<ng0>this buildsystem is stupid. <ng0>if i only had uclibc-ng completed already.. or if musl was merged already.. I could try and test if those would not send bbs100 in a rebuild loop, causing the setup to fail <iyzsong>jmd: yes, same here. and add '--fallback' works. <jmd>iyzsong: Yep. I guess something's screwy with hydra ***lostcoff` is now known as lostcoffee
<mark_weaver>(the machine I'm currently on, a mips64el box, doesn't have elogind, but I vaguely recall learning that elogind contains a program that, when passed a suitable subcommand, will suspend the machine) <lostcoffee>mark_weaver: not seeing it in the man page, at least not immediately <mark_weaver>lostcoffee: I found it in my logs. alezost said that "loginctl suspend" worked for him. <mark_weaver>and I remember it working for me too. even run as my normal user. <janneke>lostcoffee: mark_weaver, yes that works <janneke>/gnu/store/4mmn5y6syzv7wwz1y6bl1ab4g0yvkdq1-elogind-219.14/bin/loginctl suspend <jmd>What does (lambda _ indicate ? <rekado>jmd: it means that we don’t care about the argument(s) <jmd>Wouldn't (lambda () be the same, then? <rekado>no, () means that no arguments may be passed. <rekado>“lambda _” means: throw anything at me, I don’t care. <rekado>Emacs 25.1 comes with xwidget support, but I found that it uses the old webkitgtk API, which is only provided up to version 2.4. <jmd>Is that a general scheme thing or is it guix specific? <rekado>you can bind all arguments to “abc” if you wanted. <rekado>((lambda abc (display abc)) 'hello 'world 'and 'goodbye) <rekado>the “_” is a conventional name for “whatever” <alezost>I think "_" is a general convention, for example, it is also used in elisp <jmd>How can I find out what the builder is currently building? <jmd>janneke: That's one way. I didn't think of it. Thanks. <janneke>prolly doesn't work with offloading, but hey ;-) <dvc>on hydra the jobs that aren't crossed off are still pending? <dvc>somehow a linux-libre-headers build snook in to master or am I missreading something? <dvc>and what's with all the wip-python rebuilds? <janneke>ng0: since i'm running guixsd i need to surf on my phone <dvc>janneke: packaging inox for guixsd would be pretty awesome... <dvc>it's like icecat for chromium <ng0>i will package firefox at some point for myself.. i don't like the slow development of icecat, lagging behind with cves etc <dvc>inox is current and laggs at most one version behind - no long term support thing <dvc>and there is a way to install proprietary addons manually if you really want to (like the chromecast one :) <janneke>dvc: a browser that allows us to interact with people yet to learn about free software, even if it means giving up our sw-freedom while interacting, that would be nice <ng0>the way is not straight forward.. to install addons. and the store plugin is dropped <dvc>yep, I think it's FSDG compliant - it's possible - not encouraged <dvc>I'm sure it's also possible to install non-free addons in icecat if you really wanted to <ng0>well not all the free addons are in the enabled store in icecat... i install some from the mozilla store and external websites. <dvc>and I for me personally - if I manage to get by with only firmware for my wifi card and an addon for chromecast - that's ~free overall... <dvc>and what I mean by manage is that our university doesn't care about free software, so doing classwork with free alternatives can be a huge amount of extra work - I'll probably have to package quite a bit of software for class in the next couple of years. (kicad for example) <ng0>:( citadel uses no version numbers for their release files. But I think they release slow enough that this doesn't matter <dvc>you know what I had to listen too this week? our english for academic purposes teacher told us that wikipedia is full of crap and can't be taken seriously since anyone can write an article. I'm pretty sure she's never actually read a wikipedia article... <jmd>What is the "replacement" field in package definitions? <quigonjinn>dvc: I plan for packaging kicad for guixsd, soon enough hopefully <ng0>well... i lost the links, but since the beginning wikipedia had flaws. i will dig those cases up again if you want to. <dvc>ng0: weren't those cases related mostly to biographies? I'm sure it's not perfect and you have to read everything with a critical mind - if it's from wikipedia or from a scientific journal... <ng0>tl;dr circle of false news travels fast. want to create a new species? consider it done with wikipedia whic hthen reeaches the international news, which then cites wikipedia, years later other news cite the news which do no longer cite wikipedia. .. happened before too often. <ng0>no. one story i got firsthand, the others are in wikipedia magazine <dvc>so did they have fake papers they referenced from the wikipedia article too? <dvc>I think that scientific journals are expensive - and actually subscribing to one doesn't actually benefit the authors (don't know if that's true) and wikipedia in most cases does a pretty good job... <ng0>i will come back to you when i find the link again. <ng0>scientic journals are also oppiniated.. if they don't like your research you won't get published, and sometimes they are so terrible they just publish whatever.. see examples like 'take me off your fucking mailinglist' or what it was called <jmd>It really depends on the journal. <jmd>And some are expensive. Others are free of charge. <jmd>wikipedia is not a scientific journal and should not be treated as such. <ng0>example are tests on animals, friend's a psychologist with other research backgrounds and there were studies and research that some cruel and brutal testing on animals are not even leading to any usable results, yet they are still used. most magazines did not 'believe' in the research so it got rejected. <jmd>Most reputable journals have an ethics policy, and have a criteria that papers must demonstrate results which have not been previously published. <jmd>There are some less reputable ones, but they probably haven't even been assigned an IF. <dvc>quigonjinn: that would be awesome! <quigonjinn>dvc: my notes tell me that all of its dependencies are packaged in guix, so it shouldn't be much work <need_help1>guix says `guix package: error: failed to connect to `/var/guix/daemon-socket/socket': No such file or directory' <dvc>how did you install guix? <jmd>need_help1: You need to read the manual again. <dvc>did you start the daemon? <jmd>If you are running fedora, then that was the wrong section. <jmd>need_help1: Are you wanting to install GuixSD (the operating system) or just guix (the package manager)? <jmd>the package manager ? <dvc>does guix always need qemu? it's not listed when performing a dry-run... <jmd>dvc: I have never used qemu <dvc>well that's where my daemon socket is so if guix is looking for it there - it's the right place... <quigonjinn>need_help1: you need to add the build users. see my last link <need_help1>quigonjinn: the groupadd & for loop ? I already ran it. <dvc>jmd: are you sure that when you run guix system build config.scm qemu doesn't end up in your store? <jmd>There's a qemu-minimal <dvc>it uses the same codepath for guix system init I think, I don't think substitutes are available for arm - getting build failure <jmd>dvc: Ahh it looks like grub uses qemu-minimal for testing. <dvc>I'm passing --no-grub <dvc>guix system build or guix system init <jmd>What happens if you try to gc qemu-minimal ? <dvc>jmd: On my arm machine? qemu fails to build so I can only use gc after it worked succesfully... <dvc>jmd: building the same configuration with --no-grub --system=armhf-linux on my laptop causes a failure due to qemu too, but guix size the result when run without --system=armhf-linux doesn't include qemu, so it must be a native input somewhere... <jmd>Well grub is the only thing I can see which has that. So I wonder if the --no-grub option is working properly? <dvc>jmd: yes it's grub 1.3.1, it must be that - I already removed one hardcoded reference to grub - they might be others <dvc>jmd: found it - looks like the u-boot patch I'm using introduced this bug <jmd>dvc: So it was your fault all along :P <dvc>and I don't know if it was there before - I just know that that line was modified by danny last... <jmd>I'm not sure what u-boot is. <jmd>(apart from German for submarine) <dvc>bootloader for non x86 systems <dvc>also performs stuff that is done by the bios on x86 systems <jmd>Grub works on non x86 doesn't it? <dvc>I'm not sure - I think there is some experimental support for other archs - but I don't know <dvc>uboot is an industry standard and virtually every board has an uboot port <jmd>Well I have grub on my mips anyway. <dvc>I was thinking more about arm - didn't know that it works on mips <dvc>what kind of machine do you have with a mips cpu? I thought that was only used in some routers <dvc>jmd: so you have a full guixsd install running on mips? <jmd>Err no. I have gnewsense on that machine. <jmd>I think guixsd is possible, but there is no binary installer currently. <dvc>have you tried guix system disk-image gnu/system/install.scm? I'd be really interested to know if it works... <jmd>I haven't tried. No. <jmd>But I think someone said there are issues with the kernel which need to be solved. <jmd>Probably mark_weaver could say more. <dvc>what kind of problems? <dvc>initrd fails to build on arm for example because it expects a kernel module <dvc>is that the kind of problem? <jmd>dvc: Actually, I don't know. <FNTM-U3>I want to install Guix. but I don't know if it is just a package manager or we have a full linux os for Guix too. <FNTM-U3>and if yes where is the amd64 image to install it on the hard driver <jmd>FNTM-U3: There is both. You decide. <FNTM-U3>snape: Image link would be better than installing on the other linux os <dvc>does anyone know if the initrd can be disabled? I can remove the kernel modules for testing - but I'm not sure if arm with a custom kernel and u-boot needs an initrd <FNTM-U3>USB installer of the Guix System Distribution. <FNTM-U3>snape: 151? is it like stage 3 for gentoo <quigonjinn>I am packaging openocd. This package has libusb and libftdi as inputs, as do other flashing tools currently in guix. In my current system (guix on parabola), i am getting permission errors, when trying to use it. Can this be an issue of our libusb, which is being built with udev disabled? <dvc>ah I'm trying to load root through a label, I do need an initrd <dvc>quigonjinn: enabling udev seems like a good idea <quigonjinn>dvc: There a note in the package definition, that it is disabled because eudev indirectly depends on libusb. <dvc>then that libusb should be named libusb-eudev and libusb should depend on eudev <quigonjinn>dvc: thats what i though of. I'll try to fix that and see if it helps. <quigonjinn>actually, using guix graph, i don't see eudev depending on libusb. Am i missing something? <quigonjinn>quigonjinn: well, still doesn't fix the problem. There a chance it's parabola's fault, but i should find a way to test it out <brendyn>quigonjinn: You're talking to your self. <dvc>got a system to build for arm :) <quigonjinn>so, i have a similar issue using flashrom from guix with another programmer. It *should* be a parabola issue. The flashing tools have been tested to work, right? <ng0>i booted into ubuntu live dvd for flashing so far.. there's a /dev/mem thing you need to set, search for a thread about 'flashrom update' or something on the mailinglist. <quigonjinn>ng0: I understand this error applies to internal flashing. In my case, i get permission issues trying to access usb device, for use with external programmer. <quigonjinn>ng0: side note, this /dev/mem issue has happened to me in the past, while using linux-libre-grsec on parabola <lfam>I haven't seen this failure before. I'm trying to build Bash 4.4 on core-updates. The compress-documentation phase succeeds, and then the build fails with "cycle detected in the references of `/gnu/store/fk15p13agmj5zcylp388m4kmmvgvfxmw-bash-4.4'" <lfam>Ah, this comes from the daemon, 'nix/libstore/misc.cc' <lfam>The check 'if (parents.find(path) != parents.end())' fails <lfam>I don't know C++, but that does look like a cycle checker <ng0>cool. i just cloned pijul to see if darcs would work without solving the ssl problem already, and they include a nixos file. <ng0>pijul is one of these many packages i need cargo and rust for <lfam>quigonjinn, dvc: For kicad, please search the guix-devel archives. The package is almost there and only requires someone to finish packaging wx-python, if I remember correctly <lfam>Hm... I wonder what is the best way to debug the bash cycle. <dvc>nice, looks like guixsd is coming together for electrical engineers. Has anyone looked at dannys yosys and iverilog patches yet? <lfam>dvc: I wanted somebody with some domain knowledge to review them. But if enough time passes, I will just review them from the perspective of a Guix packager. If they don't work, I expect somebody with domain knowledge to complain :) <lfam>And "enough time" has passed by now. <lfam>Maybe I will get to them tomorrow. <lfam>But if you or quigonjinn have some domain knowledge, it would be awesome to get your review <quigonjinn>lfam: I can most likely check both of them in my workflow, but I probably won't have enough time til next week <dvc>they should just be normal guix packages. the only thing that might be a problem is having to run icestorm as root, because udev rules aren't set up correctly <lfam>That sounds like a configuration problem more than a packaging issue, right? <lfam>quigonjinn: If so, can you reply on guix-devel saying so? I'm sure Danny would appreciate it, and it might remind some other people about the packages <dvc>lfam: not sure - haven't done anything related to udev yet, but probably <dvc>lfam: adding a comment on what lines to add to /etc/config.scm might be nice... <snape>does anyone know how to setup a static ip address with connman? <snape>oh nevermind, I'll use static-networking-service <taylan>sad news: https://libreboot.org/gnu/ I'm guessing it won't have any effect on distros using libreboot though, since it will continue to be free software (even if they move away from the GPL...) <Gottox>well, at least their unprofessional behavior killed my trust in libreboot. <jmd>I certainly won't be ordering from minifree.org again. <Gottox>I don't think they're the issue here. <Gottox>The problem is, that a single person has blamed people for mysogony publicly without giving them a chance to explain themselves. <ng0>i have too little information, as does everyone who feels suddenly entitled to being a judge on this, to comment on this. <Gottox>even if they're mysogonists and transphobs, you can handle it professional. <Gottox>that means: handle your project independent from your personal issues with people. <jmd>I think the FSF, would have a prima faca case for libel. <jmd>Anyway. Off topic here. <ng0>my countries law does not use these words ,what does this ( a prima faca case for libel ) mean? <jmd>ng0: As I said: Off topic. <taylan>Gottox: issues of social justice can be very fierce and un-diplomatic. I'm not sure if ever in human history an oppressor class gave up its power over the oppressed through diplomacy. so, without knowing what really happened at the FSF, I'm absolutely withholding any judgment... <kyamashita>Anyone have any idea why I can't boot GuixSD on my old desktop? <kyamashita>It doesn't even look for root, I'm just dropped to a Guile prompt. <ng0>januar 2015 is not in my local archive, i can't bump this to attention. <ng0>david is dthompson here or was it davexunit? <ng0>oh, thanks. I need this for package preparation i do, taler ones :) ***jonsger1 is now known as jonsger
<kyamashita>Guix doesn't want to accept the php patch. I feel like I'm missing something obvious. <ng0>could you try to just version bump it to 7 and see if some problems are fixed that way? <kyamashita>I'm getting the "guix package: error: php: unknown package" error. <ng0>what did you do so far? just git am? <kyamashita>I think it has to do with the gnu-system.am part of the patch. That's since changed in the Guix tree? <kyamashita>Oh wait, I can do that. So I just copy the patch into a file and apply it with git am? <ng0>probably.. i'm doing gforth atm, can't help that much <ng0>it seems like nix answer is sometimes just 'screw this'.. i think gforth needs to be bootstrpped when you don't do it like that. <ng0>In file included from ./main.c:23:0: <ng0>./forth.h:108:21: fatal error: machine.h: No such file or directory <ng0> #include "machine.h" <ng0>compilation terminated. <ng0>where is machine.h coming from normaly? <ng0>would be fun to find a sourcecode location of colorforth some day. <rekado>sneek: later ask lfam I can’t seem to find the Lua patches you wanted me to test with Prosody. Could you please send me a pointer if this is still to be done? <kyamashita>Either this PHP package refuses to acknowledge the laws of computation, or I have become a fool, lol. <kyamashita>gnu/local.mk is where one declares new scheme package files and patches? <ng0>but it should build without declaring them there. <ng0>i hvae some feature brnaches where i need to go back in history and add the new files before i send patches <ng0>new emacs is released :) <rekado>I’m again suffering from this (old) Libreboot + Linux time bug. <rekado>After each shutdown my system clock is reset. <rekado>ntpd doesn’t seem to do anything <rekado>I always have to set the time manually. <rekado>I’ll upgrade to a later release of Libreboot eventually, but for the time being is there anything I should do to make ntpd work? <ng0>hm. i don't know how i solved it. i use utc everywhere, maybe that's part of the fix. <rekado>I don’t even know where I’d configure this. <rekado>(running ntpdate out of the store manually now, but that doesn’t feel right)