IRC channel logs

2018-10-28.log

back to list of logs

<rekado>civodul: yeah, it’s not a detached signature. It’s only a validly signed message.
<rekado>I honestly overlooked that.
*rekado —> zzzZ
<luciddreamzzz>do you ever find problems with 'guix gc --verify' or is it more of a disk space thing
<luciddreamzzz>--verify=contents
<nckx>luciddreamzzz: Who's 'you'?
<luciddreamzzz>s/do you/does one/
<nckx>(And were you able to update LO, I wonder?)
<luciddreamzzz>LO?
<nckx>Weren't you...?
<nckx>LibreOffice
<luciddreamzzz>OK nope that was not me
<nckx>luciddreamzzz: You are right. Never mind. Scrollback yay.
<luciddreamzzz>problems like things listing as installed but not being able to run them
<nckx>To answer your question: Yes, a few times. Once self-inflicted (knowingly modifying a file in the store for giggles & profit), more than once due to failing hardware.
<luciddreamzzz>ahh, ok cool thanks. it sounds like it may take some time
<luciddreamzzz>I am not sure how to troubleshoot so much yet
<luciddreamzzz>it almost seems like my $PATH is messed up
<apteryx>hi! Is there a way to run guix bulid --check from guix-devel-mode?
<tune>is a gateway timeout on my end or guix's end? I was getting build errors that told me to use --fallback, but bow I leave it overnight multiple times to find it was stuck on a gateway timeout forever
<tune>and then I ctrl-c out of it and start updates over and it seems to happen again
<buenouanq>I had the same issues this week.
<buenouanq>found it takes 5-6 days to build Gnome and everything on a Pentium Duo btw
<efraim>Interesting, it takes less than that on my fast aarch64 machine
<Elephant454>I'm getting a gateway timeout on my end, too
<efraim>hydra is currently still down, the machine it runs on is getting upgrades
<Elephant454>Ah, ok 😊
<buenouanq>is there some sort of guix weather checker where we could check for stuff like that?
<mfzap>hi all, is mirror.hydra.gnu.org having problems? I'm getting 504's for substitutes from california
<efraim>sneek_: later tell civodul sorry, I should've done a better job with mailutils
<efraim>mfzap: hydra is currently down, the machine it runs on is getting upgraded
<mfzap>ah thanks
<efraim>sneek_: later tell civodul overdrive2 should be back up, the ssh-tunnel service didn't start at boot
<ng0>buenouanq: no
<ng0>well if it's down you should get 0 substitutes if you check it
<grall>Besides installing a wifi card with a free driver in my Thinkpad T400, I shouldn't have to do any hardware modification to run GuixSD without installing blobs, right?
<ng0>yep
<grall>Awesome.
<ng0>there's a probability that I will be moving into an entirely different direction, implementation and language with my design. if I do, I might set aside some time to send a large number of packages I found no time for sending to guix in the last year(?). I'm still contributing, my attention is just elsewhere. Oh, Buildbot is more or less functional now. We need to test it in development deployment to see if my
<ng0>rush on packaging it in the short given time was okay.
<efraim>ng0: how's it going with the taler stuff?
<ng0>what do you mean?
<ng0>development server will be moved on monday, at least the git part. I'm making some finishing moves on the server config tonight.
<ng0>the new server is already running with guixsd
<efraim>that's what I meant :)
<ng0>then we have details left, like documentation of the deployment etc
<ng0>well, I
<ng0>there's this onboarding document for - iirc - production servers which are supposed to run exchanges and what not, so this should be written with GuixSD in mind. If I'm writing nonsense right now, I have this noted elsewhere so I don't have to remember all of it
<ng0>christian's idea is that Taler should be able to deploy 2 kinds of servers with a very easy to replicate state.
<ng0>I'm sure at some point I will be motivated to write in length about it, but it's just moving from Debian to GuixSD.
<ng0>while I'm at this, I also sync changes to our side in gnunet because new gnunet servers will run comparable configs
<efraim>cool
<ng0>if you feel lucky, I have the buildbot published
<ng0>git clone https://c.n0.is/git/taler.net/guix/buildbot-channel
<ng0>heads up, it includes a bunch of python related updates, so read the code.
<tune>is there no way to update stuff while hydra is down/
<Gamayun>tune: You can use berlin.guixsd.org for substitutes instead :)
<tune>awesome, how do I go about doing that?
<tune>I might've figured it out
<Gamayun>tune: Cool! Otherwise: https://lists.gnu.org/archive/html/guix-devel/2018-10/msg00490.html
<tune>Gamayun: Where do I get the .pub file for the berlin server to authorize it? I tried moving on to using the substitute without authorizing it, and it seems to be working, so not sure if this matters.
<ng0>you don't rely on binary substitutes unless you have low resources to build. so --fallback or --no-substitutes are options
<ng0>*you don't have to
<janneke>ng0: what's `the buildbot'?
<janneke>it sounds like something like cuirass, but must be something entirely different...
<janneke>ah, it seems to be cuirass written in python, with most probably a whole different feature set/interface
<Gamayun>tune: It should already be under /share/guix in the current system as I recall..
<tune>Gamayun: I haven't got a /share folder that I know the location of. I just haven't done any of this before.
<Gamayun>tune: I'm installing atm, so mine is under /run/booted-system/profile/share/guix .. Can't remember where it is on an installed system or with guix under a foreign distro...
<tune>ah, found it. thank you
<tune>/run/current-system/profile/share/guix
<Gamayun>Great :)
<ng0>janneke: CI
<janneke>ng0: ok, a vaguely cuirass/hydra/jenkins link thing
<janneke>*like
<g_bor>hello guix!
<g_bor>I've just got this: Backtrace:
<g_bor> 3 (apply-smob/1 #<catch-closure 692300>)
<g_bor>In ice-9/boot-9.scm:
<g_bor> 705:2 2 (call-with-prompt _ _ #<procedure default-prompt-handle?>)
<g_bor>In ice-9/eval.scm:
<g_bor> 619:8 1 (_ #(#(#<directory (guile-user) 747140>)))
<g_bor>In guix/ui.scm:
<g_bor> 1583:12 0 (run-guix-command _ . _)
<g_bor>guix/ui.scm:1583:12: In procedure run-guix-command:
<g_bor>Bad Read-Header-Line header: #<eof>
<g_bor>substitution of /gnu/store/h5r6bbj22rc74g7qmkhzi3g96l1khfmw-mercurial-4.7.2 failed
<g_bor>killing process 12996
<g_bor>guix package: error: build failed: some substitutes for the outputs of derivation `/gnu/store/0y6h2z162aynlw2bdrhb5d0ddi2p0kv2-mercurial-4.7.2.drv' failed (usually happens due to networking issues); try `--fallback' to build derivation from source
<g_bor>any idea?
<g_bor>hydra mirror gateway timeouts on ghostscript source instead :-(
<efraim>Add --keep-going and it'll try the next source
<cbaines_>I think https://berlin.guixsd.org/ is down at the moment, as I'm seeing the same issues
<rekado>oh?
*rekado checks
<rekado>berlin is up
<rekado>I can connect via ssh
<rekado>but cuirass…?
<rekado>cuirass is stopped and disabled
<rekado>ah, out of space…
<rekado>running “guix gc -C 1G”
***felicien_ is now known as felicien
<rekado>it’s still deleting things
<rekado>I’m using “guix gc -F 5G” to have a buffer of 5G now.
<rekado>I’ll look into removing older stuff later
<cbaines_>thanks for fixing it :)
<rekado>space is shrinking fast, though
<rekado>I’m reluctant to remove more, because hydra is still down.
<rekado>I’ll do “guix gc -C 10G” now to buy us some more time.
*rekado goes afk
<g_bor>thanks for the info
<XWill1>Hello
<XWill1>I'm a new user of GuixSD, and i wonder if someone got a config.scm for a thinkpad x200 tablet with X and stylus working so i can use it as a base
<janneke>XWill1: did you already install GuixSD? I advice for the initial instal to use the `bare bones' template and create the real system once GuixSD is installed.
<XWill1>yes with the desktop tempate
<XWill1>I expect to make the real system from the desktop template
<Laalf>XWill1: do you know what kernel modules/other drivers the stylus uses? afaik gnome just works with the thinkpad tablets
<XWill1>i think it's wacom, with the desktop template with gnome, stylus does not work
<Laalf>XWill1: we do have xf86-input-wacom. your xorg conf might need some adjustments though
<Laalf>you might also want to install libwacom if that is not a dependency.... and maybe get rid of thinkpad tablets since they are far from sturdy:D
<derpsky>is there a list of mirrors? I'm having a 504 error when downloading from mirror.hydra.gnu.org
<Laalf>derpsky: hydra is offline. berlin.guixsd.org exists but doesnt have everything build atm
<derpsky>ah alright thanks Laalf i'll try that
<XWill1>thank you Laalf i will start with that
<luciddreamzzz>how to make my guild more faster
<luciddreamzzz>s/guild/build/
<luciddreamzzz>can i skip tests
<Laalf>luciddreamzzz: afaik only by modifying the packages. if you know any other way, lemme know
<luciddreamzzz>hmm that would be nice to put in a config maybe
<g_bor>luciddreamzzz: this has been discussed several times, and it was decided that this would not be well aligned with the goals of providing strong guarantees. It is also much less of a problem, when substitutes are available.
<g_bor>In the development phase, when I need that I usually define a no-tests package inheriting from the original, with test disabled.
<g_bor>Good night!
<luciddreamzzz>ahh OK good explanation g_bor, thank you.
<luciddreamzzz>makes sense
<Laalf>just bought a new notebook today to build python.
<luciddreamzzz>Laalf, nice! A development environment is pretty demanding
<Laalf>luciddreamzzz: was meant as more of a joke. I need more than 16gb of ram. I doubt I can build python even with 32.
<luciddreamzzz>wow had no idea python took that much to build hmm
<luciddreamzzz>i was thinking a notebook wasnt optimal but wasn't going to say anything :)
***jonsger1 is now known as jonsger
<luciddreamzzz>I have laptop that is plenty fast but source distros could get painful nonetheless
<luciddreamzzz>doesnt stop me though
<Laalf>I do have a server I plan to use for offloading. But not now since hydra is still offline
<Laalf>And I can't build python on it either. I can't install guilw-ssh for some reason without compiling some stuff that break my ram
<rk4>is this a viable idea, or one someone has had before: have the package origin git-fetch+uri-fetch be overrideable to query some centralized store of source code. that makes offline use feasible, if you have the store
<rk4>then as another effort, have a means to generate the centralized store by searching through the package definitions and downloading sources
<rk4>it's a little bit different to pulling down pre-built packages [substitutes], but i think it involves trusting less internet connected systems
<atw>using Guix has spoiled me. Now trying to use Docker at work is really hard