builds successfully on i386 on core-updates when numbers.o is recompiled with make CFLAGS="-mfpmath=sse -msse2" <lhp22>It's strange. Sometimes I get "error in finalization thread: Success \ udevd[123]: no sender credentials received, message ignored \ Error: Driver 'pcspkr' is already register, aborting <lhp22>I'm on VirtualBox on Windows <apapsch>lhp22: "error in finalization thread: Success" is not an error, it's like outputting exit code 0 <apapsch>the other two messages are not the cause either (likely) <apapsch>I had the same problem when a services g-exp was failing <apapsch>(same problem = system hanging on boot) <lhp22>I'm reading dmesg around the messages <lhp22>And it seems a bug between virtual box and guix <lhp22>Some initializing fail randomly <lhp22>Hm. Iwas trying to send to myself the dmesg report by mail, and so install `sendmail` by 'guix install sendmail' but then 'sendmail: command not found' <roptat>do you have ~/.guix-profile in $PATH? <roptat>well, actually sendmail doesn't provide sendmail ^^' <roptat>I don't know what would provide that command though... <lhp22>'mailx' doesn't exist on guix repo <lhp22>In my path there is '/root/.config/guix/current/bin' and my GUIX_PROFILE is '/root/.config/guix/current' <roptat>oh, actually the sendmail package is broken <roptat>so, it does provide sendmail is usr/sbin/sendmail <roptat>I guess nobody uses sendmail, if no one noticed <maximed>lhp22: "guix environment --ad-hoc THE-PACKAGE -- THE-COMMAND" may be useful here <roptat>so you can call it as $(guix build sendmail)/usr/sbin/sendmail <maximed>if you don't want to install it, and just temporarily need it <lhp22>maximed: "guix environment: error: execlp: No such file or directory sendmail" <roptat>it can't work because it's not in bin/ <lhp22>roptat: which sbin are you talking about ? <roptat>I see /gnu/store/yn6dlixw6cnbmgjggk640vbv8hidfhi7-sendmail-8.15.2/usr/sbin/sendmail <lhp22>Oh ! That what you meant by "store item" <roptat>well, once installed, it should be ~/.guix-profile/usr/sbin/sendmail <lhp22>nevertheless, the sendmail package seems to have many problems in itself x) <lhp22>would you know simple package to send mail through line command ? <bricewge>I struggle to know what modification should go through a review and what I can push <leoprikler>bricewge: those both look like you'd want some comments on them, so to the ML they go <leoprikler>In general, unless you're sure you don't need a review (such as in trivial version bumps), you might want a review ***iyzsong- is now known as iyzsong
<jonsger>bricewge: i would have pushed the second directly and the first one to the ML... <apteryx>is it possible to run guix packs on android? <maximed>apteryx: I see no theoretical reason for why that wouldn't work. Android is ‘merely’ Linux with an unusual user space. I dunno if you can create bind mounts without root on Android though <maximed>the graphics stack is different, so graphical applications would be out <maximed>there was a blog post on running guix on android <civodul>maximed: re Guile on i386, sounds... interesting! <sneek>civodul, you have 5 messages! <sneek>civodul, maximed says: Maybe the numbers.test failures have something to do with the wackyness of floating-point on i386 (see ‘excess precision’à <sneek>civodul, maximed says: maybe "-mfpmath=sse -msse2" will help <sneek>civodul, maximed says: I'll try adding that to CFLAGS <sneek>civodul, maximed says: Guile builds successfully on i386 on core-updates when numbers.o is recompiled with make CFLAGS="-mfpmath=sse -msse2" <civodul>but note that "make CFLAGS=..." also means you're removing -O2 <maximed>civodul: I didn't think about that (removing -O2) <maximed>Anyway, I don't think we can use "-mfpmath=sse -msse2", as apparently SSE2 support cannot be assumed (there are a few package definitions that explicitely remove "-msse2" compilation flags) <civodul>maximed: yeah, that doesn't sound like a valid workaround for us <maximed>I've read that casting to (double) gets rid of the excess precision <maximed>do you know of an easy way to run a single test (numbers.test) of the test suite *civodul builds it locally <civodul>once we have that fixed, we can very much freeze, i think <civodul>well, we also need to check the powerpc64le status, uh <bricewge>leoprikler, jonsger, efraim: Thanks! When in doubt, start a review? <bricewge>Sometime sending such small patch fell like spam to me <bricewge>I don't have a good workflow to review patch and don't want to "saturate" the bug tracker <lhp22>I don't see why the sandboxed app space could be worse than for other package manager <lhp22>Ok, that was an answer about android guix, didn't scroll down >.< <maximed>civodul: "CFLAGS=-O2 -fexcess-precision=standard" + adding (double) casts everywhere makes the tests pass <maximed>(maybe the casts are unnecessary though) <apteryx>maximed: I too think "it should just work". Good point about the graphics stack. Last time I was blocked by not having root (you can't just run a guix pack from /sdcard as that's (ex)FAT, with all its limitation, e.g. no exec bit nor symlinks support). I'm trying to get passed that now. <maximed>civodul: I removed the casts (i.e., reset the code to the original version) and compiled with "-O2 -fexcess-precision=standard" and the tests pass <leoprikler>bricewge: don't worry, you're not alone in sending small patches <leoprikler>people without commit access have to send two-liners just to bump their favourite package :) <civodul>sneek: later tell maximed re -fexcess-precision=standard, well done! \o/ <iskarian>Is there a lint that checks for same-named package variables across modules? Because, surprisingly, there are no duplicates <ruffni>sooo, what's the easiest way to get the version string of an input package in a (modify-phases) expression? i tried un-/quoting and (with-imported-modules) but to no avail <ruffni>iskarian: not sure, but i guess two (define foo) blocks would raise a guile error 