IRC channel logs

<str1ngs>I mean it could be something else. just from experience

<bavier1>I think I fixed the 'ledger' package build

<bavier1>eww, did I really start working on this patch in Oct 2016?

<sneek>Welcome back civodul, you have 1 message.

<sneek>civodul, ng0 says: why was there no #:log-file for the mcron service provided? are there any reasons at all or just because it didn't make sense to you at the moment you were writing it?

<civodul>no idea

<civodul>bavier1: it's been broken for a while i think

<bavier1>civodul: right, all the builds on hydra are fail

<bavier1>ok, patch sent

<bavier1>it even passes '--check'

<civodul>good job!

<civodul>ACTION -> zZz

<civodul>later!

<taylan>amz3: you're Amirouche, right? I just released bytestructures 1.0.1, fixing the bug you found :-) BTW if you face further issues, feel free to send me a direct email (linking to a guile-user thread or so if you like) to make sure I respond faster. today was a coincidence; I often skim the MLs very quickly and might miss things.

<taylan>whoops, wrong channel, was supposed to be in #guile. anyhow. off to sleep with me.

<vagrantcish>haven't updated in a few weeks, and now "guix build system /etc/config.scm" seems to no longer parse.

<vagrantcish>$ guix system build /etc/config.scm

<vagrantcish>guix system: error: failed to load '/etc/config.scm':

<vagrantcish>/etc/config.scm:23:9: /etc/config.scm:23:9: In procedure allocate-struct: Wrong type argument in\\

<vagrantcish> position 2: 3

<vagrantcish>there seem to be a couple patches related to checking mapped devices

<vagrantcish>notably 42ff7d3be642d66ba567f64882a1f2301b1a7bd9 and 4ca90ff5976434a2b6e758df38df54387ae70c1b

<vagrantcish>my /etc/config.scm https://paste.debian.net/1003937

<vagrantcish>there don't appear to be any changes to the documentation regarding this...

<vagrantcish>apparently, putting the mapped-devices all on one line worked around the issue

<vagrantcish>didn't think whitespace was significant with guile...

<davexunit>it's not

<vagrantcish>well, removing some newlines made it work: https://paste.debian.net/1003939/

<vagrantcish>suppose it's worth a bug report

<davexunit>that's just not possible

<buenouanq>guix pull on the beagleboneblack has got me into a zeno"s paradox

<buenouanq>first 30% took an hour

<buenouanq>first 60% took all day

<buenouanq>and now we're approaching the end of day 2 and are only at 93.2% (;-___-)

<vagrantc>ouch

<buenouanq>any ideas on how to speed this up? some sort of swap configuration i'm unaware of

<buenouanq>you have to use a swap on this because the bbb only has .5GB of memory

<buenouanq>I'm right now using a microSD card for the swap, might a USB flashdrive be faster?

<clacke[m]>not likely

<clacke[m]>unless, you know, it is :-)

<clacke[m]>depends on the card and the drive in question

<buenouanq>I would love to do some tests, but at a week per trial it isn't going to happen.

<clacke[m]>and the USB on the BBB

<cehteh>buenouanq: zswap or zram helps a lot

<clacke[m]>at 0.5 GB there is nothing that helps

<buenouanq>can someone possibly explain what exactly is up with guile that requires the 3GB or whatever?

<clacke[m]>I'm assuming it goes toward holding all the compiled package references in memory, and them cross-referencing each other so that a large part of them are in the active set. Someone insightful would have to explain why this happens in compilation, but just using them is ok. Is there a particular package that pulls in everything and we're usually lucky enough not to try to install that package?

<clacke[m]>Come to think of it, I haven't tried compiling guix from git on that machine. I wonder if doing that acts any different.

<buenouanq>I guess I don't really understand what guix pull is actually doing.

<civodul>Hello Guix!

<clacke[m]>buenouanq: It almost compiles a new version of guix. It's more than just the package definitions.

<buenouanq>clacke[m]: is there any way to make it not do all that? we need a `guix tug' that we can use on resourceless systems

<civodul>buenouanq: there's a wip-pull-reload branch that aims to improve on that by building things in pieces and allowing for substitutes

<civodul>it's not ready yet!

<efraim>I need to pull out my PPC laptops and see how much RAM they have, although apparently PPC G4s are also affected by SPECTRE

<efraim>s/spectre/meltdown

<civodul>yeah :-/

<mubarak>hello guix!

<mubarak>did anyone of you use his Android phone as "Mobile hotspot" as wifi device and connect to internet using wpa_supplicant.

<mubarak>does configuring Android device as "Mobile hotspot" is a little bit complicated than regular wireless devices?

<clacke[m]>it's the same as connecting to any WiFi, no complications

<clacke[m]>(generally)

<mubarak>here are my configuration: cat wpa_supplicant.conf network={ ssid="AndroidAP" key-mgmt=WPA-PSK psk="ShareMyWifi"}

<mubarak>I used "wpa_supplicant -Dcfg80211 -iwlp3s0 -c wpa_supplicant.conf" to connect

<clacke[m]>I remember back in 2011 some devicesbdidn't have proper hostap and had to make do with ad-hoc mode. But since 2012 connecting through any phone of mine has been trivial using NetworkManager, which uses wpa_supplicant under the hodd.

<clacke[m]>hood*

<mubarak>and here is the output:

<mubarak>wpa_supplicant -Dcfg80211 -c wpa_supplicant.conf -iwlp3s0

<mubarak>Successfully initialized wpa_supplicant

<mubarak>Line 2: failed to parse ssid 'AndroidAP'.

<mubarak>Line 3: unknown network field 'key-mgmt'.

<mubarak>Line 4: Invalid PSK 'ShareMyWifi'.

<mubarak>Line 4: failed to parse psk 'ShareMyWifi'.

<mubarak>Line 5: failed to parse network block.

<mubarak>Failed to read or parse configuration '/home/sahil/wpa_supplicant.conf'

<clacke[m]>ok, it's a near decade sknce I had to actually look at wpa_supplicant.conf, so I don't have anything to contribute except double-check with the manual that you got the syntax right

<mubarak>double-click? where?

<clacke[m]>manpage says key_mgmt for examome

<clacke[m]>example*

<clacke[m]>double-check: read the manual again. read your file again. verify that you typed things correctly.

<mubarak>i have read wpa_supplicant manual page & I tried many things. one time wpa_supplicant said cfg80211 unkown drive and I replace it with wext, and also it give me error

<efraim>The last time I did it by hand I don't think I used the D or i flags

<efraim>When I tether my phone with a USB cable it just shows up in nmtui as an Ethernet type device

<mubarak>I'm currently connecting to Mobile hotspot(AndroidAP) using networkmanager, on another GNU/Linux live-image but if want to install GuixSD I have to use wpa_supplicant

<mubarak>efraim: and nmtui able to configure it and connect?

<efraim>Its worth a shot, to your computer its just another wireless network

<mubarak>Yes indeed it worth. But nmtui nor networkmanager does not come with guixsd 0.14.

<mubarak>I mean guixsd 0.14 .iso

<rekado_>Hi Guix!

<mubarak>hi rekado :D

<mubarak>how are you?

<rekado_>tired but motivated :)

<mubarak>haha ;) great

<buenouanq>I've actually had problems with the 0.14 iso the last few times I've tried to use it - Had to install from the 0.13 image and upgrade.

<rekado_>buenouanq: we no longer serve substitutes for 0.13

<rekado_>what problems did you encounter?

<rekado_>please consider sending bug reports

<buenouanq>I really should.

<buenouanq>luckily, I've saved all the old .xz since 0.9 I think

<buenouanq>problem was, after burning it to a flash drive, fdisk would show an error on it about size somethings not being equal, and though you could boot to it, it would fail and drop you into a repl saying something about a missing partition maybe

<buenouanq>i'll maybe try again tomorrow and write this stuff down

<buenouanq>it was weird

<buenouanq>and at the time I just needed my machine working again so I used the 0.13 image and moved on

<rekado_>buenouanq: you don’t need to keep copies of the releases; they are all available on the GNU servers.

<buenouanq>I have no control over your servers - Local copies of things are never a bad idea.

<rekado_>yes, not a bad idea. Just saying that we do keep an archive of previous releases.

<buenouanq>though I suppose in this particular case if they are down, whether or not I have the installation image really won't matter as everything else will be gone too

<buenouanq>having my own copy was nice too as I could just pull it up right then and there, I didn't have to try and find it and redownload it

<civodul>heya rekado_!

<rekado_>Hi civodul!

<Rnytom[m]>Hi guixers

<Rnytom[m]>I have a newbie question:

<Rnytom[m]>I've been using GuixSD for a few weeks and I can no longer do 'guix system reconfigure <file>'

<Rnytom[m]>I'm getting: guix system: error: symlink: Permission denied: "/var/guix/profiles/system-3-link.new"

<Rnytom[m]>I looked the permissions of the folder and it's drwxr-xr-x 3 root root

<rekado_>Rnytom[m]: are you running “guix system reconfigure” as root?

<Rnytom[m]>no do I have to ?

<rekado_>yes.

<rekado_>try “sudo -E guix system reconfigure the-file.scm”

<rekado_>this ensures that you’re using *your“ version of guix, not root’s version of guix.

<Rnytom[m]>...indeed it's working... I think I confused package installation which I run as a user and 'guix reconfigure' which I guess I have been running as root from the start... until now

<Rnytom[m]>By the way should I always use the -E switch ? Or should I somehow update root's guix version then do reconfigure ?

<bavier>good day guix

<rekado_>Rnytom[m]: either way is fine. I usually don’t use guix as root, so I always run ‘sudo -E guix’

<Rnytom[m]>Yes it makes sense. Thanks for your help!

<roptat>and I usually use guix as root, so to update it, I have to use "guix pull" twice: for root and for my user. Then I use "guix system reconfigure" as root

<clacke[m]>rekado_: interesting about sudo -E. I have a function 'sudomy', which does 'which' on $1 and then calls sudo with that absolute path and the rest of the args

<clacke[m]>It would be nice if a second 'guix pull' could just reuse the existing output instead of compiling everything again

<efraim>Its possible, but it would be nice if it could be a flag for guix pull

<efraim>You can always download the snapshot and then run guix pull against the file for both

<efraim>But 'guix pull --latest-cached' would be nice

<davexunit>are we doing anything to address spectre and meltdown?

<efraim>That even sounds like it could be fast enough to put in a lazy user's login script

<davexunit>nvm, just found the mailing list post :)

<roptat>clacke[m]: isn't it already the case? I'm quite sure when there is no update between two guix pull, it just downloads it twice but doesn't build it twice

<efraim>But its not so easy to force it to do that

<roptat>true

<marusich>rekado_, FYI sudo -E will retain the env var HOME, so invoking a command like "sudo -E guix package -i foo" will not have the desired effect of installing foo into root's profile.

<marusich>For 'guix system' commands that's fine, but I'll bet someone will sooner or later shoot themselves in the foot by doing this.

<g_bor>Hello guix!

<marusich>Hi g_bor, and goodbye, since I need to leave :)

<marusich>Have a nice day!

<g_bor>Anyone here familiar with the installer effort?

<g_bor>I'm interested in implementation details.

<snape>g_bor: you can watch that talk: https://archive.fosdem.org/2017/schedule/event/guileappuserinterface/

<snape>g_bor: there is a wip-installer-2 branch maintained by Danny

<g_bor>Thanks!

<g_bor>Will look at that. It's just I think I have a quite powerful idea, and I would like to know, if the same occured to the current developers, or is it difficult to steer the effort into that direction.

<snape>g_bor: why don't you ask about it on the mailing list? I believe lots of people would be interested. I'm sure John would answer any technical question you have too.

<efraim>Librsvg builds on aarch64 on core-updates

<fps>hmm, how do i get unsubscribed from the debbugs mails about guix?

<Rnytom[m]>running guix pull with a user account took 2 hours and 20 minutes for me, and running guix pull as root it seems it wants to recompile everything, so I guess I'll have to skip updating the root account

<catonano`>Rnytom[m]: recently I have updated my system both for root and for my user

<catonano`>Rnytom[m]: I hhave to say that it is improved since a few months ago. Remarkably, it DIDN'T recompile everythhing for both root and the normal user

<catonano`>Rnytom[m]: so maybe once you have updated and reconffigured, thhis won't happen anymore

<catonano`>maybe

<alezost>fps: look at https://lists.gnu.org/mailman/listinfo/guix-patches

<zero21>"clocksource switched to clocksource tsc" how to fix it?

<fps>i wasn't subscribed to guix-patches..

<fps>hmm, the emails i get are just bug reports and answers to them..

<fps>sadly the mails lack unsubscribe info like mailman mails usually have

<fps>i'll write a filter to route them to /dev/null in the meantime

<fps>oh it was bug-guix

<fps> https://www.gnu.org/software/guix/contact/ had only the address to send bug reports but not a link to the mailman interface

<fps>[sent a bug report ;)]

<bavier>no compiz package yet?

<davexunit>people still use that? ;)

<bavier>idk, maybe I'm behind the times :)

<bavier>I remember using it back in the day.

<ijp>weirdly the debian packages for that got a mention on #emacs today

<bavier>was actually the thing that got people to not think I was weird running a GNU/Linux system

<davexunit>oh I had my spinning cube days as well

<amz3>bavier: I think compiz is now part of xorg/wayland

<bavier>amz3: got it, cool

<amz3>except the spinning cube is gone

<efraim>So it turns out I have two PowerBook6,7 laptops

<amz3>Rnytom[m]: if you configure both root profile and user profile to point to guix git, you will need to compile things only once, also a fix to this issues is under development

<efraim>One 12 and one 14 inch

<amz3>efraim: IMHO, one is already too many ;)

<efraim>Lol

<amz3>it's powerpc processor?

<efraim>yep

<amz3>hopefull it can compile guix

<efraim>its capable of 1.5 GB of ram so its possible

<efraim>csan<tab> proved it was possible so I might actually do it, not sure I'd put as much effort into it as aarch64

<amz3>what is goal / interest in porting guix to powerpc?

<amz3>seems like it's not supported by debian anymore

<efraim>as far as I can tell, its debian oldstable, gentoo and openbsd

<efraim>for the fun factor mostly, and i think it's one of the few laptops I have with a working battery

<Rnytom[m]>amz3: cool! Is the process of pointing a profile to guix git in guixsd's online manual ? I didn't read everything of it yet, but I can't seem to find that

<rekado_>today I add info.js to a project using texinfo. I think we should add it to the HTML version of the Guix manual.

<rekado_>*”I added“

<bavier>rekado_: looks like a good idea to me. does it clutter the texinfo source much?

<amz3>does the javascript need a review

<amz3>?

<bavier>amz3: I noticed the source doesn't include license info

<rekado_>texinfo 6.5 seems to automatically generate enough HTML for info.js to work.

<rekado_>it works by creating iframes for individual pages.

<rekado_>our stylesheet would have to be adjusted.

<rekado_>info.js is part of texinfo and under the same license, so it only needs to get a license header that librejs can understand.

<bavier>rekado_: oh, I see. Sorry, I was confused by https://introjs.com/

<rekado_>info.js adds a navigation sidebar (which I’m not so fond of), allows for navigation with the common info reader bindings, and—my favourite feature–gives access to index search with ‘i’.

<buenouanq>ok, guix pull took the better part of 3 days and I am now actually building the installation image for my bbb

<buenouanq>i've wrapped it in a time this time some i can see how long this will actually take

<buenouanq>s/some/so/

<amz3>nice

<amz3>Rnytom[m]: it's footnote 30 https://www.gnu.org/software/guix/manual/guix.html#FOOT30

<amz3>rekado_: let me know where this new info.js is and how to use it, I might want to provide additions to that js file

<amz3>rekado_: our maybe point me to the project that use info.js

<bavier>amz3: https://svn.savannah.gnu.org/viewvc/texinfo/trunk/js/

<amz3>tx

<Rnytom[m]>amz3: thanks !

<amz3>Rnytom[m]: I think compiling from git is also faster... but not sure i don't do it yet

<Rnytom[m]>I'll test this tomorrow

<amz3>rekado_: I'd like to add something like that http://hyperdev.fr/projects/reactive-menu/demo/long-read-with-table-of-content/#vectors-products or like that http://hyperdev.fr/projects/reactive-menu/demo/long-read-with-breadcrumb/#seize-holisticly-red-had

<amz3>wdyt?

<efraim>i think I have a fix for clisp on armhf and aarch64, if it works then I can try to fix clisp on core-updates

<buenouanq>failed after 49 minutes with no space left on device

<buenouanq>how do I run the guix system disk-image command mentioned https://www.gnu.org/software/guix/blog/2017/porting-guixsd-to-armv7/ to utilize a different output path (SD card)?

<amz3>read the code

<amz3>:)

<efraim>To create the disk image?

<buenouanq>yes

<buenouanq>I need it to use an SD card rather than the BBB"s eMMC as that does not have enough free space.

<buenouanq>If what I'm needing to do is mentioned in the manual, I have yet to find it.

<buenouanq>I also don't see the -e flag mentioned anywhere...

<amz3>I have two tests in tests/lint.scm failing

<amz3>with very recenter master

<amz3> https://dpaste.de/UDJP/raw

<catonano`>ok, I discovered that I was having a hard time using my qemu based vm becase it was given too little memory. Given more memory I could use a Gnome session to test my Tryton. It turns out that a postgres "tryton" role doesn't exist. That means that the activation of my system service is not working, because it should create it :-/

<catonano`>sigh

<buenouanq>is the only/best way to do this to make a build user with their home dir on said SD card, and then run guix system disk-image from there?

<buenouanq>oh no, the problem is /gnu/ itself running out of space...

<buenouanq>ok, so I've moved /gnu/ to it"s own partition

<buenouanq>let's see how this works

<buenouanq>it's chugging away, so I'll hope for the best

<catonano`>buenouanq: good luck

<buenouanq>thank you thank you

IRC channel logs

2018-01-08.log