IRC channel logs

2021-03-23.log

back to list of logs

<lispmacs[work]>I just checked and I have 22 open bug reports. Do I get some kind of Free Software Award for that?
<leoprikler>jlicht: Aren't licensing discussions more or less bikeshedding, but with lawyers?
<lispmacs[work]>Most Annoying Guix User?
<leoprikler>As long as you're not a tree, you should be fine.
*raghavgururajan never understood the reference bikeshedding, looks it up on web.
<jlicht>leoprikler: fair enough :).
<leoprikler>Not that I'd complain if half the world's lawyers would convince Oracle to choose any other license, as long as it's compatible with the GPL.
<jlicht>but slightly more on topic wrt the zfs patches; I won't be using it myself, and since it touches file systems, I can't just throw it on a box all fearless like other Guix-related changes
<raghavgururajan>lle-bout: So are we gonna do gnome 4.xy or say with the 3.36 plan? If its the former, then we have to update glib and gtk again.
<leoprikler>I'd personally go for 3.36 or maybe 3.38/3.40 since there's already some work put into it
<leoprikler>but that's just my 2c and I'm not the one who was asked
<charles`>Hi everyone. I'm trying to add the CursorTheme configuration to
<charles`> sddm, and I got it so /etc/sddm.conf has "CursorTheme=Adwaita" in
<charles`> it after system reconfigure. However, the cursor actually used is
<charles`> still the default one. Any ideas?
<lle-bout>raghavgururajan: I would prefer if we could get up to present day GNOME so it's easier to keep up incrementally then but whatever's the easiest
<leoprikler>Is Adwaita not the default cursor theme in SDDM?
<boomerchad>terpri_: thanks, I'll submit a bug report right now.
<charles`>leoprikler no, I'm not sure wat it is, but it is lower resolution and has completely different shapes for everything except "normal pointer"
<charles`>lle-bout, If we move up to latest version of GNOME, would older versions still be installable?
<lle-bout>charles`: they would not, unfortunately we can't put resources to maintain different versions of GNOME in parallel now
<jackhill>raghavgururajan: psi problem reported: https://issues.guix.gnu.org/47331
<raghavgururajan>lle-bout: Cool! I am working on patches in #42958 that didn't build. Let me see I can update glib and gtk.
<lle-bout>raghavgururajan: what do you think about it?
<raghavgururajan>jackhill: Thanks!
<lle-bout>raghavgururajan: do you use guix refresh?
<jackhill>you're welcome, thanks again for packaging them
<lle-bout>raghavgururajan: e.g. ./pre-inst-env guix refresh -t gnome -u
<raghavgururajan>jackhill: I was looking briefly at the problem. It might be something to do with QCA package.
<raghavgururajan>lle-bout: I have pulled-in and bootstraped c-u branch
<lle-bout>raghavgururajan: since GNOME 40 is not released yet I would say go to 3.38 directly if it isnt adding bunch of additional work
<lle-bout>maybe by the time we are finished GNOME 40 will be released
<lle-bout>In fact some release candidates are already out so.. I would say GNOME 40 directly..
<raghavgururajan>lle-bout: 3.38 sounds like good target.
<leoprikler>btw. why the jump from 3.38 to 40?
<lle-bout>leoprikler: they changed versioning scheme
<raghavgururajan>Question: Does substitutes for c-u available in ci.guix.gnu.org?
<lle-bout>raghavgururajan: not that much really, just tell me I can generate substitutes for any command you like
<lle-bout>On my substitute server I sent you
<lle-bout>My machine's speedy
<lle-bout>I can also give you access to some machines
<lle-bout>To setup offload
<lle-bout>jlicht: do you still want machine access?
<raghavgururajan>lle-bout: Cool! Since I am re-editing the patches, I would need substitutes for stuff before I apply and test my patches. So with current c-u (b35581bd63d929e83d18f42b067f63efc867353c), can you do `guix build glib gtk+ gtk+2 gobject-introspection cairo pango gtk-doc yelp-xsl yelp json-glib atk gtdk-pixbuf gdk-pixbuf+svg`
<lle-bout>raghavgururajan: doing it
<raghavgururajan>*no separate yelp
<raghavgururajan>Thanks!
<raghavgururajan>lle-bout: By any chance you setup a domain pointing to your machine?
<lle-bout>raghavgururajan: yes I did, I sent you earlier: http://store.guix-xps.schmilblick.org
<raghavgururajan>Oh I missed it, sorry. and thanks
<lle-bout>raghavgururajan: This is what I ran: ./pre-inst-env guix build glib gtk+ gtk+@2 gobject-introspection cairo pango gtk-doc yelp-xsl yelp json-glib atk gdk-pixbuf gdk-pixbuf+svg
<raghavgururajan>What was that package you mentioned for using IPv6?
<lle-bout>raghavgururajan: miredo
<raghavgururajan>Ah. So how to use it with --substitute-urls=http://store.guix-xps.schmilblick.org?
<lle-bout>raghavgururajan: just run miredo as root it will get you an IPv6 address then that will work
<lle-bout>And make sure to authorize my signing key in your GNU Guix System config for the guix-service
<raghavgururajan>So `sudo miredo` then `guix build foo --substitute-urls=http://store.guix-xps.schmilblick.org`??
<raghavgururajan>Yes, I have that in my system config
<lle-bout>raghavgururajan: yes, only once sudo miredo is needed then you'll have IPv6 networking until you change network config
<raghavgururajan>Awesome
<lle-bout>raghavgururajan: what's your SSH public key?
<raghavgururajan>ssh-ed25519 AAAAC3NzaC1lZDI1NTE5AAAAIErzQMgnxFJChAVuQaDOWSUyA1GdLll00HBhIU7BlJ0V rg@secondary
<charles`>lle-bout, since old versions of GNOME are no longer being developed what maintenance would need to be done?
<lle-bout>charles`: keeping them in parallel is the maintenance
<lle-bout>individual versions of glib, gtk, etc. have to be maintained for security fixes etc.. it's very troublesome
<lle-bout>We would have to duplicate every single package..
<lle-bout>raghavgururajan: thanks
<lle-bout>charles`: you can always use time-machine to access them but there will not be any maintenance there
<lle-bout>time-machine or inferiors within manifests
<raghavgururajan>UNRELATED: Is there a way to enable IPv6 globally in system config?
***jx97 is now known as jx96
<charles`>lle-bout, I guess GNOME is too complmicated to have a package function that just takes a version and makes the package object
<lle-bout>raghavgururajan: what do you mean? run miredo automatically? Otherwise your network should provide you IPv6 access directly, miredo is an IPv4 to IPv6 tunneling software with NAT traversal
<lle-bout>charles`: That's also another topic to solve
<raghavgururajan>lle-bout, Ah so I am not able to access IPv6 address because of my ISP? I thought its a setting.
<lle-bout>raghavgururajan: yes, your ISP doesnt provide you with native IPv6 access
<raghavgururajan>In that case, is it possible to run miredo automatically?
<lle-bout>raghavgururajan: well someone would have to write a service for miredo, maybe with simple-service you can write one that just runs a command, I would really like to do that for many cases but I don't know how to
<raghavgururajan>lle-bout: I see.
***j is now known as jess
<lispmacs[work]>raghavgururajan: miredo is not the way to go if you need an ipv6 tunnel
<lispmacs[work]>6in4 tunnel is much easier
<lle-bout>lispmacs[work]: miredo involves running a command as root and be done with it so I don't see how this can get easier
<lispmacs[work]>lle-bout: the main problem is getting connected to Toredo relays, which are provided less and less by ISPs
<lispmacs[work]>where as you can sign up for an HE tunnel in about 2 minutes
<lle-bout>lispmacs[work]: harder than miredo
<vagrantc>i thought you needed to go through the he ipv6 certification to get tunnels
<sneek>vagrantc, you have 1 message!
<sneek>vagrantc, roptat says: I have substitutes for 13953f0e7ec230e701faa679bcf2c1d6a3ab4693 on armhf if you're interested
<lispmacs[work]>vagrantc: you just fill out an online form, but you do need a static IPv4 address
<vagrantc>roptat: i'll try building on another machine first, but might take you up on some shared substitutes :)
<vagrantc>it's been a while since i've played with alternate substitute servers ... was one of the really exciting things when i first started using guix
<lispmacs[work]>raghavgururajan: if miredo works great for you, that is wonderfully, but it did not for me in the past. I wrote this guide once to setting up 6in4 on a librecmc router:
<lispmacs[work]> https://librehacker.com/2018/03/09/ipv6-internet-connected/
<roptat>it built the kernel tarball, now building the kernel itself ^^
<vagrantc>i must say, it's tempting to switch to using linux-libre tarballs for armhf
<raghavgururajan>lispmacs[work], Thanks!
<apteryx>vagrantc: you'll welcome to beat me to it with a patch!
<apteryx>you're*
<apteryx>not just for armhf
<vagrantc>well, i dipped my toes in that conversation before and ... am not excited to get back into the water
<vagrantc>:)
<vagrantc>though, i think it would be interesting to be able to compare the linux-libre tarballs with the guix-generated ones in some systematic way
<apteryx>vagrantc: there's a patch doing just this on the tracker, but my position, which was discussed with maintainers and OK'd, is that it'd be reasonable to use the linux-libre git repo directly, as it was set up for the very reason that caused Guix to move away from the tarballs: volatility.
<vagrantc>yeah, i remember some of the discussion ... the linux-libre didn't just set up their git workflow for projects *like* guix ... but actually for guix specifically
<apteryx>exactly; that's why there was some resentment with us not using it after they went through the effort of setting it up.
*vagrantc sighs
<apteryx>so we could make some people happy and make the build of linux-libre friendlier to lesser machines, while lowering the complexity. Perhaps the verification of linux-libre could be extracted in a channel for folks interested in the experiment; seems like a win to me.
<apteryx>all this to say that if you were afraid of making a patch, don't ;-). I'll be happy to review it.
<apteryx>and if you don't, I'll get to it eventually :-)
*apteryx returns trying to crack the go module build challenge
<vagrantc>yeah, i kind of forget about channels to explore interesting ideas
<vagrantc>probably because i can hardly keep up with the standard channels :)
<Gooberpatrol66>Does anyone know where service-extensions go in config.scm? I keep getting errors.
<lfam>Gooberpatrol66: In general, service-y things go in the services field. If you could give more details I'm sure someone could assistyou
<Gooberpatrol66>Do they go alongside services or underneath them?
<Gooberpatrol66>i put service-extension openssh-service-type under openssh-configuration and got "invalid field specifier"
<Gooberpatrol66>(service-extension openssh-service-type (const (quasiquote (("root" (unquote (local-file "/root/guix-configs/guixrig/ssh/root-gentoodesktop.pub"))))))): invalid field specifier
<Gooberpatrol66>outside of openssh-configuration:
<Gooberpatrol66>(service openssh-service-type (openssh-configuration (permit-root-login #t) (password-authentication? #t)) (service-extension openssh-service-type (const (quasiquote (("root" (unquote (local-file "/root/guix-configs/guixrig/ssh/root-gentoodesktop.pub"))))))) (service-extension openssh-service-type (const (quasiquote (("root" (unquote (local-file "/root/guix-configs/guixrig/ssh/baksnapper-gentoodesktop.pub")))))))
<Gooberpatrol66>(service-extension openssh-service-type (const (quasiquote (("root" (unquote (local-file "/root/guix-configs/guixrig/ssh/baksnapper-gentooserver.pub"))))))) (service-extension openssh-service-type (const (quasiquote (("root" (unquote (local-file "/root/guix-configs/guixrig/ssh/nathan-guixlaptop.pub")))))))): source expression failed to match any pattern
<lfam>It's hard to read here, can you put it on <https://paste.debian.net>?
<Gooberpatrol66> https://paste.debian.net/1190603/
<lfam>Thanks
<lfam>Are you trying to authorize some SSH public keys?
<Gooberpatrol66>yes
<lfam>Did you try using the authorized-keys field of ssh-service-type's openssh-configuration, as described here: https://guix.gnu.org/manual/en/html_node/Networking-Services.html
<Gooberpatrol66>yes, it only recognizes the last key in the list
<lfam>Hm, that's no good
<Gooberpatrol66>i guess it assumes that you would only specify one key per account
<lfam>The manual specifies that it accepts an arbitrary number of keys
<lfam>So, if that doesn't work, it's a bug
<lfam>"This is the list of authorized keys. Each element of the list is a user name followed by one or more file-like objects that represent SSH public keys."
<lfam>Are you working on this over SSH?
<Gooberpatrol66>yes
<lfam>Did you already grant access to a key that you are currently logged in with?
<Gooberpatrol66>i'm using password authentication temporarily
<lfam>I see
<Gooberpatrol66>so am i supposed to only put one "root" followed by a list of files?
<Gooberpatrol66>what's the syntax for that?
<lfam>Do you see the example in the manual section?
<lfam> https://guix.gnu.org/manual/en/html_node/Networking-Services.html
<lfam>There's an example if you search on that page for 'chris'
<lfam>The root user in the example has two keys
<Gooberpatrol66>i see
<lfam>Oops, I just wiped my test server's authorized keys by accident 🤣
<lfam>Good thing I just got my "test" key onto the server and I'm still logged in
<lfam>I'm setting this up myself to make sure it works
<Gooberpatrol66>it's not asking for a password now on this machine
<Gooberpatrol66>of course, it wasn't before because it's the last key in the list
<jackhill>lfam, Gooberpatrol66: https://paste.debian.net/1190604/ works for me
<lfam>Thanks jackhill
<lfam>That's a great example
<jackhill>:)
<lfam>I'm currently waiting for my server to reboot to test it as well
<lfam>I like that your example uses absolute paths. It's very clear
<lfam>It's working for me with a similar example
<lfam>Let us know if you need further assistance Gooberpatrol66
<Gooberpatrol66>i reordered the keys and it's still working
<Gooberpatrol66>i'd rather not test the other machines right now as there's a thunderstorm
<Gooberpatrol66>thanks for the help
<lfam>Does anyone know how Rust handles time zones?
<lfam>I'm seeing Rust failures when updating the time zone database
<lfam> https://data.guix-patches.cbaines.net/compare-by-datetime/package-derivations?base_branch=master&base_datetime=2021-03-23+04%3A16%3A12&target_branch=wip-next-release&target_datetime=2021-03-23+04%3A16%3A12&system=x86_64-linux&target=none&build_change=broken&after_name=&limit_results=&all_results=on
<lfam>I found the chrono-tz package: It seems we have a github action for this but the job was broken.
<lfam>I meant to paste this link: https://github.com/chronotope/chrono-tz
<lfam>It looks like they aren't keeping up, however
<lfam>It seems hard to believe that updating the time zones would break a bunch of Rust packages...
<lfam>It's hard to tell what's going on since the Rust package dependency graph is not available to Guix tools
<jackhill>That also seems like a wierd package "The impls are generated by a build script using the IANA database and parse-zoneinfo" It seems like we'd want to re-run the scripts anyway
<lfam>We'd want for the package to find the time zones at run-time
<lfam>They change often and we don't want binaries to "go bad" just because some time zones were adjusted
<lfam>I'm not sure python-pytz has that problem or not
<lfam>One of the Rust packages fails to build like this: https://paste.debian.net/1190607/
<lfam>I don't really know what that means
<jackhill>I find some of the ways Rust does thing confusing.
<lfam>I don't really understand it any more or less than most languages
<lfam>It does seem kind of immature, in terms of actually using it. There are lots of circular dependencies and things like that
<samplet>lfam: Looks like Efraim’s commit 2d24b80f3c1fe1e510c7fe161ffced4cd12f59db might be relevant.
<lfam>Hm, interesting
<lfam>I didn't realize that cfg-if was the name of a package
<lfam>It does seem quite relevant!
<lfam>I think my development environment was broken :/
<lfam>I thought this problem manifested after updating the time zones, but now I see that is not the case
<lfam>If anything, the problem doesn't exist anymore, and was tied to an older Git commit
<samplet>It looks like you’re in the clear. :)
<samplet>BTW, thanks for your work on the release!
<lfam>Phew :)
<muradm>hi
<muradm>lfam: i have an answer to you regarding sysctl thing
<lfam>Okay
<muradm>what is your TZ?
<lfam>USA east coast
<lfam>So, I think it's UTC -4
<muradm> should be around midnight there, for me it is early morning, just on my desk
<lfam>Yeah, it's almost 2 o'clock in the morning here
<muradm>ok, around lunch time I will write to 47013 and link 47323 to it
<lfam>Cool
<muradm>may be will meet in your morning :)
<lfam>You might ask about it here during the European midday, too
<lfam>That way you'll be more likely to get some other opinions about it, from some people who have more Scheme knowledge and stronger vision for how services should work
<muradm>no issue, for now I have other things to do any way
<lfam>The issue in 47013 was that I lack that vision
<muradm>the problem with hardlink/symlink setting might be solved in different ways, depending on the severity, but in short, that discussion in 47013 raised due to missings provided 47323 in my opinion :)
<muradm>any way, have to do some other things here, will get back to it then
***jx97 is now known as jx96
<rhou[m]>why does cabal give me an error about invalid GHC_PACKAGE_PATH and missing package db when I call `cabal list --installed`
<nly>hi
<sneek>nly, you have 1 message!
<sneek>nly, mwette says: lost track re libfuse, here is the ffi file: https://paste.debian.net/1189035/, you just get nyacc and type "guild compile-ffi ffi/fuse.ffi"
<nly>thanks
<genr8_>what does "mesboot" mean exactly
<leoprikler>GNU Mes for bootstrapping :P
<genr8_>whats a mes
<leoprikler>Mes is a piece of software; see `guix show mes` :)
<genr8_>ah
<genr8_>so this is the key to everything
<civodul>Hello Guix!
<genr8_>Is there a way to confirm this is actually correct?
<genr8_>cb13f4e0a83b2ecbcfd11c66865189add0c03c57 /gnu/store/jn0cm31xx0kpxbc924fjskbxnxjhyznq-gzip-mesboot-1.2.4/bin/gzip
<genr8_>I've finally uncovered it as the source of my "sterror: unknown error:" error
<genr8_>guix hash /gnu/store/jn0cm31xx0kpxbc924fjskbxnxjhyznq-gzip-mesboot-1.2.4/bin/gzip = 04lhgzlf1il57xjny9gzfp28w5az5z843mqbgb2wamqmh4vrsk2w
<nckx>Good morning, Guix.
<sneek>Welcome back nckx, you have 1 message!
<sneek>nckx, screwdupgrub says: that I get same "Could not prepare Boot variable: No space left on device" will try with live usb later
<nckx>Haso.
<nckx>genr8_: That's what the Guix build farm has!
<genr8_>thanks
<nckx>genr8_: I cheated with ssh, but you could check yourself with ‘guix challenge --substitute-urls=https://ci.guix.gnu.org /gnu/store/jn0cm31xx0kpxbc924fjskbxnxjhyznq-gzip-mesboot-1.2.4’.
<genr8_>thanks.
<PotentialUser-86>Hello!
<nckx>Hullo PotentialUser-86.
<PotentialUser-86>Friends, Is the famous Bombsquad game can be placed in Guix packages? (I will really be sad if not!)
<PotentialUser-86> http://bombsquadgame.com/
<PotentialUser-86> https://github.com/efroemling/ballistica
<PotentialUser-86> https://files.ballistica.net/bombsquad/builds
<nckx>“Q: Does this mean BombSquad is open source? A: Yes and no. All code contained in this repo is MIT licensed and free for use anywhere. This includes game scripts, pipeline tools, and most of the binary engine sources. Anything not directly contained in this repository, however, even if automatically downloaded by build scripts, is still proprietary and cannot be redistributed without explicit consent. This includes assets and game libraries/binaries. So in
<nckx>a nutshell: create and share mods to your heart's content, but please don't distribute your own complete copies of the game without permission.”
<nckx>Yes and no, but mainly no. We can package the engine but (note that I'd never heard of this famous game and take my responses accordingly) I don't think people will be able to ‘guix install bombsquad’ and actually play the game you love.
<nckx>This isn't our choice; we're simply not allowed to legally do so.
<leoprikler>"binary sources" mhh :)
<nckx>IKR.
<PotentialUser-86>@nckx: right, thanks!
<nckx>You're welcome, and sorry :)
<davidl>how can I run cuirass from within a container? Im getting an error about not allowed to remount /gnu/store when starting cuirass from inside the container.
<davidl>Can you use --share or --expose options to solve this? --share is supposed to share read+write so not sure thats a good idea..
<leoprikler>I don't think running cuirass in a container makes much sense unless you're completely simulating /gnu/store
<leoprikler>[or don't care about what happens to it in which case you might as well run it outside]
<davidl>leoprikler: the reason is that I want a recent version of guix for my system and an old version of guix specifically for cuirass, but without running a full vM.
<leoprikler>so you more or less want a system container with shared store?
<davidl>leoprikler: yeah I guess so.
<leoprikler>in that case --share=/gnu *might* do what you want, but I'm not sure if some vague stuff keeps you from remounting even after that
<leoprikler>it's worth a try
<davidl>I actually wouldn't mind having a container with it's own store - just didn't know how to. As long as I don't need to virtualize since Im already in a VM.
<davidl>leoprikler: so it errors with the following when running /gnu/store/..-run-container gnu/build/linux-container.scm:327:16: In procedure mount: mount "/gnu" on "/tmp/guix-directory.PA0n41//gnu": Invalid argument
<davidl>it was created with guix system container config.scm --network --share=/gnu=/gnu
<efraim>Try expose instead of share
<efraim>You can't write to /gnu
<davidl>efraim: Ok thx. Also btw, I needed to do --share=gnu and not --share=/gnu or else it errors like above.
<davidl>efraim: same error with --expose.
<rndd>hi everyone! does anybody tried to pack golang cli apps in guix?
<nckx>rndd: (golang == Go, right?) Surely we already have some, like restic.
<nckx>I don't think they're any more rare than other Go packages. Just a pain to package.
<nckx>(...like other Go packages.)
<rndd>nckx: ohhh, thanks. yea, i have some pain to. trying to pack - simple go build doesn't produce binary =(
<rndd>btw, i also wanted to ask - is there any plans to pack racket packages?
<mariofoo>Hi, I was trying to install the GNU Guix system today and I'm having trouble with Xorg, can anyone help?
<rndd>mariofoo: sure, what kind of problem?
<rndd>what hardware you use?
<rndd>does guix boot?
<rndd>need more info
<mariofoo>I'm on a Pentium E5300 HP Compaq dx2420 Microtower, don't know if it's enough info. Guix boots, but gdm gives me a white screen saying that there is a problem
<mariofoo>I tried to drop to a tty, installed xorg-xinit and created an .xinitrc
<rndd>mariofoo: what gpu?
<mariofoo>Intel 82G33/G31
<mariofoo>Always worked fine with i915 module
<mariofoo>And lspci -kk says it is loaded
<rndd>mariofoo: you can try another dm
<mariofoo>VGA compatible controller: Intel Corporation 82G33/G31 Express Integrated Graphics Controller (rev 10)
<rndd>for example slim
<rndd>and check if it works
<mariofoo>Ok will do, it's the same machine I'm on atm, just a different HDD. brb
<rndd>mariofoo: ok, also, give error messages you get
<rndd>if you do
<mariofoo>Yes i'll try to copy them, if they're too long I'll put them on pastebin if it's ok
<rndd>mariofoo: use debian paste, please
<rndd>
<mariofoo>Ok
<mariofoo>brb
<lle-bout>hello! :-D
<civodul>o/
<nckx>Hullo lle-bout.
<lle-bout>civodul: do we do anything for mongodb? should we issue a news entry?
<lle-bout>it seems people don't agree but have nothing against the remove seeing the situation?
<lle-bout>raghavgururajan: everything has built since a little while now and is available through the substitute server at http://store.guix-xps.schmilblick.org
*lle-bout wonders how they could at the same time advertise access with avahi to their guix store and enable HTTPS with nginx and certbot
<roptat>they're two separate things, right?
<lle-bout>roptat: advertise? parameter is within guix-publish
<lle-bout>but if I use nginx I wont be using the same port as guix-publish so advertise wouldnt work
<roptat>you'd have guix publish publish on http internally and on avahi, then nginx would use the http and proxy that
<roptat>oh
<roptat>right, sorry
<lle-bout>roptat: probably that's easily solved by adding a manual avahi service advertise but I don't feel at ease with that yet
<civodul>lle-bout: hi! so far there seems to be a rough consensus on removing it
<civodul>when removing it, we should probably have added a news entry, yes
<civodul>we can still add one, but those who've already updated won't see it
<bone-baboon>I am working in Emacs on a virtual terminal using `emacsclient --tty`. I have added `%desktop-services` to my configuration file. Services are outputting to whatever virtual terminal I am using over top of Emacs. I can remove the output of these services using the Emacs command `redraw-display` but I am having to do that very often. How can I configure these services not to output to my current virtual terminal?
<pranavats>Hello, is there a way to skip the check phase for some package (or all packages) while executing `guix system reconfigure`?
<pranavats>One of the packages (ffmpeg) takes more space that I can provide it during check phase, and I'd like to disable the check.
<yoctocell>pranavats: You can use package transformations, see "8.3 Defining Package Variants" in the manual.
<pranavats>Okay
<abcdw>hi guix!
<lle-bout>abcdw: hello! :-D
<pranavats>If I pass --keep-going option to guix system reconfigure, would the derivations that fail to build still persist or be available in /gnu/store/?
<pranavats>ffmpeg derivation is failing for me at check phase due to lack of space during a system reconfigure, but the build step finishes. If I choose to pass --keep-going, would the incomplete build of the derivation still be available?
<yoctocell>pranavats: You can pass --keep-failed and then you can find the source code for ffmpeg somewhere in /tmp.
<pranavats>Thanks yoctocell
<yoctocell>np
<smartineng>Hello, what "guix environment --ad-hoc git:svn" does exactly? And what to do to use "git svn" ?
<roptat>guix environment --ad-hoc creates a profile with the given packages and puts you in a new shell where that profile is activated
<roptat>git:svn means the svn output of the git package, so "git svn" should be available
<roptat>oh, you might need to add git too
<roptat>otherwise the "git" command would not be available
<smartineng>roptat: ok thx it's a bit confusing but it works
<roptat>(and if you already have git outside, then I think it's still needed in the environment so guix knows it needs to set some variables for it to find the svn module)
<roptat>git:svn only contains the module, not git itself
<smartineng>roptat: actually it make sens you can add more modules that way
<roptat>right, you can add git:send-email, git:svn, etc
<smartineng>awesome
<roptat>they're all listed by "guix show git"
<roptat>in the outputs section, where out actually means the package itself, with no colon (so "git" instead of "git:out")
<roptat>success, my armhf server is finally reconfigured
<roptat>lle-bout, I saw your patch for java-mxparser, I'll have a look now
<roptat>first, you didn't get the right sources, you can't use the -sources.jar, because those are not sources, despite the name
<roptat>they're generated from the sources, and used only for debugging
<lle-bout>roptat: okay I see, I saw this in other packages that's why
<lle-bout>I tried the git repo earlier but didnt work either, the upstream only has pom.xml and we don't have a maven-build-system do I didnt know
<lle-bout>thank you for picking it up
<roptat>lle-bout, we do have one
<lle-bout>roptat: oh well!
<roptat>but that's not usable for this package, because it's in the path to maven itself :/
<lle-bout>oh..
<roptat>the best you can do is use the ant-build-system and include dependencies that are in the pom.xml
<roptat>lle-bout, ouch, this depends on xmlpull1, and the website where we download xmlpull2 is "deprecated" and redirects to the internet archive
<lle-bout>roptat: ahh..
<roptat>unfortunately they don't seem to have the page I'm interested in :/
<lle-bout>roptat: so the code has disappeared? :-S
<roptat>lle-bout, seems so
<roptat>I'll ask them on github if they have the source of their only dependency
<lle-bout>roptat: maven didnt archive it?
<roptat>nope, there's nothing there: https://repo1.maven.org/maven2/xmlpull/xmlpull/1.1.2.1/
<roptat>well, there's the binary and a very minimal pom file
<kcurtet>Hi, I'm installing a third party package and after the build succed the package outputs an error becouse they can't run /usr/bin/wget. What can i do? My only solution was adding '/usr/bin/wget' special path but its not a good one
<sneek>Welcome back kcurtet, you have 1 message!
<sneek>kcurtet, nckx says: ‘-c’ specifies the number of different processes/threads that builders are advised to use. You usually want to set it to the number of cores your CPU has, which is what the ‘nproc’ command returns.
<roptat>got the wrong version, but same: https://repo1.maven.org/maven2/xmlpull/xmlpull/1.1.3.1/
<lle-bout>roptat: sad sad
<nckx>kcurtet: Fix the package not to hard-code wget's location but use $PATH instead.
<roptat>lle-bout, https://github.com/x-stream/mxparser/issues/3
<nckx>(Then when it's added to Guix we can sigh because that makes it harder to patch in a hard-coded wget.)
<roptat>lle-bout, https://paste.debian.net/1190665/ here's what the package should look like when we find xmlpull
<roptat>basically, the #:jar-name ensure the ant-build-system generates its own ant.xml file
<roptat>might be missing a copy-resources phase too
<lle-bout>roptat: I see, thank you!
***kiroul3 is now known as kiroul
<civodul>oh oh! the Matrix bridge is going through difficulties
<rekado>hmm, looks like “guix pull” no longer installs /lib/systemd/system/gnu-store.mount and the like
<rekado>that’s what the installer script depends on.
<rekado>so when a user upgrades the root user’s Guix and removes the old profile generation they no longer have the systemd unit files.
<GuixNewbie321>Hi!
<htsr>Hi! guix reconfigure --no-substitutes fail on lsof test phase, how can I fix this?
<nckx>htsr: Mount a tmpfs on /tmp and retry.
<nckx>hstr: Please give more details when reporting bugs. Now I'm forced to assume you're talking about the btrfs-incompatible LTlock test, which might be completely wrong. :)
<g_bor[m]>Hello, civodul, rekado, nckx
<nckx>Hiyo.
<g_bor[m]>I have just read the outreachy mailing about fsf sponsorship
<nckx>Oh, I haven't, if I'm even a recipient.
<g_bor[m]>I will forward it.
<nckx>Hi GuixNewbie321.
<g_bor[m]>Last one I saw was that they were asking us if we are willing to accept the offer from Software Freedom Conservancy to sponsor one intern.
<Whyvn>what is the difference or reasons for having/installing emacs packages in the guix repo compared to elpa or melpa?
<htsr>Sorry nckx, I have a btrfs partition and yes, the LTlock test fail for lsof-4.94.0
<nckx>htsr: Not much we can do until upstream fixes it (or someone who knows why it fails submits a patch that we can ship) <https://github.com/lsof-org/lsof/issues/152>
<nckx>Building it on anything but btrfs works, so building it in a tmpfs TMPDIR is a work-around.
<g_bor[m]>nckx: I've just sent it to you. I've also sent an answer that I will contact them as soon as we come up with a soultion.
<htsr>nckx: Thanks
<nckx>Whyvn: That they are managed by Guix, with all the concordant advantages (and possibly disadvantages). That's it.
<nckx>You can still use other methods.
<GuixNewbie321>I'm trying to install threadscope haskell package, by importing it but I run in some errors. I think it's chocking with one of the packages
<GuixNewbie321>so I'm following "building from git" from the info manual to see if I can debug it; does somebody have some experience doing this?
<lfam>Many of us have experience building Guix from Git and then using the result for working on Guix packages
<GuixNewbie321>great!
<GuixNewbie321>btw, when importing *threadscope* doesn't fail, but it does when I try to do it recursively... I pinpointed the package:
<g_bor[m]>the more I use tools that have limited templating capabilities, the more I enjoy the freedom of doing config management with guix :)
<lle-bout>yay cuirass release!
<lle-bout> https://guix.gnu.org/en/cuirass/
<lle-bout>oo we even have live logging now!!
<lfam>Wow, 1.0.0!
<nckx>Sweet!
*nckx .oO that the site header & logo background colours are different shades of grey bothers me more than it should
<lle-bout>and there is a full documentation wowow!!
<lle-bout> https://guix.gnu.org/cuirass/manual/cuirass.html
<roptat>lle-bout, https://github.com/aslom/xmlpull-api-v1 seems to implement that missing package
<roptat>but not sure if it's the same or if there are differences
<roptat>lle-bout, you'll have to be careful, you need to remove the lib directory in a snippet
<GuixNewbie321> https://pastebin.com/weSYUZeK
<GuixNewbie321>these are the problems
<lle-bout>roptat: if you can do it please do :-) - Thanks for investigating
<roptat>not sure, they lost all history when importing, so it's not clear which version it is or if it is compatible
<roptat>I'd prefer waiting for an answer from xstream devs, to be sure
<lle-bout>roptat: okay!!
<GuixNewbie321>I can, but as my nickname says, I'm newbie, so I'm not shure how to. I begin with building guix from source, then I'll try to debug the import mechanism... does anybody that did this has some tip?
<lfam>GuixNewbiew321: So, basically, the problem is that a program is named ghc-events upstream, but you can't import it as either 'ghc-events' or 'events'?
<lfam>GuixNewbie321 ^
<GuixNewbie321>ifam: yes, with ghc-events, it fails with the same input error that I see when importing recursively threadscope package from hackage
<lfam>We both spelled each other's names wrong :)
<lfam>I'm not sure what's wrong with the importer in this case, but I recommend just writing the package definition for ghc-events by hand
<lfam>You can try adapting from of the other package definitions that the importer gave you
<GuixNewbie321>plz be patient I'm really bad at English, learn it by video games university XD
<lfam>No worries :)
<lfam>You can copy the package of ghc-threadscope, but change the name, version, uri, sha256, home-page, and erase the inputs
<lfam>Of course, you'll also change the synopsis and description, and check the license
<lfam>GuixNewbie321: What's your native language? Maybe somebody here can help you by speaking your language
<GuixNewbie321>mmm that might be the least effort solution, I'll give it a try... anyway I'm interested in learning guix internals
<GuixNewbie321>lfam Spanish, but it's ok, if you don't mind reading a couple of poor redacted questions, I don't mind write them
<lfam>Okay, I'm happy to use English with you
<lfam>nckx: I pushed a followup to your Borg update. The origin snippet usually needs adjustment
<lfam>GuixNewbie321: It's good to want to learn Guix internals :) For me, the easiest thing is to write the package by hand, but it would be great to understand why the importer doesn't work for this package
<GuixNewbie321>well... recursive failure get in the way.
<GuixNewbie321>make check, in guix sources failed with errors ...
<lfam>Okay, you can probably ignore that for now. Does `make` succeed?
<GuixNewbie321>yes
<lfam>Great
<lfam>Now, you can use the Guix you built by doing, for example, `~/tmp/src/guix/pre-inst-env guix build hello`
<lfam>That will use the package definitions in the Git repo. It's how we test our changes
<lfam>The 'pre-inst-env' script is the key
<lfam>So, if you added new packages, you can use it to build them
<lfam>All of the Guix commands can be used in this way
<lle-bout>is there no way to figure out where an error is in a Scheme file using GNU Guile..?
<civodul>GuixNewbie321: there are a couple of limitations with "guix import hackage", such as https://issues.guix.gnu.org/39844
<GuixNewbie321>guix build: error: failed to connect to `/usr/local/var/guix/daemon-socket/socket': No such file or directory
<GuixNewbie321>mmm I'm running inside a container with: guix environment guix --pure --ad-hoc help2man git strace --container
*civodul looks a backlog
<civodul>Cuirass release, woohoo!
<civodul>🎉
<lle-bout>civodul: :-D
<lle-bout>I hope to run mine
<lle-bout>soon
<roptat>lle-bout, you should be able to find that in the backtrace, but sometimes it's hard
<lle-bout>With all the docs it will be so much easier and predictable
<civodul>yup
<g_bor[m]>civodul: have you seen the last mail from Sage?
<civodul>g_bor[m]: hi! maybe not, i just saw the one i replied to this morning
<civodul>anything urgent?
<lle-bout>roptat: ice-9/eval.scm:293:34: In procedure list-head: Wrong type argument in position 1 (expecting pair): () -- that's all I have, after bunch of ./pre-inst-env guix refresh -t gnome -u changes
<lfam>GuixNewbie321: You'll need to exit the container
<lfam>You don't use to be in a special environment to use 'pre-inst-env' to build packages
<roptat>lle-bout, sure you don't have a few lines above that? if not, then ok, it's not very useful ^^
<lle-bout>roptat: nope nothing useful above
*nckx forgot: g_bor[m]: thanks for forwarding it.
<lle-bout>I think it's due to version and version-major+minor
<lle-bout>guix refresh considers 2.68.rc is newer than 2.68.0
<g_bor[m]>civodul: They offered to sponsor one intern. Do we accept it?
<civodul>oh!
<zimoun>civodul: which mail from Sage? Or on which list?
<civodul>zimoun: a private message
<GuixNewbie321>lfam: ok, I needed it because with out that I can't use --pure as the info manual says. I'll check with out --conteiner and --pure
<zimoun>civodul: ah ok. Thanks.
<g_bor[m]>I don't see any drawbacks, as we only have one project anyways.
<civodul>g_bor[m]: i answered you by PM :-)
<raghavgururajan>Hello Guix!
<abr>Hello !
<civodul>g_bor[m]: but yeah, it sounds like an appealing offer
<raghavgururajan>lle-bout: Thanks!
<lle-bout>raghavgururajan: I'm looking at upgrading to GNOME 40 on my end, but just an experiment
<raghavgururajan>Cool!
<GuixNewbie321>with in an environment without --pure nor --container I get the same error. In a normal shell, I get a backtrace
<lle-bout>roptat: respond to the issue by email please, on java-xstream
<abr>I can't find anything about using NetworkManager's applet, I seem to lack the right to create connections (but I can edit them for some reason ?!) any useful pointer on that ? I've been searching the doc for hours and I can't find anything past the herd service
<roptat>lle-bout, ok
<lle-bout>roptat: 47342@debbugs.gnu.org
<lle-bout>roptat: because lfam just said they have pinged Julien so they need to be aware of your work
<GuixNewbie321>I'm running guix on top of a distro
<nckx>g_bor[m]: After saying just once that this makes me very uncomfortable & I'm disappointed in Outreachy, I won't object further. That wouldn't be right either, after all the effort you've put in. Cancelling now wouldn't be in Guix's best interest anyway.
<roptat>lle-bout, I'm Julien ;)
<lle-bout>roptat: ohhh!!!
<raghavgururajan>lle-bout: For some reason, the I dont get ssubatitute for librsvg (/gnu/store/vq8swv5cb71bf8plw1w3sgf5s4kbsqgh-librsvg-2.50.2.drv)
<raghavgururajan>lle-bout: I tried building though ssh, it is already built there. So the .drv is different?
<lle-bout>raghavgururajan: the substitute server is on my machine
<lle-bout>not the server
<lle-bout>but I have lots of the store items from the server
<lle-bout>Let me see
<raghavgururajan>lle-bout: Ah I see.
<lle-bout>raghavgururajan: I got /gnu/store/x3ygcj630b7bahyl6a0zi59jzry5hbcy-librsvg-2.50.3.drv
<lle-bout>raghavgururajan: the server is not running GNU Guix System so as much as I would like to setup a guix-publish thing here, it's more complicated
<raghavgururajan>AH.
<raghavgururajan>lle-bout: What's the output of your `./pre-inst-env guix describe`?
<lle-bout>raghavgururajan: https://bpa.st/raw/AGXQ
<lle-bout>raghavgururajan: I have a local commit, but more like a9d540cfa87ef3a5de3296188f650fb0d037efbd
<lle-bout>Latest available in core-updates branch
<abr>so no-one creating connections in nm-connection-editor ?
<raghavgururajan>lle-bout: Ah, I think that might be it. Mine is ff594dc1c83cdb6dd5661b03fbefe890d82b3e44. Could you git pull and retry the build?
<raghavgururajan>The current c-u is at ff594dc1c83cdb6dd5661b03fbefe890d82b3e44
<lle-bout>raghavgururajan: current c-u is a9d540cfa87ef3a5de3296188f650fb0d037efbd
<lle-bout>I pushed to it recently
<lle-bout>I am at current c-u
<lle-bout>+ one local commit (irrelevant to you)
<raghavgururajan>Ah okay, I;ll git pull now
<raghavgururajan>lle-bout: What's the output for `guix environment guix --pure --ad-hoc guix` then `./pre-inst-env guix build librsvg --dry-run`?
<lle-bout>raghavgururajan: in the checkout or outside?
<raghavgururajan>in the checkout
<lle-bout>raghavgururajan: I don't understand how this can impact anything
<lle-bout>raghavgururajan: since you also use ./pre-inst-env from within the environment
<raghavgururajan>Since the system is not guix system, I system, I suspect its picking something in unpure env
<raghavgururajan>You already run ./pre-inxt-env commands inside an env right or no?
<lfam>I don't
<raghavgururajan>lle-bout, Never mind, guix build is already restricted.
<lle-bout>raghavgururajan: yes I do run them inside an env but I don't think we are totally on the same page
<raghavgururajan>lle-bout: Never mind! :D
<raghavgururajan>lfam, on foreign distros?
<lle-bout>raghavgururajan: Just to clarify, I have my laptop with my powerful machine set up as offload, I work on my laptop not the offload machine, so my laptop runs GNU Guix System, the machine does not but as you said it runs builds in isolation
<lfam>Yes
<raghavgururajan>lfam, Intresting, i have seen missing guile dependency errors.
<lfam>The pre-inst-env script should make it so you don't have be in `guix environment guix`. If it doesn't, it's likely that some environment variables are being erroneously exported in ~/.bashrc or similar
<lfam>It's missing Guile?
<raghavgururajan>Oh you must have installed installed on the system.
<lfam>Yeah, I do
<raghavgururajan>Makes sense then.
<raghavgururajan>lle-bout: Cool!
<abr>how can one get a "%sudoers-specifications : unbound variable" error when explicitly having «(use-modules (gnu system))» in the file ?
<abr>s/specifications/specification
<abr>it's written without the 's' in the file, I've doubled checked
<genr8_>how can I obtain the official buildfarms build log for /gnu/store/jn0cm31xx0kpxbc924fjskbxnxjhyznq-gzip-mesboot-1.2.4 <-- i want to compare the working one to my broken one
<lfam>genr8_: What command did you run to start the build that failed?
<lfam>In general, `guix build` takes a --log-file argument
<genr8_>i have my own log. Look at lines 132-138. the build succeeds but the binary is rendered broken https://dpaste.com/B7RMQ5XRV
<lfam>You can also append --derivations to the command that caused the failing build. That will print the derivation of the build, which I could use to look up the build log on the server
<genr8_>this ? /gnu/store/w37macy41dk7828nr768d3ax4hbk9h9i-gzip-mesboot-1.2.4.drv
<lfam>Thanks
<lfam>It's here genr8_: https://ci.guix.gnu.org/log/jn0cm31xx0kpxbc924fjskbxnxjhyznq-gzip-mesboot-1.2.4
<lfam>It's a gzip-compressed text file
<genr8_>hmm yours has the exact same thing
<genr8_>im going to run "challenge" on every item in the store
<genr8_>1 (100.0%) were inconclusive
<genr8_>what does that mean. how can something be inconclusive. are we using SHA1 hashes that miniscule bitflips can slip through the cracks
<genr8_>hmm theres more of them
<lfam>Can you share the command that you used?
<lfam>We use SHA256 but, even if we used SHA1, your interpretation would be incorrect.
<lfam>SHA1 is still strong against random bit flips
<genr8_>i picked a suspicious item in the store and ran challenge with the --substitute-url. im gonna go run some more commands.
<lfam>I think that, if you are worried about bitflips or filesystem corruption, you should try `guix gc --verify=contents`: https://guix.gnu.org/manual/en/html_node/Invoking-guix-gc.html
<avalenn>I think that inconclusive is if either there is no subsitute or no local build
<genr8_>ok
<lle-bout>roptat: hey, well still stuck on that mysterious syntax error..
<genr8_>ooooh boy. yep it found stuff. lots of it
<lfam>Dang :/
<lfam>At least we have something to go by now
<genr8_>omfg, i was investigating the wrong gzip...
<genr8_> https://dpaste.com/79LW8E5VQ
<lle-bout>maybe now change disks or guix gc --verify=contents,repair
<lfam>Oof
<lfam>Before changing disks, I'd test the RAM
<lfam>I've actually never experience a disk with random errors that weren't caused by faulty RAM
<lle-bout>sure, RAM also
<lfam>And, I've had about 10 computers in my life, and one of them had faulty RAM, provem with memtest86. So, 10%
<lfam>It's more common than we'd like :(
<genr8_>ok. it fixed everything, except my -profile.
<lfam>You might be able to gc that profile
<genr8_>i could do that, how
<lfam>If it is "alive" and can't be gc-ed, you can look in /var/guix/gcroots to see how it is alive and take the appropriate steps, with `guix package --delete-generations NNN`
<roptat>lle-bout, I'll be busy for at least the next hour, but if you share what you're doing someone might be able to have a look
*raghavgururajan is having Smoke's Poutine
<genr8_>yes its alive
<lfam>So, check the list of generations in `guix package --list-generations` or `guix package -p ~/.config/guix/current --list-generations`, switch generations if necessary, delete the generation, then you can gc
<lfam>You'll have to match the generation numbers with the gcroots in /var/guix/gcroots to make sure you've got the right one
<genr8_>I wonder if this is the actual problem. guix package: error: profile '/var/guix/profiles/per-user/root/guix-profile' does not exist
<genr8_>cant delete. cant even list.
<genr8_>oh wait. guix package -p ~/.config/guix/current --list-generations <-- worked.
<genr8_>i think my env variable got lost or something ?
<lfam>I think the actual problem is very likely to be data corruption
<lfam>Caused by disk, RAM, or even miscompiled software :/
<genr8_>i will take that into account
<lfam>Actually, can you copy that "bad" profile out of the store, to preserve it?
<lfam>I'm curious what it looks like
<genr8_>ok
<lfam>It could be useful to understand the nature of the corruption
<lle-bout>roptat: I shared on #guile
<genr8_>I'm at a loss of how to delete this and proceed. https://dpaste.com/8AEY352YY
<lfam>Something is broken, if that directory does not exist
<genr8_>it does exist
<genr8_>just rm -r it ?
<lfam>And, I don't understand the first line of your paste
<lfam>That's not something you should need to od
<lfam>need to do
<lfam>But, your paste doesn't show which directory you're trying to remove, or that you are trying to remove any directories
<lfam>I assume you want to remove the /gnu/store/66dqq6nr8g6sn9drizpwjzyblrb0iyis-profile directory?
<genr8_> i was just trying to make sure, i dont get why it needs it explicitly specified as a -p variable to --list. all my attempts with --delete-generations have failed in syntax error
<genr8_>yes
<lfam>Okay
<lfam>The reason you needed to use -p there is that you were trying to inspect a profile that is not the default profile for the `guix package` command
<lle-bout>rekado: hey, etc/committer.scm doesnt work with gexps, if you don't know
<lfam>That, is the default is ~/.guix-profile
<lfam>But, you are dealing with a different profile, at ~/.config/guix/current
<genr8_>this is what I don't get. I only had the "current" one
<lfam>This second profile is the profile of `guix pull`
<genr8_>ah
<genr8_>also part of what i didnt get
<lfam>So, you can create another generation of it by running `guix pull`
<lfam>Then, you'll be able to delete previous generation, and then finally gc it
<genr8_>thats gonna be an issue. guix pull was the main command that failed, first time around, that led me down this trail of gzip-mesboot, gzip, blah blah
<lfam>Yeah, it's a tough situation
<genr8_>I believe its because I'm not pulling any substitutes, and its happening on the wrong distro for whatever reason. im gonna try a new VM
<lfam>I really recommend you test your RAM
<lfam>If it is faulty, you're going to have more problems
<genr8_>i really dont understand how this could be the thing that suddenly faulted. ive been compiling kernels on this system and a bunch of VMs
<raghavgururajan>lle-bout: I'm such a dumbo. I have been working in master checkout instead of c-u. After switching I get all substitutes from your machine, including librsvg.
<lle-bout>raghavgururajan: happens.. are you tired?
<genr8_>i will be back later i guess. at any rate i'm learning the innards of guix, which is teaching me more than having a functional system
<lfam>Okay
<raghavgururajan>lle-bout: Not at the moment.
<rekado>lle-bout: it only works with changes to packages.
<roptat>lle-bout, I got an answer, and I did find the right fork
<roptat>if you give me ~1-2 hours, I should be able to push the update :)
<atuin>hi
<roptat>hi atuin :)
<atuin>hi roptat :)
<luis-felipe>Hi, has anyone experienced problems with pulsaudio lately?
<luis-felipe>I upgraded guix and guix system to 1ab03fb (5.11.7-gnu) recently and sound settings got messed up.
<luis-felipe>For example, couldn't record in Audacity, no sound in Jami, and pulseaudio using a lot of RAM...
<luis-felipe> https://luis-felipe.gitlab.io/media/2021/03/pulseaudio-abnormal-use-of-ram-2021-03-20.png
<roptat>oh, I had a similar problem where I couldn't record from audacity on a foreign distro
<roptat>but that problem also happened with audacity from the foreign distro, so probably not related
<luis-felipe>After a reboot, pulsaudio stopped eating all RAM, but I had to reconfigure applications manually to previous settings to be able to use the microphone.
<luis-felipe>Audacity works alright now, but I can't use the mic on Jami any more, which was working alright before the upgrade.
<luis-felipe>Back to isolation :)
<luis-felipe>Anyways, I just wanted to see if someone else had similar problems, because it seems hard to find what went wrong
<luis-felipe>Jit.si couldn't detect the microphone either.
<mbakke>oh my, building librsvg on core-updates is almost as slow as the entire Rust bootstrap chain
<civodul>mbakke: that's what it does, no?
<lle-bout>roptat: Thanks a lot!!
<lle-bout>zimoun: mariadb 10.5.8 compiled and tested fine for me with zstd 1.4.9 on core-updates
<roptat>lle-bout, xstream is still failing to build, it's missing a new dependency
<lle-bout>roptat: ow
<roptat>oh no, it looks like mxparser miscompiled
<roptat>missing org.xmlpull.mxp1.MXParser ^^'
<lle-bout>rekado: hey well it was changes to packages
<lle-bout>rekado: it's just in gnome.scm there's gexps and it seems somehow etc/committer.scm can't process anything in gnome.scm at all
<roptat>mh... the class is present, but under a different name
<roptat>uh, it's the same as what's on maven central, so I don't get it...
<roptat>mh.. probably a good reason why I should reduce the path to Maven and use the maven-build-system more
<roptat>there's some magic happening that I don't understand
<roptat>oh, it's from xpp3
<roptat>alright, the pom file includes it too, and it builds with it so...
<roptat>I'll check the dependents and push :)
<lle-bout>roptat: thanks!!
<roptat>lle-bout, do you mind if I become author for the mxparser (I'll keep you author for the xstream update), and put you as co-author? I can do the reverse if you prefer
<lle-bout>roptat: I don't mind anything as long as this issue is fixed
<lle-bout>I don't believe packages are worth much copyright
<roptat>:)
<roptat>they're quite some work sometimes
<lle-bout>they are, but I feel like copyright is the wrong metric to define it, since most of the work is testing
<roptat>no failure so far :)
<zimoun>«./pre-inst-env guix pull -p /tmp/new» says “Migrating profile generations” with “guix pull: error: symlink: Le fichier existe”. I should have done something wrong but I do not remember what and I am confused on what could be wrong. Or why even this “pull” checks “current”?
<civodul>zimoun: check the code that does that in (guix scripts pull)
<PotentialUser-48>I just installed a the guix sd with gnome & tor services but gnome doesn't start up. is this normal?
<civodul>(and consider it read-only :-))
<civodul>PotentialUser-48: hi! no, looks like something is wrong
<civodul>PotentialUser-48: do you see more info on tty12, by pressing ctrl-alt-f12?
<rekado>lle-bout: oh, okay. Could you file a bug report with details please?
<lle-bout>rekado: will do, also need to figure out minimal repro
<zimoun>civodul: for the why, reading the code I see. :-) But I miss what I have wrong. Hum!
<PotentialUser-48>civodul: Yeah there is some info. What stands out is : localhost gnome-session-binary[367]: WARNING: Falling back to non-systemd startup procedure due to error: GDBus.Error:org.freedesktop.DBus.Error.ServiceUnknown: The name org.freedesktop.systemd1 was not provided by any service files
<zimoun>civodul: «export SUDO_USER=1 && guix pull» to bypass the test and restore my brokenstuff. Arf! Well, time to go home. :-)
<zimoun>lle-bout: thanks for checking. Maybe I have seen an artefact. Well, I will to try to reproduce.
<rekado>lle-bout: it’s probably because etc/committer.scm uses “read” and the gexp syntax requires a reader extension
<rekado>I just noticed that you can do “guix environment --ad-hoc” and just hit return.
<rekado>I don’t know why it surprised me, but … I wonder if this can be useful for something.
<civodul>rekado: "guix environment -C" (alone) is almost useful sometimes
<rekado>yes, that one seems obvious
<roptat>lle-bout, pushed!
<PotentialUser-52>hello everyone, I'm trying to setup sway + wayland (which I understand requires sddm?) and am having little luck finding any information on that. even though I see the screenshot on guix.gnu.org indicating that its possible. i would truly appreciate any tips on how to do this...
<PotentialUser-52>i added sway/swayidle/swaylock to system packages, and added the sddm service type, but machine still boots into gdm... for one
<PotentialUser-52>sway-desktop-service type would be ideal, but I don't think that exists
<mbakke>PotentialUser-52: you'll need to remove gdm from the system services in order to use SDDM
<mbakke>there is an example in the manual, search for (eq? (service-kind service) gdm-service-type)
<mbakke>PotentialUser-52: I can give you an example config that starts sway directly without going through a display server if you want
<mbakke>warning: may contain nuts...
<PotentialUser-52>thank you! yes that would be much appreciated, i can live without a display server
<genr8_>is there a way to replace the upstream source package downloads to from https://ftp.gnu.org/gnu/guix/ to a new GNU mirror ?
<lle-bout>roptat: thanks!!
<lle-bout>rekado: yep that's probably it!
<lle-bout>rekado: It would've saved me 117 manual commits to do in a row if it worked aaa
<genr8_>the file /gnu/store/6fc4hzrcspcrdq4rdvmggf6867w5ychx-mirrors contains the list of mirrors . there must be a way to switch
<mbakke>PotentialUser-52: it is probably way overkill for your purposes, but here you go: https://gitlab.com/mbakke/guix-sway-example :)
<PotentialUser-52>mbakke: not at all, thank you so much for sharing this, real configurations help a lot!
<mbakke>PotentialUser-52: you're welcome! I recommend trying it with 'guix system vm' before reconfiguring ;-)