IRC channel logs


back to list of logs

<rekado_>can you show us the error?
<PotentialUser-79>one moment
<PotentialUser-79>how do i copy the error of the other machine?
<PotentialUser-79>i can't
<rekado_>what command did you run to get the error?
<rekado_>you may not be working on the same file
<rekado_>because the error does not match your latest paste
<ngz>I tried to update X-Moto package (see <>), it builds, it runs, but I encounter rendering issues, ("unable to render the biker" while playing). Could someone tell me if they observe the same problem?
<rekado_>PotentialUser-79: you should remove (dependencies mapped-devices)
<rekado_>the core-updates merge should have made Guile 3 the default, but there are quite a few Guile packages that still use Guile 2.
<PotentialUser-79>we have lift off!!! hank you
<lfam>Woop woop!
<PotentialUser-79>in the end I did a manual install ignoring the tutorial. I have pepared partition with another OS, booted guix, connected to wifi via dongle so I could use same network as my laptop, started ssh service on guix and done the rest running commands over ssh and editing the config over a shared folder using sftp
<PotentialUser-79>copying to /mnt about 50%
<PotentialUser-79>oh no!
<PotentialUser-79> could this be a solution?
<PotentialUser-79>i got it
<PotentialUser-79>it should be sda not sda1
<PotentialUser-79>what doi do if I need to re run the guix system init
<rekado_>you can just re-run it.
<PotentialUser-79>I have started all over again
<blaze_cornbread>hello, I have guix installed and would install and update packages via tor
<PotentialUser-79>another disappointment guix system: warning: at least 3,258.1 MB needed but only 648.2 MB available in /mnt
<PotentialUser-79>guix system: warning: at least 3,258.1 MB needed but only 648.2 MB available in /mnt
<PotentialUser-79>guix system: warning: at least 3,258.1 MB needed but only 648.2 MB available in /mnt
<PotentialUser-79>now I am running out of disk space
<anadon>Where is civodul? I need help tracking down a bug report.
<bandali>sleep, probably :-)
<ryanprior>PotentialUser-79: how much disk space do you have? Guix is pretty disk-intensive.
<PotentialUser-79>I tried with another partition 6.6 G used so far still installing, i guess i might have missed a step ind it was trying to install itself on pendrive
<lfam>Probably missed the step where you start the cow-store service
<PotentialUser-79>bootloader installed
<PotentialUser-79>what now?
<PotentialUser-79>I will reboot
<PotentialUser-79>ok, boot work, good night thank you for your help
<lfam>Hey gagbo
<gagbo>Hi :)
<lfam>Glad to hear it, PotentialUser-79!
<lfam>Alright, gagbo, the basic workflow for patching Guix packages is to clone our Git repo, build Guix (basically, `make`), and then edit the package recipe for guile-sdl2 to use your patch file
<gagbo>It's pretty late here, I assumed I would just come and ping david in guile, but if I can quickly clone and use my local copy of guile-sdl in my guix-profile / environment that'd be great
<lfam>gagbo: These steps are covered in our manual, section Building From Git and the next section, Running Guix Before It Is Installed
<lfam>The first time you build Guix, when you have to build the whole thing, can take a while, so maybe you should leave it for tomorrow
<gagbo>I already have a "custom" version of guile-sdl2 because I started using guile3.0 before it was really out in guix packages so I have a "patched" define-package version to build against guile3.0
<lfam>But once you do, you can use the Guix you've built from the Git repo, and edit the package, and quickly test the build and deployment of your edited package
<lfam>Oh, cool
<gagbo> which is this
<gagbo>So I suppose I can put my patch somewhere here directly
<lfam>Yup! You can redefine the source field of the package and include your patch there
<gagbo>Ok, I'll try that then, thanks :)
<lfam>Here's an example of using patches in a custom package repo:
<lfam>The main thing to look at is where the actual patch files go, which is at the top level of the repo
<lfam>I'm not sure if you are using a Guix channel or the older GUIX_PACKAGE_PATH mechanism, and tbh I haven't converted my custom packages repo into a channel yet, so I don't know if it will translate to channels or not
<lfam>Let us know if you have any questions!
<gagbo>I'm just using guix package -f path/to/package.scm :)
<lfam>Cool :)
<anadon>What is the bot's name for delayed messages?
<vagrantc>$bot later tell $nick $message
<vagrantc>i think
<vagrantc>sneek: later tell anadon how to later tell anadon
<vagrantc>sneek: botsnack
<anadon>sneek: later tell civodul What is the bugtracker number for the package.json timestamp issue on the website?
<sneek>Welcome back anadon, you have 1 message!
<sneek>anadon, vagrantc says: how to later tell anadon
<anadon>sneek: botsnack
<vagrantc>never quite replied ... wonder if it got the message
<vagrantc>it was so busy remembering what to tell you...
<anadon>I got the reply?
<lfam>anadon: Is it #37207?
<vagrantc>anadon: sneek didn't ACK your message with "Okay."
<shmiggles>Are there any guides/resources on how to set up a Guix build farm?
<lfam_>shmiggles: The code we use to run the infrastructure is contained in our maintenance.git repo:
<lfam_>Specifically, in the Hydra folder (we used to use a program called hydra to run the build farm)
<lfam_>berlin is the head node of the primary farm
<shmiggles>Thanks, Ifam_!
<dftxbs3e>it's crazy the amount of packages one has to build to simply *install* hello - just building it require much fewer but installing, that's a different story.
<lle-bout`>hey, spacemacs's really good for an emacs beginner like me :-)
<reepca`>dftxbs3e: installing *should* require strictly less packages than building - if there are substitutes only the runtime dependencies need to be downloaded, if there aren't it's still the same amount as building.
<lle-bout`>I wish the packages spacemacs manages could be handled with GNU Guix. I wonder if it's possible to transparently stick GNU Guix somewhere in packages.el
<lle-bout`>reepca`: I'm building on a new platform so substitutes arent yet available
***apteryx_ is now known as apteryx
<blaze_cornbread>hello, I have have installed guix onto my existing os and would like to install packages from guix using tor
<lle-bout>blaze_cornbread: GNU Guix does not have Tor endpoints for the git or substitute server, however, you can probably, after Tor is installed and started, prefix every GNU Guix command with torsocks, but I would not rely on it so heavily to not leak anything.
<blaze_cornbread>lle-bout: ok
<blaze_cornbread>how can I setup guix so I don't have to run these commands each time?
<blaze_cornbread>hint: Consider setting the necessary environment variables by running:
<blaze_cornbread> GUIX_PROFILE="/home/ryan/.guix-profile"
<blaze_cornbread> . "$GUIX_PROFILE/etc/profile"
<blaze_cornbread>Alternately, see `guix package --search-paths -p "/home/ryan/.guix-profile"'.
<blaze_cornbread>lle-bout: when I install packages with torsocks such as "torsocks guix install nnn" it downloads at speeds faster than tor
<blaze_cornbread>I was told I can run the daemon w/ tor but I do not know how to do this
<lle-bout>blaze_cornbread: I'm not sure whether it is the daemon or client that downloads, maybe it is the daemon. Either way, run both to see - and about the GUIX_PROFILE thing, I'm not sure. I don't run them, it always worked.
<xelxebar>blaze_cornbread: Make sure that guix-daemon also has GUIX_PROFILE set correctly.
<lle-bout>If you use systemd as a service manager, you can try to modify the service, you can get its content and path with "systemctl cat guix-daemon"
<lle-bout>I'm not sure if the service name is guix-daemon or guix alone, so try both
<lle-bout>You can modify the command used to launch the daemon, prefix it with torsocks
<blaze_cornbread>lle-bout: it is guix-daemon
<bricewge>blaze_cornbread: Looks like you just need to set the environment variable “http_proxy” of the guix-daemon to a proxy using tor
<blaze_cornbread>bricewge: i don't know where to start doing this
<blaze_cornbread>bricewge: I have seen it in the manual
<xelxebar>blaze_cornbread: What foreign OS are you on?
<blaze_cornbread>xelxebar: pureOS
<bricewge>As lle-bout said tweak your service file for guix-daemon
<xelxebar>Dose pureos use systemd?
<bricewge>blaze_cornbread: ?
<blaze_cornbread>xelxebar: yes
<blaze_cornbread>bricewge: i do not know how to do that
<xelxebar>Okay, are you familiar with systemd service files?
<blaze_cornbread>xelxebar: no
<bricewge>blaze_cornbread: sudo systemctl edit guix-daemon.service
<lle-bout>It's true that Tor now exposes an HTTP proxy by default, so you should be able to use it that way, port is 8118 by default IIRC.
<lle-bout>It should be cleaner, since it's documented.
<blaze_cornbread>lle-bout: is it 9050 or 9150?
<bricewge>Then add en try “[Service]
<bricewge> Environment="http_proxy=http://localhost:1337"”
<xelxebar>The tor proxy typically defaults to port 9050 btw
<bricewge>Change the proxy value to your own
<blaze_cornbread>xelxebar: ty
<lle-bout>blaze_cornbread: The SOCKS5 proxy listens on the first when in daemon mode and the latter when started by Tor Browser - but the HTTP proxy uses neither of those
<blaze_cornbread>bricewge: ok
<bricewge>BTW the proxy have to be a HTTP one not a SOCKS5
<blaze_cornbread>bricewge: this can be done using privoxy?
<xelxebar>I dunno. Instead of mucking about with http_proxy, why not just edit the service file to launch via torsocks?
<xelxebar>That should pretty much Just Work
<blaze_cornbread>xelxebar: does not matter to me which ever is more secure
<blaze_cornbread>xelxebar: do I prefix ExecStart w/ torsocks?
***dingenskirchen1 is now known as dingenskirchen
<xelxebar>Probably want to prefix with the full path to torsocks, but that should pretty much do it.
<xelxebar>Anyway, gotta run!
<blaze_cornbread>xelxebar: ty
<blaze_cornbread>bricewge: lle-bout: I have used port 8118 how can I confirm if it using tor?
<bricewge>blaze_cornbread: What route did you choose torsocks or http-proxy?
<blaze_cornbread>bricewge: right now http_proxy
*bricewge is downloading blender substitutes from tor
<bricewge>You have a tor service running with the default settings?
<civodul>bricewge: through privoxy?
<bricewge>civodul: Yes
<civodul>i wonder if there's a simpler proxy we could recommend
<bricewge>blaze_cornbread: You need a config file for privoxy with “forward-socks5 / localhost:9050 .”
<rekado>reepca` dftxbs3e Installation can require *more* packages to be downloaded because of profile hooks.
<blaze_cornbread>bricewge: ok one moment
<bricewge>civodul: I look around a few weeks ago, privoxy seems the most straight forward
<rekado>1.9T free space on
<rekado>the guix publish cache takes 4.9T
<bricewge>blaze_cornbread: Then start privoxy with the config “privoxy my.conf”
<divansantana>I want to submit a patch to add another package for emacs-exwm . The current one is bound to the version in github release page. I'd like to add another to a particular newer git commit. Would the name for this be like emacs-exwm-next or what is the convention?
<blaze_cornbread>bricewge: my first time using privoxy where do I create the config and what is the file name?
<divansantana>May want to do the same for telega.el and tdlib.
<ngz>divansantana: is there a strong reason to provide a new package? I.e., would it make sense to replace current emacs-exwm with newer one? (e.g., it fixes an important bug).
<civodul>rekado: last i checked there were few disk-image* files in /var/cache/guix/publish, but maybe worth rechecking
<blaze_cornbread>bricewge: i think i understand
<bricewge>blaze_cornbread: I'm a noob using it too. privoxy don't care about anything, just put it in a file and give it the file name
<blaze_cornbread>bricewge: okay
<ngz>divansantana: in some Emacs packages, we prefer non official releases, but the reason why must be explained in a comment.
<blaze_cornbread>bricewge: i am running privovy w/ "forward-socks5 / localhost:9050 ."
<bricewge>blaze_cornbread: So when using guix-daemon with http_proxy=http://localhost:8118 it will connect to privoxy's HTTP proxy that will forward it tor's SOCKS5 proxy
<blaze_cornbread>bricewge: okay
<blaze_cornbread>bricewge: I sitll have fast download speeds
<blaze_cornbread>bricewge: where and how can "--substitute-urls=http://bp7o7ckwlewr4slm.onion" be added to confirm tor is working?
<bricewge>blaze_cornbread: Stop your tor service, the download should end
<blaze_cornbread>bricewge: okay
<bricewge>blaze_cornbread: Here is how I do it in Guix
<bricewge>herd is the equivalent to systemctl
<blaze_cornbread>bricewge: okay stopping tor does not stop the download I will try to follow ur link
<bricewge>blaze_cornbread: Here is what happen when I stop tor
<bricewge>blaze_cornbread: Can you share what returns “systemctl cat guix-daemon.service”?
<reepca`>rekado: ah, okay, makes sense.
<civodul>bricewge: great summary, we should add it to the cookbook!
<divansantana>ngz:Can one _add_ like emacs-exwm-next or emacs-exwm-dev . Sometimes it's nice to have an easy way to install the latest or the current release. Have both options.
<bricewge>civodul: I was looking into it, but I don't really know where to put it there.
<bricewge>Doing the same with SSH forwarding could be usefull too.
<blaze_cornbread>bricewge: sudo systemctl set-http-proxy guix-daemon http://localhost:8118
<blaze_cornbread>Unknown operation set-http-proxy
<bricewge>blaze_cornbread: Yup that wont work
<bricewge>The systemctl edit and systemctl restart on guix-daemon you did is the equivalent
<bricewge>Your config seems correct
<blaze_cornbread>bricewge: ok
<bricewge>Did you restart guix-daemoin
<blaze_cornbread>bricewge: i think so
<mothacehe>rekado: I just pushed a commit to maintenance, that might improve a tiny bit the storage situation.
<ngz>divansantana: You're not answering my question :) But, to answer yours, I think that, pragmatically, we cannot offer to keep up with upstream, and update packages every time they commit something. IOW, I doubt emacs-exmw-dev makes sense in Guix proper (but it should be easy enough to add the package on your side). For emacs-exwm-next, sometimes it makes sense, sometimes it doesn't. There should be a good reason to add a ...-next
<ngz>of a package, since we cannot do it for all of them.
<blaze_cornbread>bricewge: does not work
<bricewge>blaze_cornbread: Where? Please share the output of your commands in the paste service.
<rekado>mothacehe: thank you!
<blaze_cornbread>bricewge: guix works but when stopping tor service the download continues
<blaze_cornbread>bricewge: guix configured to use port 8118 when stopping tor service the download continues
<mothacehe>rekado: I uploaded my key but I can't login to berlin. Has it been reconfigured recently ?
<bricewge>blaze_cornbread: Looks like guix-daemon doesn't have the http_proxy variable set...
<bricewge>blaze_cornbread: get the PID of guix-daemon with “systemctl status guix-daemon”
<blaze_cornbread>bricewge: I have it
<blaze_cornbread>the PID
<bricewge>Then sudo cat /proc/PID/environ
<bricewge>You should see the http_proxy in there
<blaze_cornbread>bricewge: it is there
<bricewge>pkill privoxy
<bricewge>Can you share the content of the environ file?
<bricewge>Ohhh, mine contains http_proxy and https_proxy!
<blaze_cornbread>bricewge: share the whole environ file?
<bricewge>Try adding that one to your service file too with “systemctl edit“
<bricewge>My environ file
<bricewge>blaze_cornbread: Yes
<blaze_cornbread>bricewge: how do I add the https_proxy?
<bricewge>blaze_cornbread: Like you did for “http_proxy“: systemctl edit, systemctl restart
<blaze_cornbread>bricewge: do I make another line?
<bricewge>blaze_cornbread: Yes
<bricewge>blaze_cornbread: I have to go cook lunch. Good luck!
<blaze_cornbread>bricewge: Thanks!
<ngz>divansantana: note that I don't use, or even know, emacs-exwm. So I have no strong opinion on emacs-exwm-next. I am merely asking about the reason.
<blaze_cornbread>bricewge: did not work maybe I am not configuring it right?
<reepca`>(port-revealed (open-input-file "/dev/null")) => 32539
<reepca`>hmmmmm... that doesn't seem quite right
<civodul>reepca`: oh!
<civodul>same here on 3.0.2
<civodul>could you email that to
***sirmacik_ is now known as sirmacik
<rekado>mothacehe: I’ll reconfigure once ‘guix pull’ finishes
<rekado>(SQLite database is busy)
<lle-bout>bricewge, blaze_cornbread: Tor in recent versions now features an HTTP proxy natively, no need for privoxy. sneek later tell
<lle-bout>sneek later tell blaze_cornbread Tor in recent versions now features an
<lle-bout> HTTP proxy natively, no need for privoxy
<lle-bout>sneek later tell blaze_cornbread Tor in recent versions now features an HTTP proxy natively, no need for privoxy.
<sneek>Got it.
<lle-bout>Got to love Emacs yanking that also yanks lines heh, ERC mode lacks context awareness for yanking I guess
<srandom>I updated my system using guix pull && guix package -u, but the python-parso build failed because the check in the build failed, what should I do
<civodul>hi srandom
<civodul>srandom: could you report the failure to
<reepca>civodul: done
<civodul>srandom: core packages were updated, and perhaps this particular package didn't like the upgrade
<reepca>I'm mildly curious how it wasn't detected for so long...
<civodul>it'd mean we're leaking all the ports, no?
<reepca>all the ports not explicitly closed, aye
<civodul>which in practice includes bytevector ports and the likes
<blaze_cornbread>lle-bout: which version of tor uses http proxy & which port does it use?
<sneek>blaze_cornbread, you have 2 messages!
<sneek>blaze_cornbread, lle-bout says: Tor in recent versions now features an
<sneek>blaze_cornbread, lle-bout says: Tor in recent versions now features an HTTP proxy natively, no need for privoxy.
<civodul>reepca: actually no, it's only file ports, silly me
<srandom>civodul, ok
<reepca>and I can confirm file descriptors do get leaked when not explicitly closed. I know this because fibers currently doesn't explicitly destroy its peer schedulers (bug), which is why I discovered this in the first place
<civodul>bah, good catch!
<civodul>2.2.7 has the same problem
*civodul tries add SCM_REVEALED(port) = 0 in scm_i_fdes_to_port
<sirmacik>hey guix!
<sirmacik>is anybody using vpnc with nm?
<sirmacik>I'm getting error insufficient privileges while adding new connection via nm-connection-editor (not a gnome user)
<wxie>civodul: 'make check' stucks at PASS: tests/import-utilsscm. What's the next check?
<rekado>wxie: tests/inferior.scm
<rekado>you can see this in
<rekado>but this won’t help you when parallel tests are enabled
<wxie>rekado: thanks. How long will it take to finish the tests?
<wingo>wow, wondering why the configure for bash is taking forever, then i see that it's guile / gash that's running configure :mind_blown:
<guix-vits>some aside: Is anyone had any troubles with "creating temporary rules on command line using `nft`"?
<guix-vits>`uname -r` -> 5.4.38-gnu
<apfel>hi there, i want to use cuirass to buld some of my packages. During system reconfigure the build of cuirass failed during the testing stage: - (unbound-variable #f "Unbound variable: ~S" (current-processor-count) #f) - any hints or ideas?
<guix-vits>apfel: what if: try install it not system-wide?
<apfel>guix-vits: well, i want to run it as a normal system service, like everything else. Why do you suggest i should not run it as a system service? I tried to just build it with guix build, but i got the same result.
<rekado>current-processor-count is defined in (ice-9 threads)
<rekado>so perhaps Cuirass is missing an import of that module?
<rekado>apfel: can you tell us what version of Guix you’re using?
<civodul>rekado: yes, it's in (ice-9 threads) only on 3.0
<civodul>on 2.2 it was available globally but deprecated
<apfel>rekado: i ran guix pull an hour ago, as well as guix package -u, so it should be the most recent version?
<rekado>apfel: yes, looks like we need to fix cuirass
<apfel>rekado: guix system (GNU Guix) cd23eb067eacce112214944a7f0fecd1af1facb5
<rekado>Guix very recently merged in a bunch of significant changes including the use of Guile 2 to Guile 3 as the default Guile
<rekado>Cuirass hasn’t been updated yet to work correctly with Guile 3
<rekado>Cuirass commit ae8c067d19aa4d544e5aae0abacbfb5e56a1a218 already includes the required change
<rekado>so we just need to update the Cuirass package in Guix
*rekado does that now
<apfel>rekado: ah ok nice, thank you
*rekado waits for fibers tests to complete…
<rekado>apfel: pushed an update with commit 09db2343f6
*rekado reconfigures
<guix-vits>apfel: i'd not supposed, just 'tinkering'.
<apfel>rekado: reconfigure works, thanks again
<wxie>rekado: in "", the link is missing. I think this is why guile-dbd-mysql is not available in guile-xyz.scm. Actually the guile-dbd-mysql-2.1.6.tar.gz file can be downloaded from as part of How could I upload it to
<rekado>wxie: I don’t understand. What is the problem you have?
<rekado>there’s no link between Guix and
<wxie>rekado: the is not available.
<rekado>yes, but where does that URL get used in Guix?
<bandali>wxie, went offline for good a few years ago. feel free to come by #fsf on friday noon (12pm ET) for Free Software Directory meeting, and propose an updated link for use in the above page you linked
<wxie>No, it is not. But if I want to add guile-dbd-mysql to the guile-xyz.scm as a package, I need the download link, right?
<wxie>bandali: ok.
<rekado>wxie: you need a URL to download sources, yes, but you already said that it’s available on Github, so I don’t see the problem.
<wxie>rekado: understood.
*rekado updates all R packages again
<rekado>I think we need someone who would be willing to keep our Python packages up to date on a regular basis
<guix-vits>ok, if someone will notice temporary rules not working with `nft`: sneek.
<divansantana>ngz: Sometimes there is a commit ahead of a release which is useful. Last time it was a major bug fixed, which I wanted to use. With telega.el, I find I just want to try out the lastest code to try it out.
<ngz>divansantana: IMO, the "major bug fixed" is a good reason to upgrade emacs-xxx instead of providing emacs-xxx-next. For the "I want to try it out", I suggest to define a "emacs-xxx-dev" package locally. The "useful commit" reason really depends on how useful it is. There is no hard rule about it. If it is, e.g., very useful, I could be a reason to upgrade emacs-xxx. If it is not so useful, well…
<rekado>looks like custom GCCs are broken since the core-updates merge
<rekado>one example:
<rndd>hi everyone, i described (specifications->manifest) with sbcl and libx11 but when i'am trying to load common-lisp package trivial-gamekit with quicklisp it gives me an error " cannot open shared object file: No such file or d...". Does anybody had something similar?
<rekado>rndd: does it work when you set LD_LIBRARY_PATH to where is located?
<civodul>rekado: prolly CPATH vs. C_INCLUDE_PATH, no?
<rekado>haven’t been able to try fixing it yet
<rndd>rekado: i put "/gnu/store/q9y0km8682v0x5rwjvk2pd1w9v06bws6-profile/lib/gnu/store/q9y0km8682v0x5rwjvk2pd1w9v06bws6-profile/lib" in LD_LIBRARY_PATH and then in C_INCLUDE_PATH. In both variants it doesn't work
<bricewge>Writing the cookbook entry about tor and guix-daemon, I wonder if it's a good idea to write about it.
<rekado>rndd: really without a colon?
<bricewge>Since guix-daemon is using a http proxy the DNS query doesn't go trough tor at all
<bricewge>It may give a sense of false security
<rndd>rekado: you mean, LD_LIBRARY_PATH=/gnu/store/q9y0km8682v0x5rwjvk2pd1w9v06bws6-profile/lib/gnu/store/q9y0km8682v0x5rwjvk2pd1w9v06bws6-profile/lib
<rndd>ye, it too
<raghavgururajan>Hello Guix!
<bricewge>Setting a dns resolver to use tor, isn't straight forward and has some drawback
<bricewge>Maybe it's better to use torsocks in the end
<bricewge>Hey raghavgururajan !
<jonsger>mysql-5.7.27 fails for me on master
<rekado>rndd: you’re missing a colon
<rekado>(I apologize, this sounds wrong)
<rekado>(why is it duplicated?)
<rekado>LD_LIBRARY_PATH=/gnu/store/q9y0km8682v0x5rwjvk2pd1w9v06bws6-profile/lib should be fine
<rndd>rekado: checked it. nope. don't work
<jonsger>it's because of boost-for-mysql. Working on a patch
<civodul>guix pack -RR over fakechroot got nicer thanks to codonell:
<civodul>now it supports dlopen
<civodul>neat hack
<buffet>morning guix, is there a way to get a debug version of a library?
<janneke>oh nice!
<bricewge>sneek: later tell xelxebar: You were right about not “mucking about with http_proxy”, using it will query DNS in clear. Better to use torsocks as you suggested.
<sneek>Got it.
<bricewge>sneek: botsnack!
<bricewge>buffet: Hey, see the docs
<buffet>bricewge, im talking about a something that doesnt use the gnu build system and that doesnt have a debug output
<civodul>buffet: by default, when there's no "debug" output, debugging symbols are stripped
<civodul>if you want debugging symbols for a package, the easiest way is to add a "debug" output
<civodul>whether or not it uses 'gnu-build-system'
<buffet>is there a nice way to achieve that?
<buffet>besides just copying the entire (package ...) thing over?
<buffet>just copied it over and added `(outputs '("out" "debug"))`
<buffet>now the build fails because something doesnt have access to /gnu/store/...//lib/debug/gnu/bla.debug
<Guixguy>Hey Guix, I am trying to pin flatpak to 1.4.3 but in my config.scm when I put flatpak@1.4.3 under (map specification->package, it says it can't find the package
<Guixguy>What am I not getting?
<alextee[m]>is that even supposed to work? i think you can only pin commits not specific package versions, unless you put them in your channel or something.
*alextee[m] is curious to find out now
<pkill9>Guixguy: that would look for a flatpak package at that version, it doesn't pin the version
<pkill9>if that package at that version isn't pakcaged, then it won't find it
<pkill9>you need to use inferiors to pin the packages like that
<Guixguy>pkill9: OK, thank you
<Guixguy>When I did an upgrade last night it moved flatpak from 1.4.3 to 1.6.3 and it has a bug
<jonsger>inherit is strange
<pkill9>i would install packages like flatpak to my user's profile, but if you want to put it in the configuration then see
<pkill9>you'll want to use a guix commit with the previous flatpak version
<Guixguy>OK. I rolled back a generation and used guix describe to get the commit
*jonsger filed for boost-for-mysql
<jboy>If I modify-phases and use add-before to set an environment variable before the configure stage, that variable should still be defined during the config stage, right? (I'm using meson-build-system.)
<ngz>jboy: yes
<jboy>ngz: thanks. then it's weird that meson isn't picking up on it..
<labancle>I have guix installed on a foreign distro. I want to play with guix in guile, but I'm not sure what I need to add to the %load-path
<labancle>there are no .scm files beneath my $GUIX_PROFILE
<labancle>sorry, my mistake there are plenty of .scm files in my guile installation, but no guix.scm file
<labancle>In $_GUIX_PROFILE I can find 3 different guix.scm files
<labancle>what do I need to add to guiles %load-path at the REPL in order to use guix?
<rndd>how to install libgl on guixsd?
<gagbo>How can I change the %patch-path for one command ? I was just using a "package.scm" file before, and I have a patch I'd like to apply but "search-patches" doesn't find my patch
<catonano>rndd: what do you need to do, exactly ?
<catonano>ah I just read what libgl is
<rndd>and some of
***slyfox_ is now known as slyfox
<cbaines>rndd, searching my system suggests the mesa package has it
<ecbrown>i'm taking the recent commit of cuirass for a spin, using a reduced variant of what is in the manual. getting a permission denied in /var/log/cuirass.log
<ecbrown>(don't even know if i am doing this right.) i'm in ~root/src/guix where i issue ./pre-inst-env guix system reconfigure /etc/config.scm where this file is "doctored up"
<ecbrown>(permission denied upon a mkdir) not even (user root) can barrel through it
<ecbrown>(user "root")
<rndd>cbaines: thanks, it helped
<cbaines>ecbrown, the permission issue could be Linux trying to tell you that it can't do that. I think Cuirass trys to create a gcroots directory.
<cbaines>ecbrown, does /var/guix/profiles/per-user/cuirass/ or /var/guix/profiles/per-user/cuirass/cuirass exist on your system?
<g_bor[m]>cbaines: hello!
<rndd>cbaines: how you find out that package mesa contains it?
<cbaines>rndd, I just searched /gnu/store on my system. I could have easily missed a package, since I don't have everything installed
<cbaines>g_bor[m], hey o/
<g_bor[m]>It was just brought to my attention that mentors should double check that they signed the mentor agreement. Could you do that?
<cbaines>g_bor[m], I believe I have. Do you know how best to check?
<g_bor[m]>I think it should be available at your outreachy dashboard.
<rekado>buffet: you don’t need to copy everything. Use (package (inherit the-other-package) …) instead
<cbaines>g_bor[m], I can see a button "Export Mentor Contract"... I guess that's a yes
<ecbrown>cbaines: one sec
<ecbrown>cbaines: neither exist. just root and ecbrown
<buffet>rekado, oh thats a lot nicer
<cbaines>ecbrown, hmm, which user are you trying to run cuirass as?
<ecbrown>cbaines: i am placing it in config.scm, and reconfiguring. i believe that it defaults to user "cuirass" though i have tried to insert (user "root") for good measure
<buffet>still errors due to missing persmissions on the .debug file (perms are 400)
<ecbrown>wonder if it's mkdir'ing "" or something
<ecbrown>or #f
<cbaines>ecbrown, I think it tries to mkdir /var/guix/profiles/per-user/cuirass/cuirass, but that'll fail unless /var/guix/profiles/per-user/cuirass exists
<cbaines>ecbrown, maybe try creating /var/guix/profiles/per-user/cuirass/ owned by cuirass
<ecbrown>cbains: *boom*
<cbaines>it works now?
<ecbrown>well, it doesn't crap out. now i am seeing "next evaluation in 60 seconds" written to cuirass.log
<cbaines>that sounds good :)
<ecbrown>wonder if now i need to tweak the cuirass spec
<cbaines>Out of interest, why do you want to run Cuirass?
<ecbrown>cbains: first to build guix. second for a $work project
<ecbrown>(a CI system for my own packages)
<sammich>Hey, when using the graphical installer I get kicked back to the start after it finishes formatting, is there anyway to fix this?
<sammich>It appears to have worked when I chose manual partitioning
<eta-expanse>can anyone point me in the right direction of setting up an emacs daemon? I expected a service but could not find one
<raghavgururajan>Does anyone use flatpak?
<raghavgururajan>When I try to add remote, I get seg fault.
<raghavgururajan>ecbrown, Guix System?
<raghavgururajan>rg@secondary ~$ flatpak remote-add --if-not-exists flathub
<raghavgururajan>Segmentation fault
<cbaines>raghavgururajan, I think there's an open issue about this
<cbaines>raghavgururajan, I think the workaround is to use --user
<raghavgururajan>cbaines, Thanks! It appears rekado mentioned "/var/lib/flatpak" as work around.
<raghavgururajan>You mean just --user or --user username?
<cbaines>raghavgururajan, I think you add --user when adding the remote, something like: flatpak --user remote-add
<raghavgururajan>Ah I see.
*raghavgururajan tries
<raghavgururajan>cbaines, That worked. Thanks so much.
*raghavgururajan does not like flatpak, but needed to install Zulip. Requied for Outreachy Communications.
<cbaines>I haven't used much, but I've just used the web interface so far
<pkill9>Guixguy might be having that problem with flatpak
<raghavgururajan>cbaines: Yeah, I thought I could just use web interface, but keeping it open in icecat, hogs resources.
<ActivateYourLinu>Purchase a license now! $89 for single computer if you purchase it before I'm banned!
<Guixguy>raghavgururajan: flatpak was at version 1.4.3 yesterday and 1.6.3 today. When adding the flathub repo, use --user
<Guixguy>raghavgururajan: Also, there is a bug with 1.6.3. I had to roll-back my upgrade
<raghavgururajan>Guixguy, Thanks, --user worked.
<Guixguy>raghavgururajan: I am learning how to use inferiors to use the older version of flatpak
<raghavgururajan>cbaines: Does AppImage work in Guix System?
<cbaines>raghavgururajan, I don't know, I've never tried AppImabe
<rekado>Guixguy: what’s wrong with version 1.6.3? Upstream problem or something we did?
<Guixguy>When I looked up the error, it has happened before in 2018
*raghavgururajan thinking of packaging Zulip
<raghavgururajan>Uh oh, Zulip is an electron application.
*raghavgururajan whispers Amen!
<rekado>yeah, better not them
<raghavgururajan>rekado, me?
<rekado>raghavgururajan: I mean, Eletron apps are essentially unpackageable.
<cbaines>at the moment, maybe we'll be able to tame the NPM mess at some point...
<Guixguy>It's great to move completely to Guix now =] no more Manjaro
<lle-bout>cbaines: GNU Guix is a great platform for attempting to tame the NPM mess but the NPM mess also needs work, many packages download external binaries and so on, they would need to be individually patched, and some times, the upstreams are not even cooperative to changing the way it works and are hostile to the reasons.
<cbaines>indeed, it's definately a challenge
<raghavgururajan>rekado: Yeah, I figured. IIRC, ryanprior was also looking for a way to package electron apps.
<olivuser>hey guys, for some reason "make authenticate" does not work (I just pulled guix), it says "no rule specified, terminated"
<olivuser>whats the deal?
<rekado>olivuser: did you configure first?
<olivuser>rekado, no I didnt, because thats not specified in the manual
<olivuser>so it is 'configure authenticate && make'?
<rekado>./configure --localstatedir=/var && make authenticate
<rekado>the manual puts “make authenticate” too early
<rekado>without configuration you won’t even have a Makefile
<rekado>and thus no “authenticate” target
<rekado>do note that authentication is optional
<olivuser>but it is advised, no?
<rekado>but there have been some usability discussions about this recently, so expect it to be kind of bumpy
<olivuser>alright, then I might want to ignore it for the time being
<olivuser>thanks for the input anyway :)
<pkill9>is anyone else unable to run `guix environment --container --share=/run`? I get: guix environment: error: mount: mount "/run" on "/tmp/guix-directory.3FT6UA//run": Invalid argument
<rekado>sysadmins, we’re at 1.6T free space on I’m trying to organize extra storage that we can connect over iscsi to relocate all 5TB of /var/cache.
<rekado>but this will take some time and I’ll have to go to the data centre to prepare this.
<rekado>so in the meantime I’d be happy if someone could help out with removing garbage
*rekado –> afk
<Guixguy>pkill9: I get the same error
<reepca>hmm... if we need to run ./bootstrap and ./configure before we can run "make authenticate", don't we need to bootstrap trust on those files first?
<cbaines>as with all bootstrapping, you can only go so far. If you remove the need to run ./configure, you're still trusting the make authenticate process
<civodul>reepca: the end goal is to run code already installed to authenticate code you just fetched
<civodul>but yeah, bootstrapping
<shcv>hello; I noticed that there is now a package for ZFS; does anyone have an example config that uses it as the root filesystem?
<ecbrown>shcv: i heard that guix won't work on root-on-zfs
<shcv> hmm
<ecbrown>i am currently working toward just a basic guix system (tm) that will create a tank
<ecbrown>(on another volume)
<cbaines>I think it might work, it would just require some changes to Guix
<cbaines>similar to those required for btrfs support I guess
<ecbrown>after a little bit of use, i think btrfs is probably fine for a root volume
<shcv>I guess a few decent btrfs examples might help for figuring it out
<shcv>I think my problem is that there's just a little too much magic / layers of abstraction involved in the declarative packaging
<efraim>I've been pretty happy with btrfs as root
<bricewge>But guix don't yet support btrfs subvolume
<ecbrown>dammit bricewge i was just about to reformat my ubuntu
<ecbrown>now i will dd it ;-)
<bricewge>ecbrown: Tho there is a pending patch
<ecbrown>soooo happy that tigervnc-server was fixed
*civodul stops cuirass so reconfigure can complete
<ecbrown>how great is herd? it starts things, it stops things, it restarts things...
<ecbrown>i guess that's just about it
<rekado>civodul: thanks
<bricewge>ecbrown: It does way more that this “sudo herd doc root list-actions” ;)
*janneke tries civodul's profile hack with #:system, #:target ... building guile3
<bricewge>ecbrown: Each service can define arbitrary action, for example invalidating cache of host name lookup “herd invalidate nscd”
<civodul>rekado: i've stopped a whole bunch of processes but am still getting "database is busy"
<bricewge>ecbrown: Or setting the proxy for guix-daemon “herd set-http-proxy guix-daemon http://localhost:1337”
<civodul>ah, seems to work now
<civodul>janneke: sorry for not making much progress :-/
<rekado>I’m still waiting for that reconfigure to complete
<rekado>oh, no longer
<rekado>it was terminated
<rekado>civodul: wait, are you *also* reconfiguring?
<rekado>because… that’s what I’ve been doing for the past couple of hours
<janneke>civodul: i expected this not to be easy, but not /this/ difficult -- crap!
<janneke>meanwhile i have been worrying about xattr and hurd; thought it would be easy but i'm kinda stuck (for inspiration) there now too
<civodul>rekado: yes, but it was stopped
<civodul>so i killed it
<civodul>and restarted
<civodul>and killed a few other things
<civodul>like stale "guix offload" processes
<civodul>rekado: interesting: there was at least one child guix-daemon process from a defunct parent guix-daemon process
<civodul>and that thing was holding a lock on a store item
<civodul>"guix processes" doesn't report stale children like this
<civodul>janneke: uh
<civodul>rekado: anyway, reconfigure completed
<civodul>the new mcron jobs will GC more, FWIW
<civodul>now running gc by hand
<civodul>i'm going afk but i'll arrange to have cuirass restarted
<whk>Submitted bug 41217 ( "guix pull: error: You found a bug: the program '/gnu/store/h1b3kqafcgwv5zdj98f9ka3nppkz4l09-compute-guix-derivation'" I will be logged in here for a bit if there are anything I should try to troubleshoot.
<lfam>Thanks whk
<rekado>I’m trying to reproduce this, but it looks good so far
<lfam>I have a feeling you ran out of RAM whk
<whk>Reasonable. I can add more but was just walking through the steps. I just started "free -s 1" and will try again.
<rekado>can’t reproduce this, so lack of memory is a good guess.
<pinoaffe>is there a command-line rar archive utility in guix? I was unable to find one
<bdju>pinoaffe: p7zip can unrar files
<whk>lfam & rekado: Agreed on ram. Failed again at different point but free reported only about 8MB available at the time. I will boost memory and try again.
<lfam>Okay. Please let us know how much you needed so that we have an idea of the current requirements
<whk>Rgr I will go up in increments of 1GB
<cbaines>how much RAM did it just fail with?
<whk>The default 1GB from the manual. This was my first machine.
<cbaines>Ok, I believe 2GB should work
<alextee[m]>pinoaffe: gnome does this in the file manager if you just want to decompress
<lfam>We should update the manual, although it's a shame that 1 GB isn't enough anymore
<whk>lfam, rekado, cbaines: With 2GB allocated to VM the "git pull" (using config in bug 41217 )dropped memory available (per "free -s") to about ~320MB at the lowest. Spent alot of time around ~910MB available. Feel free to close bug with use more memory as fix.
<lfam>whk: Can you close the bug report by sending a message with your findings to <>?
<Guixguy>How do I get info for my (commit ) field in a git-fetch?