<lle-bout>phossil: support for f2fs may not have been added to the installer, watch out for the 1.2.1 release which will happen in the coming weeks where we will actually test and fix any problem in the installer
<lle-bout>butlerian: what are you trying to do? please suggest improvements to email@example.com or at least document your uncovered use-case so people who know GNU Guile and GNU Guix internals will be able to study the issue (not me)
<boomerchad>So I was able to get a slightly older version of DeaDBeeF to successully compile and launch. The problem is it fails checks after building. It complains about files wich contain translations, which are not in use. Is there a way I can disable the checks or should upstream be patched?
<boomerchad>Versions starting with 1.8.5 add libdispatch as a dependency and will not compile with gcc so I figured I'd start below that.
<phossil>isnt libdispatch an apple project ? their preferred compiler is clang
<boomerchad>Yes, from version 1.8.5 and upwards libdispatch is a dependency for the core and a couple of the plugins meaning clang is needed for those parts. I decided to focus on 1.8.4 for now because libdispatch hasn't been packaged in guix yet (or nix as far as I can tell).
<wehlutyk>I would like to install guix as non-root, on a system on where user namespaces are activated. Do you have any pointers for doing that?
<marusich>Emacs just ate my bug report. I crafted a nice, detailed reply, and then it crashed. Sooo frustrating...
<marusich>Anyone know how to tell Gnus or Debbugs to autosave my articles???
<marusich>This is like the third time it's happened to me, but I can't figure out a good solution. Even pressing C-x C-s to "save" the draft doesn't work; I don't know why but Gnus does the opposite of saving a draft when I do that. It doesn't save a draft.
<marusich>And sadly, now I can't even open the bug report I was viewing in Debbugs becuase of some silly Emacs error...what did I do to you, Emacs?
***regtur_ is now known as regtur
***MidAutumnHotaru1 is now known as MidAutumnHotaru
<soouul>hello, i have a question, can i run the mesa vulkan 32 bit driver on guix? if so how? i cant figure it out
<bdju>anyone else have issues with some qt programs opening links? in both quaternion and nheko I can't seem to just click a link to open it. all my xdg stuff is set fine. links in dino work, emacs can open links, xdg-open command opens links fine...
<masiku[m]>bdju: Did you try turning it off and on again?
<bdju>masiku[m]: definitely, have had this issue for months but only recently it hit me that it was two qt programs so I thought maybe I was on to something
<sneek>mothacehe, nckx says: One of them has a dead hard drive, and the other one dropped off the edge of the Internet just when I'd almost finished reconfiguring it. These things are cursed, but it shouldn't take much work once I can get to it.
<sneek>mothacehe, civodul says: thanks for the Cuirass fix!
<cbaines>either I've got my Guix Data Service query wrong, or it doesn't know of any nars for fixed output derivations that have references
<cbaines>the narinfo data isn't comprehensive yet though, so that doesn't suggest they don't exist
<civodul>it's a feature we don't use at all currently
<civodul>because it's weird: where would those references come from?
<civodul>it's possible to create such derivations, but it's a bit convoluted
<cbaines>I guess things from the internet could contain references, but that would be a bit odd. The builder inserting references in to outputs would also be odd though, at least in the guix packaging use case
<civodul>yes, and the references of the output must be a subset of the derivation inputs
<civodul>so you have to know in advance what the references will be
*mothacehe is downloading zstd substitutes at 20MiB/s, thanks civodul!
<mothacehe>uprading to the latest emacs is much more pleasant
<cbaines>the green successful builds showing up is a little fustrating
<pkill9_>something neat to create with guix is the concept of a 'distributed-operating-system' as opposed to just 'operating-system', which would be used in `guix deploy` to deploy many operating systems that get configured to communicate with eachother, and have it's own types of services designed for working like this
<dorian_greyscale>I'm trying to set up some user services but for some reason shepherd is not accessible on a user level. The herd command reports "error: connect: /run/user/1000/shepherd/socket: No such file or directory" any time I try to use it as a regular user, but it runs fine for root.
<BlackMug>though even if guix chooses to have namespaces its better than nothing and just leaving the package to be secure to the air flow
<rekado>pkill9: you don’t need to make it mandartory.
<rekado>pkill9: if we go with SELinux, for example, it’s a user choice to set it to enforcing or not.
<rekado>one problem is that files in Guix have unexpected locations, so policies are more difficult to write.
<lle-bout>raghavgururajan: it has quite a lot of dependents
<lle-bout>it seems it's just about freezing to release series the dependencies of gtkmm-2
<roptat>oh, it doesn't find its dependencies because the perl path is not set properly
<roptat>Can't locate NetAddr/IP.pm in @INC (you may need to install the NetAddr::IP module) (@INC contains: ../blib/lib @@INSTALLSITELIB@@ /gnu/store/8zvc5mvk0xm3ygrxsgpyy5ilxb5rzjry-perl-5.30.2/lib/perl5/site_perl/5.30.2/x86_64-linux-thread-multi /gnu/store/8zvc5mvk0xm3ygrxsgpyy5ilxb5rzjry-perl-5.30.2/lib/perl5/site_perl/5.30.2 /gnu/store/8zvc5mvk0xm3ygrxsgpyy5ilxb5rzjry-perl-5.30.2/lib/perl5/5.30.2/x86_64-linux-thread-multi /gnu/store/8zvc5mvk0xm3ygrxsgpyy5ilxb5rzjry
<roptat>-perl-5.30.2/lib/perl5/5.30.2) at ../blib/lib/Mail/SpamAssassin/NetSet.pm line 26.
<roptat>this @INC is not enough, it only lists local files and perl itself, but during the build, I think it was able to find its dependencies
<lle-bout>roptat: if you look at sendmail package there's some stuff about perl
<sneek>luis-felipe, madage says: I've also been having issues with pulse audio since some days back.. eating ram, hanging and muting sound if I plug/unplug headphones. Not every single time I did tho and I also couldn't debug it until now..
<raghavgururajan>Happens. The GNOME project changed the official name from gtk+ back to gtk.
<lle-bout>luis-felipe: I think artwork is really important and we definitely need more around GNU Guix, I think it makes one more happy to be on a system that is to have some visual identity sticking to it, like in GNU Guix desktop environments for example
<luis-felipe>abr, roptat: What issues do you have with ibus? Apart from the slowness to activate the input method, I'm able to type Japanese.
<lle-bout>luis-felipe: what do you think of the idea of creating better GNU Guix out of the box configurations? When we upgrade GNOME for example, would you be interested in an out of the box configuration that works and provides all GNOME features smoothly as well as some light GNU Guix-specific art work and theming so that users of that configuration can feel right at home
<lle-bout>I think it is really important we make better out of the box configurations of GNU Guix, I don't think it's a great thing that the only way for users to use GNU GUix is to first study how a GNU/Linux desktop must be assembled.
<abr>luis-felipe: dunno, I'm starting it from XFCE's session, exported GTK_IM_MODULE and friends from .bash_profile and nothing happens when I switch IM (the selection thingy shows up, but once I've stopped over anthy and the prompt goes back to my focused window, the IM is back to bépo, confirmed by the icon of the systray ibus icon)
<abr>I noticed there was already one ibus-daemon instance started from gnome-shell (?! not quite sure since I thought I wasn't using gnome, but I guess it comes from GDM)
<abr>which leads me to this remark : lle-bout I'm not sure about this «out of the box» perspective: it's practically what gave me the most trouble since I've started giving guix a try
<roptat>luis-felipe, I can type the combo key to change input methods, and it shows on screen I'm using anthy, but when I type I still use my other layout, no Japanese
***j is now known as jess
<lle-bout>abr: I understand what you are saying but also it's that GNU Guix is written in Scheme and has a whole new service manager, is a functional package manager, etc. Those are new concepts which can make it seem obscure at first but really when you figure that out everything becomes really accessible for hacking.
<balance>Hello! can anybody help with building bisq? (https://bisq.network/) cloned git repo, installed git-lfs, but have error while building "Could not initialize class org.codehaus.groovy.runtime.InvokerHelper"
<abr>I've never used NetworkManager: «1. oh, there's a nm-applet in my tray, cool ! I'll stop being geeky girl and give it a try 2. oh, it's auto-connected on my wired connection, how great ! 3. now let's configure the lab's VPN ! -> nothing works, don't have the rights to create a connection, nm-connection-editor crashes when I run it from sudo, the «gateway» setup I have to fill in the GUI doesn't
<lle-bout>abr: with the GNOME upgrade, I hope we will solve this networkmanager mess
<lle-bout>But to be honest, this is due to complex interactions between networkmanager, polkit, etc. which are problems not related to GNU Guix itself. To be frank, if networkmanager ended up this buggy on any distribution it would still be obscure, there's something we don't do right around Polkit I think.
<lle-bout>I think NetworkManager in general is not easily troubleshooted for errors.
<abr>lle-bout: no need to apologiz, I was only documenting my experience of «oh ! it looks all clean and production-ready» but then disappoints in subtle and surprising ways (because if there's something as mainstream as gnome and nm, I expect most of the users to be using it, especially since it's a default when you give it a spin for the first time — so then I'm puzzled, how can this thing not work
<abr>and no one has noticed ? it must be me ? but can I have done, I've pretty much only configured to extra-packages in /etc/config.scm)
<abr>(zzappie: about katakana / hiragana: this is the main story, but it's trickier though, they sometimes use katakana for no other apparent reason than emphasis, and some foreign words got in so early in the language that they forgot it was foreign and spell them using hiragana like «tabako» (tobacco), introduced by the portuguese missionaries in the XVIth cent.)
<marusich>I don't really know about the --with-long-double-128 (do we need to specify it when building gcc-final?), but it works so I'm inclined to let it be. That whole long double stuff in the toolchain sort of feels like black magic to me; I guess it's an area where I'm not super knowledgeable.
<lle-bout>raghavgururajan: I mean this: $ ./pre-inst-env guix build firstname.lastname@example.org
<lle-bout>raghavgururajan: I run this command to test things as a whole (your commits): $ ./pre-inst-env guix build -k -v1 glib libsigc++ glibmm gtk-doc gobject-introspection cairo cairomm pango pangomm gdk-pixbuf gdk-pixbuf+svg vala libgsf atk atkmm
<lle-bout>marusich: did you announce POWER9 support to #talos-workstation?
<lle-bout>I'll send an email to Michael from Phoronix about it
<lispmacs[work]>hi, I hooked up an old digital® lk46w-a2 computer to my guix workstation, and it has an cool but non-standard keyboard layout. I'm trying to figure out if there is a utility I could use to display what letter/function is being triggered when I press a key
<marusich>there is an email list specifically for reviewing blog posts...I think it was guix-blog@? guix-blogs@? I can't recall exactly. Anyway, if you CC me I will comment, and I will CC you when I come up with something
<lispmacs[work]>I tried looking at key-mon and xkbwatch but they are broke or don't seem to do what I was looking for
<cbaines>how easy is it to add substitutes for fixed output derivations to ci.guix.gnu.org?
<cbaines>It's missing /gnu/store/bna7d1nlmlkgnnjqf7sz9f1hybn2jypk-sudo-1.9.6p1.tar.gz, which is a problem at the moment as the source website is down
<cbaines>guix.cbaines.net provides a substitute for it, so the bytes are available
<marusich>cbaines, you recently mentioned that it is the output of a derivation that determines whether dependent derivations get rebuilt. but I thought one factor in determining a derivation's output path was the content of the derivation file itself. If the derivation is different, how is it possible for the output to be the same? Of course, I'm talking about non-fixed-output derivations; fixed output derivations obviously can have the same output
<cbaines>marusich, and I think it's this bit that means that a fixed output derivation can change at the bottom of the graph, but it's "hash" doesn't change for the purposes of computing derivation hashes (because the derivation still produces the same bytes)
<lle-bout>raghavgururajan: let me review commits individually really quick and let's push!
<cbaines>civodul, woo, using guix download on bayfront worked, so I'll go through those same steps on berlin in case anyone else is having the same issue
<marusich>cbaines, thank you for your link. It's helpful. I think I need to re-read the code while looking at it, but it's clear I probably misunderstood how the hashing works. In particular: "One more processing step takes place before this data in hashed. The input derivation files, shown in the green area are replaced with the hash for the corresponding derivation, computed through the same process that we are currently going through."
<cbaines>every time this comes up, I remember that I wanted to make the Guix Data Service clearly display where the hashes come from, and what data goes in to it, but I haven't got around to that yet...
<nckx>raghavgururajan: I meant your Guix signing pubkey (/etc/guix/signing-key.pub). It should be the same one mentioned in the error message you get when offloading fails.
<dustyweb>I just wanted to say again, thanks for being a nice community here
<nckx>raghavgururajan: By the way, 7D... is bayfront's public key, and it needs to be added to one or both of the active build nodes. I thought you had offload (root) access to bayfront and were trying to offload to it from your own machine.
<lle-bout>raghavgururajan: do we want that libcloudproviders thing? why did you add it, do you have a precise use case in mind? is it for Nextcloud? If so that's fine
<tissevert>cbaines: thanks for your slides that you linked earlier, they're very helpful
<dustyweb>sometimes I kind of get depressed about the state of... everything, but the experiences I've had in this community have just been so nice. like everyone, we have a lot we could do better but I really like being here and part of things, even though my contributions are small
<leoprikler>btw. did we fix all bad response stuff or did I just get lucky today?
<civodul>lle-bout: two comments from the trenches: (1) i would not recommend SWIG when writing bindings (prefer nyacc's FFI tool or hand-coded FFI bindings), and (2) sometimes it's more productive to implement the thing than to write bindings
<civodul>this is hand-wavy, that's my contribution to the discussion
<lle-bout>civodul: I didnt know about nyacc! And yes, always end up more ergonomic to write them by hand but I believe generation can also help
<lle-bout>dustyweb: I am saying this because maybe we need some HTTP2 library somewhere here
<raghavgururajan>Hey dustyweb! Looking forward to collaborate with you this Sunday. :-)
<lle-bout>dustyweb: It feels refreshing to see your enthusiasm into things, I also very like much acting like that wish I saw you more around here :-D - Do you hang out in other channels now? If so which?
<dustyweb>tissevert: I hope you find it so as well!
<lle-bout>tissevert: I have a feeling that everyone here has begun and is quite advanced in the process of being self-conscious of themselves with their emotions and the ones of others so that we try to pay attention to that and when we fail recognize and try to improve together
<raghavgururajan>lle-bout: For email@example.com, can you edit the inputs to remove *-version for cairomm, pangomm, glibmm and the try building?
<tissevert>wow, if that's true that's a lot ! I don't even ask that much : ) but yeah my first experience with this chan is really encouraging, especially considering I started by being quite bitchy about not getting a quick answer to my first questions
<genr8_>oof. I ran a command "guix pack --localstatedir --profile-name=current-guix guix" and it completely ruined all my profiles somehow. I can't even get the guix command to run now, "no code for module (gcrypt hash)"
<genr8_>it seems to have re-written everything on disk , and then re-linked stuff, but wrongly
<lle-bout>raghavgururajan: no gtk-doc and all its dependents are still broken
<lle-bout>I think we should solve it somehow but heh
<lle-bout>genr8_: woops, not sure what that involves here..
<lle-bout>raghavgururajan: you can't test it yourself?
<lle-bout>raghavgururajan: substitutes are available from my machine
<lle-bout>raghavgururajan: would be easier if you had commit access
<lle-bout>raghavgururajan: or we could agree to cooperate on a single branch somehow
<lle-bout>pineapples: I am actually not sure if you can get those compiled from an existing install without recompiling GNU Guix
<lle-bout>pineapples: to compile: "./bootstrap && ./configure --localstatedir=/var && make etc/guix-daemon.cil"
<lle-bout>raghavgururajan: please only make new commits by the way, otherwise it could be annoying for me to integrate changes in between my current history
<pineapples>lle-bout: thanks! I'll try the command out. If it somehow doesn't work, I'll give Debian a try. As much as I love Fedora, I need Guix installed on my backup OS, just in case something goes wrong.
<lle-bout>pineapples: it's always possible to "setenforce 0" as root temporarily if you need GNU Guix.
<lle-bout>raghavgururajan: could you get substitutes by the way?
<dongcarl>Anything that needed to be done for the openssl 1.1.1k release?
<lle-bout>dongcarl: what do you mean? I grafted it to the openssl package few hours after it was released
<lfam>Hey roptat, regarding the Python optimization changes, do you have patches online anywhere? Or a git branch?
<pineapples>ill-bout: I think I'll just give a Debian try :-) Thanks for your help!
<lle-bout>also must binutils be upgraded in tandem with GCC?
<lfam>There are still some run-time uses, if I understand correctly
<lfam>We should think about if the build-time uses are a security problem for Guix or not
<morgansmith>I was wondering if I could get some help. I'm trying to build something using the arm-none-eabi-toolchain that requires newlib-nano. If you read the nano.specs in our newlib-nano package, it really looks like the library should be named libc_nano.a but the only files available are libc.a. Any ideas? I'm not the greatest with gcc stuff
<lle-bout>lfam: I think nckx's CVE lint coloring patch would really help prioritizing for these issues, unfortunately the patch has a bug I am unable to find. I can't wrap my head around how json parsing works in GNU Guile
<raghavgururajan>lle-bout: So sorry, just saw your message. I had create few new commits and edit few existing ones. But mostly the former. This will be the last change for this batch. Fixed (glib|cairo|pango|atk|gtk)mm-foo builds. Updated my repo.
<raghavgururajan>Once again thanks a lot for helping me out. I couldn't have managed to finish it without you.
<lle-bout>civodul: do you think tooling could be made to transfer any removed package to guix-past automatically?
<lle-bout>civodul: sure hmm.. whether it is feasible to actually NOT use everything from back then when doing old things.. I think for archaeology you may be better simply using an older GNU Guix version altogether with time-machine probably
<lle-bout>oreoking[m]: which step exactly? what command do you want to speed up?
<lle-bout>oreoking[m]: guix pull compiles GNU Guile code to some byte code, and that can take some time, it also generates various caches which can also take time, but there's substitutes for those operations available from ci.guix.gnu.org where a server computes them for you and you can trust that server to provide you with them pre-computed instead.
<oreoking[m]><lle-bout "oreoking: guix pull compiles GNU"> how to set up
<lle-bout>oreoking[m]: substitutes are basically precomputed results of compilations or computations that would otherwise be done on your own machine, it can save you from re-doing them locally but it also means you trust the substitute server to not harm your computer by providing you false/malicious results of these computations
<lle-bout>oreoking[m]: how do you currently run GNU Guix? Do you have a VM with GNU Guix System installed?
<lle-bout>oreoking[m]: does it say "Downloading from ... ci.guix.gnu.org" anywhere?
<lle-bout>oreoking[m]: note: it's also possible the substitute server didnt have time to pre-compute the results yet therefore that you must wait to get them, also some times it doesnt have substitutes for some older revisions of GNU Guix because storing all of that is expensive
<lle-bout>raghavgururajan: let's hope Cuirass can help us now
<lle-bout>I also will try to run some more builds on core-updates myself
<lle-bout>cbaines: hey! hope you are doing well :-) - I would really love to run guix-build-coordinator myself on powerful machines to build core-updates fast along with the policy you were speaking of that cancels builds if a new commit alters any derivation that affects it
<lle-bout>cbaines: else I would run Cuirass if it had such policy mechanism to cancel expensive builds but I don't think it has yet
<lle-bout>genr8_: if the reproducibility issues with parallel building are fixed then we can enabled back parallel builds
<cbaines>lle-bout, cool, the way that works for the guix-patches instance I'm running is through build tags, for each new revision of core-updates, builds are queued (just those on the branch, based on a comparison with the latest master revision), and then any build tagged for the branch, but not relevant to the latest comparison is cancelled
<lle-bout>cbaines: do you already run those core-updates build? is there any place I can see results easily? Also can I add more machines to this? What's the priority of it now?
<lle-bout>mbakke: it seems accessing https://ci.guix.gnu.org on ungoogled-chromium doesnt work at all now. It seems related to the new sorting icons that were added, I am thinking this might be due to some image parsing stuff
<cbaines>I'm trying to setup the Guix Data Service, Patchwork, Guix Build Coordinator, ... such that it helps with QA stuff, so testing branches and patches, as well as things like periodically testing fixed output derivations
<cbaines>as for core-updates, I'm not currently queuing builds, but that's just because I rebooted the machine, and I haven't started the script up again
<cbaines>I think you've already got some connected machines
<lle-bout>cbaines: yes I got powerpc64le-linux and one x86_64-linux but they could be even more powerful :-)
<lle-bout>I havent been checking out builds for them because I didnt know where to find information easily for them
<jackhill>lle-bout: rekado: are we collecting pages ungoogled-chromium doesn't handle well? I think I may have found one: https://git.hcoop.net which, uh, as far as I know doesn't have anything wierd on that page.
<lle-bout>jackhill: yes.. gitweb seems to crash, also sourceware and qemu gitweb pages make it crash
<lle-bout>It's surprising how ungoogled-chromium never crashes for me on various really crappy sites but crashes on like gitweb or cuirass who are the most minimalist sites ever