<lle-bout>hi, ever got "Bad page state" errors in dmesg on GNU Guix? <lle-bout>not sure what causes them, not much online <Blackbeard>lle-bout: hey, why are you getting that? what are you trying to do? <lle-bout>Blackbeard, nothing in particular, just using my computer, ungoogled-chromium, emacs and terminals <lle-bout>I have Xorg with i3, linux-libre-4.19, also tried linux-libre and now trying out linux-libre-4.14 to see <lle-bout>linux-libre seems to crash early on, while linux-libre-4.19 lasts a little more, and for now 4.14 hasnt crashed, it could at any moment though. <Blackbeard>rekado: ok, seems like I got everything running, sorry about earlier messages, I can read localhost:1234 <Blackbeard>the web interface is running and I connected to Geiser <Blackbeard>lle-bout: what website did you visit with ungoogled-chromium <lle-bout>Blackbeard, it doesnt happen for ungoogled-chromium only, it's just an example, and I had many sites open, it didnt happen after I opened a particular site, it also happened while in emacs <Blackbeard>lle-bout: seems like it is a kernel problem from a quick duckduckgo search <lle-bout>Blackbeard, yup but it does not make sense, it points to diverse issues that don't really have things in common besides the actual error <lle-bout>I'm confused.. been suffering this for weeks now <lle-bout>Blackbeard, stable and 4.19, yes, also 5.4 <lle-bout>with stable, it happens DURING boot and freezs shortly later *dongcarl apologizes for the hotfix <nckx>dongcarl: You couldn't even finish the commit summary. *dongcarl will upstream more reproducibility patches as penance <nckx>(Nice work, you beat me by 5 seconds or so. Rebase… ‘…where did my patch go?’) <dongcarl>nckx: Haha yeah I literally got an adrenaline rush after I noticed what happened and started typing furiously <dongcarl>lle-bout: I don't have a good setup and forgot a cloing paren :-( <nckx>dongcarl: What's your set-up? If it allows this, it sounds unpleasant mainly for you. <lle-bout>Blackbeard, I'll try. Do you think it's better if I submit it upstream rather, or..? <dongcarl>nckx: I just use spacemacs... Through all this I learned to `check-parens` before pushing... *dongcarl goes to yak shave my setup a bit *dongcarl will look that up <sirgazil>I feel like I always connect to #guix to say that something didn't work... <sirgazil>I just "guix search lispy" and I didn't see it :) <sirgazil>And the package fails to install using Emacs package-install :) <sirgazil>Failed to verify signature ivy-0.13.0.tar.sig: <Guix-beginner7>How do I access gnu/build/python-build-system.scm to inspect the phases? <Guix-beginner7>I have a package that has a Meson build file but it also has a setup.py file to grab some modules. What would be a way I could tackle this? <brendyyn>lately my whole system has been freezing up unresponsive to even the power button ;( <nckx>raghavgururajan: PARTY HARD <raghavgururajan>nckx Now we have atleast working client. Its legacy GTK UI. I am still working in Qt version, but appears it requires Qt 5.9. We have only Qt 5.12. <alextee[m]>on second thought on the zynfusion thing, i think we can just have leave both as is <alextee[m]>and let users deal with it, i doubt someone will install both zynfusion and zynaddsubfx and then complain <nckx>I think you've made the right choice. *alextee[m] sends a torrent of patches to guix-patches <brendyyn>did i break a record by neglecting a 1 line patch for 3 years? <brendyyn>i think i went travelling in china at that point. <alextee[m]>guix build: error: icu4c-CVE-2020-10531.patch: patch not found ? <slyfox>i got by openldap patches merged recently. looks like i sent them in 2017 and forgot about them <atw`>mbakke: just wanted to drop a quick thank-you for the work on synapse! And also to apologize for some mail fumbling, some of it was definitely my fault, but there was at least one message that arrived on the ML out-of-order ... bad day for debbugs and me :P <brendyyn>raghavgururajan: is it on a different branch? i only see qt 4.8.7 <raghavgururajan>There is a monolithic source tarball that contains all modules. But Qt > 5 onwards, they are also released as separate module tarballs. <alextee[m]>is something wrong with the guix repo? i just did a fresh clone and i can't build anything <alextee[m]>guix build: error: failed to load './build-aux/compile-all.scm': <alextee[m]>ice-9/eval.scm:293:34: In procedure string-append: Wrong type (expecting string): #f <brendyyn>alextee[m]: you should start fresh with ./configure or even ./boostrap <brendyyn>every now and then some change occurs that invalidates the configure script <brendyyn>you could investigate why but i generally juts dump everything and start fresh <alextee[m]>i just did a bootstrap and successful configure, still same error, i guess someone made an oopsie <raghavgururajan>Folks! Is anyone familiar with the error "QMutex: destroying locked mutex"? <str1ngs>raghavgururajan: sound like the mutex should be unlocked before destroying it. <str1ngs>could be catching a bug in the program. <dhanvanthri>Hello people, I tried guix system around the time that there was a usb disk image for installation <dhanvanthri>However I was not able to get a prepackaged binary (specifically CDDA, which is now in the official repo) to work, some error involving unlinked libraries or something of that nature <dhanvanthri>I understand that guix is a source based distro, and compiling programs from source is the reccomended mo <dhanvanthri>But I was wondering if someone can give me a quick and dirty of getting prepackaged binaries to run in guix, I recently got into emacs proper, and I want a whole computing environment that feels like it <Blackbeard>dhanvanthri: guix is not source based distro if you don't want it to be <Blackbeard>dhanvanthri: are you running Guix the package manager right now? <dhanvanthri>Blackbeard: Honestly, I want to run it as such as much as possible. I've invested so much into my computing freedom that this seems like the next step <Blackbeard>dhanvanthri: ok, do you want to install it as your main distribution then? <dhanvanthri>Blackbeard: Yeah, I want to switch over. I got my emacs to be portable, and I want to start living in guix system <Blackbeard>dhanvanthri: wonderful, don't worry there won't be much or any compiling <dhanvanthri>I'm literally dding to a usb drive right now, I've installed it before, but, a lot of config seems totally different from how it's done in most distros <dhanvanthri>I don't even recognize the filesystem, it feels alien <Blackbeard>Don't worry I'll by here to help as much as possible, and guix has a graphical installer now <dhanvanthri>Blackbeard: This is the system onto which I'll be installing. The graphical installer worked great last time (I did have to edit libreboot grub for it to work last I tried). I'll be back on my phone as some anon user, tyvm for the moral support <3 <dhanvanthri>This looks like exactly what I needed the first time around :'O, I'll get at it after getting guix-sytem on here <pkill9>dhanvanthri: i just tested cataclysm dda release on my config with the fhs service and it works <dhanvanthri>pkill9: Just confirming, the "experimental" prepackaged bin? <pkill9>dhanvanthri: no, the stable one, i'll test the experimental one <dhanvanthri>pkill9: Tyvm, I imagine that if the stable bin works, experimental would too! <dhanvanthri>Please wish me luck, I'm about to install. I have logged into this irc channel from my phone as "dhanvanthri2". <raingloom>initial F2FS support in gnu/build/file-systems.scm finished ^u^ <raingloom>just tested it and it reads uuids and volids properly <raingloom>should i send the patch now or wait until everything else is ready? <dhanvanthri2>is the tor option on install mean ill be josting a relay? im a little confused <dhanvanthri2>Blackbeard: I'm seeing a flood of "substitute: upgrading subs from ci.gnu.guix..." <dhanvanthri2>Is each one a seperate package or did i hit some glitch? <apteryx>would someone have some experience with NFS? <apteryx>is it possible to keep a file under my personal directory, but have it shared in my NFS export as well? <apteryx>From what I can see symlinks wouldn't work unless it points to an exported file system, so that's a no go. <apteryx>bind mounts wouldn't work because they don't allow to modify the "origin" of the bind mounts, which is what I want. <dhanvanthri2>I just want to note that there are some substitutes with sub 1kbps downloads <dhanvanthri2>Blackbeard: I think that one of my downloads has hung (glibmm-2.56), ive waited about 5 min longer than the estimated 3 min, nothing moving on acreen <Blackbeard>dhanvanthri2: select the file and press 'insert coin' <Veera>dhanvanthri2: Are you outreachy <Veera>Blackbeard: Are you a Guix member? <Blackbeard>Veera: I am not sure how I would acquire membership haha <Veera>Blackbeard: In savannah pages members are listed! <Veera>Blackbeard: I also don't know how to become a member! <Blackbeard>dhanvanthri2: did you use graphical installer or command line? <Veera>Blackbeard: Do you have commit powers to git. <Blackbeard>dhanvanthri2: ok, let's give it 5 more minutes, if nothing had changed let me know <Veera>dhanvanthri2: What you use for net connection? <Blackbeard>Veera: no I don't. I had been a user for a while and I've sent a few patches but overall I don't know that much <Veera>dhanvanthri2: see it is showing 705B/s; the net is slow! <Veera>dhanvanthri2: no look at the current one! <Veera>dhanvanthri2: network congestions sometimes long. <Blackbeard>dhanvanthri2: ok. Cancell it please. Let me know if you are on a terminal er graphical install it <Veera>dhanvanthri2: don't; make a strategy first. <Blackbeard>I've never used the graphical installer, so I am not sure if it will put you back in graphical installer <Veera>dhanvanthri2: No use doing restart again and again. <Blackbeard>dhanvanthri2: just cancell but we can keep going from where you are <Veera>which iso version are you using? <apteryx>raingloom: sounds good! unless you'd like early feedback, you can wait to have the patch set completed, I think! <Veera>what options you selected for install? <Veera>you can select minimum and get a booting system! <dhanvanthri2>Veera: encrypted fs, single partition, ssh and tor and ff certs <Blackbeard>dhanvanthri2: are you back on the graphical installer? <Blackbeard>dhanvanthri2: is this a tty? Sorry for the dumb question, but I rather be safe <Veera>Blackbeard: Integration of desktop environments into GNU Guix <se-sm-ca>are there any packages that need maintainers/ports to guix? Looking to help out in some way <Veera>se-sm-ca: Have you done Guix-packaging tutorial? <bandali>iiuc guix participates in gsoc as a gnu project <se-sm-ca>Veera: nope I guess I'll start there, thanks! <brendyyn>se-sm-ca: there are tens of thousands of such progams. just start by finding something you use that isnt in guix <brendyyn>se-sm-ca: id like to see okular if you want to try that. <brendyyn>it might require a lot of kde packaging work though <rekado_>alextee[m]: I can’t reproduce your problem. <rekado_>I just fetched the latest version of Guix, bootstrapped, configured, and started building. <se-sm-ca>brendyyn: thanks! Yea I think I'll start off small after the tutorial but I'll take a peak <rekado_>at what point should I see an error? <raingloom>must system init copy everything all over again if i just wanna reinstall the bootloader? ;_; <raingloom>my poor SSD wasn't made for so many full rewrites... <apteryx>the above output loops for a while, while the disk is grinding <rekado_>raingloom: you can also just install the bootloader the hard way <rekado_>once you have it you can boot into Guix System and use “guix system reconfigure” <rekado_>…which will not copy things that are already in the store. <rekado_>“guix system init” is special because it can assume that nothing on the target location matters. <raingloom>that works, but setting up store paths manually is a pain. <raingloom>i might try to patch system.scm to not copy store items that are already present... <rekado_>raingloom: I mean you can run grub-install manually <rekado_>raingloom: this might be tricky because it can be hard to trust arbitrary files that happen to exist in the target location. <rekado_>will it be faster if you disable it? <raingloom>rekado_: yea, i was thinking of only skipping /gnu/store items. it's probably easier to overwrite the others. <rekado_>I don’t know… Say you’re reinstalling Guix because /gnu/store got corrupted <rekado_>so you “guix system init” and then all the bad stuff with the correct names is kept <rekado_>how did you end up in a situation where guix system init succeeded only up to installing GRUB? <rekado_>perhaps the better option here is to add a flag to *just* install GRUB for the built system. <apteryx>rekado_: how do I disable IPv6 for the NFS service? Or do you mean globally? <rekado_>I meant globally, but I’m sure there’s a knob for NFS somewhere <apteryx>cool. on the host, or both host + client? <apteryx>will try all all the options :-) thanks <raingloom>rekado_: hmm. i guess that can also work. i'm trying to use f2fs on a new system, so sometimes i run into errors due to misconfigured grub. <raingloom>hence the need to just reconfigure / reinstall that <rekado_>“sysctl net.ipv6.conf.all.disable_ipv6=1” is what I used in the past <Veera>rekado_: Is that only the Outreachy Project Mentor accepts patches for commit <rekado_>Veera: no, other people can also accept patches. <Veera>rekado_: I sent one patch of image editor for image-viewers.scm; DannyM somehow did not do any comment. Is there someone for this specific category who reviews? <rekado_>Veera: can you give me the URL of the issue? <xelxebar>Okay, so I'm booting a guix vm via qemu over ssh <Veera>xelxebar: does ssh connects? <xelxebar>Since this is over ssh, I cannot use the standard qemu window, so tried booting using qemu's -curses switch <xelxebar>Veera: I am not sshing into the guest, I am sshing into the *host* ***rekado_ is now known as rekado
<xelxebar>The problem is that the guixsd qemu image, by default, doesn't stay in text mode, so qemu -curses ends up just showing "1024x768 Graphics Mode" and nothing else. <rekado>xelxebar: yes, I had the same problem. <Veera>rekado: is there some change I have to do. <rekado>xelxebar: I don’t know of a good workaround, sadly, but I think Danny suggested rebuilding the image with some changes. <xelxebar>Annoyingly, grub also goes into graphics mode, but that's somewhat manageable by blindly manipulating the interface by memory <rekado><20191221200040.16d49dc3@scratchpost.org> on guix-devel <rekado>it’s really about installing via serial console, but I guess it’s a similar problem <xelxebar>The real goal here is creating a guix system image for google compute <xelxebar>Looks like some people have tried in the past, but I'm giving it a shot. *guix-vits Hi Guix; *afk* <raingloom>hmm, the init can't find fsck.f2fs, yet it's right there in $PATH. <raingloom>btw, i don't understand why Guix uses bournish or guile for its init. they are quite cumbersome and sh is already there, so why not use that? <raingloom>(at least until a scheme based solution becomes ready for regular use) <rekado>is sh really guaranteed to be there already? <rekado>the REPL also appears when the root disk cannot be mounted <rekado>so we’re guaranteed to have Guile <xelxebar>Finally! Putting vga=off on the cmdline ends up displaying a deprecation message, saying to use gfxpayload=text. However, since the guixsd iso and qemu images both use grub's graphics mode by default, this error message wasn't getting displayed on the serial connection... <xelxebar>What are the default login credentials on the guixsd qemu image? <raingloom>rekado: i guess it's not "there" there in the initrd, but it's widely available and isn't that large. <xelxebar>raingloom: Ugh. That was one of the obvious options I didn't try. <raingloom>then it should work. not sure about the qcow download on the website. <xelxebar>Woot. Finally have a guixsd vm running inside a gce instance with a decent serial console. <xelxebar>As a bonus from the gods, networking pretty much Just Works. Was expecting that to take another inordinate amount of time <xelxebar>Part of the annoyance is that you have to turn the right knobs on the google cloud side just to get qemu running <xelxebar>At least now I have a reasonable setup to try building the guixsd gce vm. <hulten>Are there any scientists here (natural sciences, especially Earth science)? <hulten>I'm asking because at least in my field not the right tools are used to do the work; I'd like to exchange ideas on how to approach these issues. ***marmulak is now known as marmulak[m]
<hulten>I can explain in more detail if someone is interested. <raingloom>uuuugh can i go back to guile from bournish? >_> <brendyyn>hulten: if you want to reach more people you can email one of the lists, maybe guix-devel@gnu.org is the most suitable. <macc24>ummmmmm, i heard that you guys want to replace linux kernel with hurd kernel <raingloom>macc24: that was an April Fools joke. but the Hurd support is real. only the deprecating Linux part isn't. <macc24>raingloom: i hoped that hurd would finally take off <rekado>raingloom: ,L scheme should also work <rekado>macc24: help wanted! We’re pretty close to bringing Guix System to the Hurd. Cross-builds and native builds both work already. <macc24>rekado: only part that i can help would be hurd-as-coreboot-payload <rekado>macc24: how would that work? (I don’t know anything about that.) <rekado>guix-vits: working on the Hurd or anything else? <macc24>rekado: well, instead of running grub/seabios/tianocore uefi/whatever else, coreboot would run hurd <macc24>rekado: in that case, hurd will be launched in >1s from pressing the power button <macc24>but useless for debugging, as flash chip wears out <macc24>and writing/reading to it is slot <macc24>and hurd may need some bios calls to run <rekado>macc24: is this roughly equivalent to building a root disk image and writing it to the BIOS chip? <rekado>guix-vits: have you used the Hurd before? <rekado>you could start by downloading the Hurd VM image mentioned in the blog post (it’s just a Debian GNU/Hurd image) and familiarize yourself with it. <rekado>then perhaps build a few things on your GNU/Linux machine with “guix build --target=i586-pc-gnu”, copy the software over to your Hurd VM, and see if they work as they should. <macc24>rekado: how does hurd load files from hard drive? <rekado>macc24: this question is too vague. It depends on how much detail you want. <civodul>jayspeer: it's part of "gcc-toolchain" <guix-vits>jayspeer: don't forget to restart the session, if something will not work after installation of "gcc-toolchain" ("err: xyz.h -- not found"). <civodul>or just follow the instructions that suggest sourcing ~/.guix-profile/etc/profile <jayspeer>and where can I find this library after installing this package? I've installed it before asking my question :( <jayspeer>or maybe it will be easier to desribe whole issue, since this seems to be classic XY problem <jayspeer>so... I want to use cf-uaac from ruby gems. to do that I installed... ruby, then installed uaac by running `gem install cf-uaac` <jayspeer>I had exported GEM_HOME and GEM_PATH since those were required; uaac is in the path, great! but it fails with `require': libstdc++.so.6: cannot open shared object file: No such file or directory - /home/jayspeer/.gem/gems/eventmachine-1.2.7/lib/rubyeventmachine.so (LoadError) <civodul>i'm not familiar with Ruby so i can't really help <civodul>however, it seems that "gem" downloaded pre-built binaries, the rubyeventmachine.so file above <jayspeer>I've only used stuff from guix repo (apart from uaac itself) so I've assumed this was the correct place to ask <civodul>pre-built binaries "from the net" in general won't work for lots of reasons: ABI incompatibilities, incorrect file names, etc. <jayspeer>no, it's pulled from ruby gems. I'll build locally to verify <civodul>so i would recommend making a Guix pakage for this <civodul>using "guix import gem" as a starting point <jayspeer>civodul: thank's for that insight, I'll do that <civodul>but really, i'm not compentent when it comes to Ruby, so perhaps there are other solutions <jayspeer>what you said about libralies incompatibility makes sense, so I'll building this locally *guix-vits Hi zzappie; afk <zzappie>jayspeer: It may sound very bad but sometimes people pack gems or python packages that rely on some sort of object file on their say ubuntu machine and just push it to the registry never thinking about this dependency. <jayspeer>zzappie: this seems to exactly be the case :/ I can't even build package locally. Since I need this thing for work I'll just use ruby on ubuntu <zzappie>jayspeer: you can use 'guix environment' if you have to quickly install loads of packages alien and set LD_PATH there it will work, and won't break your profile <Veera>rekado: need feedback for bug #40281; apologies if you are looking into it <jayspeer>zzappie: it will take me a lot time to figure out ruby build system and hopefully I only need to use this tool once, maybe twice. If I happen to use more often I'll figure this out <zzappie>sneek: later tell jayspeer: sorry I didn't say clear enough. I meant that you can use 'guix environment' to isolate 'ruby gem' installation. Without altering your profile. take a look at https://paste.debian.net/1138220/. will create isolated HOME where you can install arbitrary packages using gem. To get rid of libstdc6++ problem you cab install gcc-objc++@ 6.5.0 in that profile <civodul>when did we loose the keyboard layout chooser in GDM? <civodul>i think there used to be one, right? <Veera>civodul: As 4 days are only left for Outreachy Deadline can patches related to it be little bit prioritized for review and acceptance <Veera>civodul: patches aren't be many <pinoaffe1>hulten: on the topic of guix in the sciences: I think that in branches where it's common to work with lots of software and lots of data, the appeal of guix might become clear <pinoaffe1>i've heard several non-tech scientists complain about reproducibility issues with data analysis, so I think that once guix is more mature, they might be interested in using it (though a lot of issues / incompatibilities in workflow might prevent this) <civodul>did g_bor[m] or cbaines comment on them? <civodul>i think i'm officially not mentoring <Veera>civodul: dannyM is my mentor <civodul>could you remind me the issue numbers, though? <civodul>Veera: i can ping Danny, but can you give me the issue numbers so i can take a look? <Veera>civodul: yesterday dannyM sent a mail reminding of outreachy application submission <Veera>civodul: #40320 #40322 they have dannyM as participants <Veera>civodul: #40281 for which in morning contacted rekado and he reviewed it <civodul>so really Danny is the one taking care of it, with g_bor[m] taking care of the overall Outreachy participation <civodul>but i'll see if i can help with patch review <roptat>civodul, I was wondering the same thing when I saw there was no keyboard chooser on gdm on my work machine (under fedora). It's probably an update? <civodul>roptat: probably, it's the second time i lose my hair on this gnome-shell/localed mess <roptat>(a pretty bad move I think, but not sure if we can do anything about it...) <apteryx>civodul: is this why the keyboard defaults to US on GDM? I had opened an issue about it. <roptat>on my fedora workstation, I have to enter the password in the default US keyboard at boot, but then when opening the session, it switches to bépo, and when I lock my screen, I need to enter my password in bépo <roptat>which would be bad if I weren't the only user of that machine, us other users would have to know their password in bépo too :/ <roptat>it's annoying enough that I thought about sending a bug report to fedora. Should I send it upstream instead? <civodul>apteryx: i had fixed it in 607fcc75404e2b1fc74affcf372b4a6a789ac55e <civodul>roptat: so you're saying Fedora has the same bug? <roptat>at least I can't find a way to change my keyboard in GDM <roptat>the top right corner has some icons, but when I interact with it I can only shutdown, reboot, etc... or (iirc) change network settings *zzappie just found out about Do No Harm License <bricewge>Maybe they have hidden it under Alt, like the suspend button... <roptat>looking at bug reports, it seems that gdm should be able to show a keyboard setting if there is more than one keyboard layout configured in gnome <civodul>roptat: yes, see keyboard.js and keyboardManager.js in gnome-shell <roptat>so we're probably missing something <civodul>it's reassuring that Fedora has that kind of bug <civodul>roptat: see 607fcc75404e2b1fc74affcf372b4a6a789ac55e <roptat>I'm using gnome on fedora, but openbox on other machines <roptat>on guix, it's not an issue because I could configure the keyboard to be bépo at boot and afterwards <roptat>but it seems I can't set gdm's keyboard layout(s) on fedora, with gdm+gnome :/ <civodul>so this is not #fedora :-) but we definitely have the same problem currently <roptat>can you pass more than one keyboard to the localed service? <roptat>I think I currently use it and that's why I have bépo at boot as expected, and there's no keyboard selector because I only have one keyboard configured <civodul>our localed is heavily modified, it just replies whatever layout was in config.scm <civodul>on master, you cannot have more than one layout <civodul>but we were discussing it on the list <roptat>oh I see. I suppose if you set more than one layout, then gdm will pick them up? <hulten>pinoaffel: Yes, this makes sense. There is some discussion/complaining about reproducibility in my field (ocean biogeochemistry and modelling) as well. <hulten>However, they are more concerned with approximate reproducibility, often based on the published equations (where code is often not available, but this is getting better). <mbakke>uff, the updated CMake on the core-updates branch has the same problem with not finding a C compiler on armhf when running through QEMUs user-mode emulation. <hulten>I also got positive response on guix-devel. <civodul>roptat: i arranged to strace localed from its start, and it replies the right "properties" (layout, variant, etc.) <roptat>I thought it was about the absence of a keyboard layout chooser <civodul>roptat: whether there's a layout chooser depends on the reply that keyboardManager.js gets from localed <roptat>so there used to be a chooser even when there was only one keyboard? <roptat>ok, but that's not really useful, is it? or is it a chooser for *any* layout, and not just configured layouts? <civodul>also, gdm would use localed to determine the default layout <civodul>i don't remember the details of the chooser <roptat>the default layout is working here at least <civodul>but currently there's also the problem that it's a us layout no matter what <roptat>absolutely, but maybe a bit outdated (on master, not core-updates if that makes a difference) <civodul>what commit does "guix system describe" report? <roptat>e2bd0af373d3abfba645647e6ec674670f2bd1e4 <roptat>(maybe not rebooted since, then the last system was built from a commit from march 17th) <roptat>ah I don't use localed, I modify the gdm service directly <roptat>with modify-services %desktop-services, I replace the xorg-configuration field in its configuration <roptat>(xorg-configuration (xorg-configuration (keyboard-layout "fr" "bepo"))) <roptat>this is also the value of the keyboard-layout field of my operating-system declaration <civodul>ok that's what set-xorg-configuration does, right? <civodul>ok, so this is broken on master AFAICS <roptat>guix system vm works, and I can enter a username in bépo in gdm <roptat>but I'm not sure if that's because gdm is configured correctly or if it's because of my host system... <roptat>ah after changing the host's keyboard, it didn't change anything in the vm, so I guess it's correct <roptat>now let me build the same, but without changing anything in the gdm service <roptat>civodul, if I don't change the gdm service type directly, I get the us layout, indeed <civodul>roptat: what if you change it with set-xorg-configuration as in desktop.tmpl? <raghavgururajan>Does guix patches (patches inside package definitions) accept only git patches/diff? Will gnu util diff work? <apteryx>raghavgururajan: it works as well IIRC. <raghavgururajan>apteryx So far I have been fetching remote patches/diffs. Where should I place the local patches/diffs? <apteryx>you can follow the naming convention of patches found under gnu/packages/patches <raghavgururajan>So regardless for what module the patch is applied, all are placed in the same directory? <apteryx>if the patch doesn't contain any reference of where it originates, I like to add this information with comments (#) at the top of the patch file. <apteryx>raghavgururajan: yes. The search-patches helper will look there. <raghavgururajan>apteryx How can I obtain the patched source? `guix build --source` gives me only original source. <apteryx>it should have been patched, or error attempting to do so. <apteryx>did you inspect the source build log? any clues there? <necrophcodr>Because I tried the example linked, and the image doesn't even boot. <necrophcodr>It also only supports i386, so that's a bit of a step back from more than 10.000 packages supporting both x86, ARMv7, and AArch64 :( <jonsger>necrophcodr: it was published on the 1st April. So don't take all too serious <civodul>necrophcodr: there's actually work on GNU/Hurd support, but the bit about deprecating Linux support is a joke <necrophcodr>civodul, oh okay, that brings my heart rate down a bit again. Thanks for clarifying for less humorous people such as myself :D <civodul>necrophcodr: sure :-) i guess we'll add a sentence on the blog post <necrophcodr>Getting GNU/Hurd support would be great, and especially getting GNU/Hurd and GNU Mach up to the point where there are significant security features in a x86-based build! <civodul>there are lots of goodies we could get from that <ArneBab_>I get failures on guix system reconfigure: ice-9/eval.scm:619:8: Throw to key `srfi-34' with args `(#<condition &message [message: "'/gnu/store/xlcbi7dc89n4wvyz4jk6j0g4590ymi6q-grub-efi-2.04/sbin/grub-install --boot-directory //boot --bootloader-id=Guix --efi-directory //boot/efi' exited with status 1; output follows:\n\n /gnu/store/xlcbi7dc89n4wvyz4jk6j0g4590ymi6q-grub-efi-2.04/sbin/grub-install: error: /gnu/store/ <ArneBab_>xlcbi7dc89n4wvyz4jk6j0g4590ymi6q-grub-efi-2.04/lib/grub/i386-pc/modinfo.sh doesn't exist. Please specify --target or --directory.\n"] 7ff06ee688c0>) <ArneBab_>i386-pc/modinfo.sh really does not exist, but x86_64-efi/modinfo.sh <ArneBab_>do you have an idea why it tries i386-pc? <ArneBab_>uname -a reports Linux fluss 5.1.2-gnu #1 SMP 1 x86_64 GNU/Linux <ArneBab_>(I cannot boot this system natively since tuesday; I’m currently running grup from a USB-stick and rewiring the disk) <civodul>roptat: "guix system vm desktop.tmpl", with a non-US layout, gives you GDM with that layout? <roptat>I set keyboard-layout to fr bepo, removed gnome-service-type and xfce-service-type because of disk size too <Veera>civodul: Thanks for doing commit of libqalculate <Veera>civodul: I should not use git commit -s. is it? <roptat>no, the commiter uses -s, you are the author already :) <Veera>how I am the author; but i don't put any author info <Veera>roptat: the From Mail address? <roptat>git adds author inforrmation for you, see "git config user.name" and "git config user.email" ***MinceR is now known as MinceR`
***MinceR` is now known as MinceR
<Veera>also I should not mention adding copyright and adding modules needed by the package in git commit msg? ***elibrokeit is now known as ztrawhcse
***ztrawhcse is now known as elibrokeit
***elibrokeit is now known as ztrawhcse
***ztrawhcse is now known as guys
<roptat>you can look at "git log" for examples ***guys is now known as elibrokeit
<Veera>i tried that I only see those made from guix members <raghav-gururajan>apteryx Diff generated by gnu diff doesn't have the header of which to file it belongs to. Like how git diff does. <raghav-gururajan>Is there way to mention which file to patch, inside package definition?/ <lfam>raghav-gururajan: What command did you use to make the diff? It should say which file it is for <lfam>I guess I would just use Git here <lfam>In that case it doesn't apply to any file, it's not a patch. Just a diff <lfam>It's not clear to me what you are trying to achieve with this diff :) <lfam>Which file do you want to apply it to? <lfam>I misunderstood your question at first and now I feel less qualified to answer it <lfam>You can pass arguments to `patch` to specify which file to apply the diff to <lfam>Check the patch man page and look in our packages for 'patch-flags' *raghav-gururajan --> afk <lfam>Does anyone have a tool for Vim that allows moving S-expressions before or after neighboring S-expressions? Basically, for re-ordering package definitions? <jakobrs>What license is the Guix package repository licensed under? <rekado>this means that the package definitions are all under GPLv3+, but it doesn’t mean that all the packages themselves are. <rekado>lfam: I just wanted to recommend paredit for vim, but transpose-sexp is actually part of Emacs, not paredit :-/ <lfam>I'm using paredit but I don't think it can do this :( <lfam>I need to reorder like hundreds of packages in crates-io.scm :/ <lfam>That module requires packages to be ordered alphabetically (easy) but they actually can't be exactly alphabetically. Inherited packages need to be reversed or the inheritance breaks <lfam>Personally I don't think it matters what order packages are in <lfam>But I already want to relax some of the guidelines so I should try to follow this one <rekado>would it be easier to delete the comment that asks people to order the packages…? <lfam>Yes but that wouldn't be friendly <lfam>I suppose the people that have been doing all the Rust packaging have a reason for it <lfam>This is also making me think that we should discourage package inheritance <lfam>I'm not sure I've ever been glad I used it <rekado>I think it’s great for personal package variants, but not otherwise <lfam>Long-term it creates a burden and causes bugs <lfam>I assume that all the package-source and package-version things in (guix packages) don't break the tools, on the other hand <lfam>There are similar issues specific to Rust for us. There is no package dependency graph for any of the Rust packages :/ *rekado doesn’t know much about Rust in Guix at all. <lfam>I'm just dipping my toes in. The cargo-build-system does not use the normal package 'input' fields, but instead uses Guile keys in the arguments field. So none of our tooling works <lfam>This is because Rust packages only use the source code of their dependencies, and this is the way that goal was achieved. Maybe we could use package-source somehow instead <jackhill>how do Go do it? Is it similar? I also no little about Rust. <lfam>Overall, it does work. My WIP package successfully integrates with FFmpeg <lfam>jackhill: The go-build-system actually installs the source code as part of the built artifact in /gnu/store, so depending packages can access it easily <lfam>The go-build-system needs serious work, too, though, especially as the upstream Go tooling has changed significantly <lfam>I'm not sure what the right path forward is for Go modules. I have some great notes about it from another Guix hacker but haven't found the energy to really understand them in this context <jackhill>indeed. It would be nice if a common abstraction for source-only-inputs could be used (if appropriate) <lfam>With Go I always felt like I was missing something... even more now that we can't re-use built objects <lfam>But I pushed the work through because I wanted to use some Go programs. I'm hoping to see some other people start working on the go-build-system :) <lfam>I think it should be possible to instrument Go's memoized build cache somehow to be able to re-use built objects <lfam>They basically copied us ;) <jackhill>I'm interested, but need to find the time to understand it better <lfam>Either that or have a generic source-only method as you suggested <lfam>I have... a lot of free time now. But I want to spend it on packaging rav1e. After that is done maybe I'll turn my focus towards this stuff. <lfam>It could be useful to just file some bug reports about the source-only thing and the missing Rust dependency graph thing <lfam>Although, it's hard to feel confident about adding rav1e without knowing I'll be able to use the Guix tools to maintain the package later. It's a huge number of packages and lines of code that will need to be maintained <lfam>The WIP recursive crate importer works great but if we can't maintain things later... that's tens of thousands of lines of packages with no real ability to work on them <jonsger>lfam: so we need a `guix maintain` feature :P <lfam>It's called `guix refresh` :) <lfam>But it doesn't work if the package dependency graph is missing <jackhill>lfam: thanks for submitting those tickets <Blackbeard>rekado: hey! Do you have time to help me get started with mumi? It is working now and I can connect with geiser <Blackbeard>rekado: what do you think is an easy problem for me to start with <apteryx>lfam: based on our documented naming conventions that'd make it 'ruby-ruby-parser' <lfam>It doesn't sound like there is a better option! <dhanvanthri>Hey guys, after 16hrs of failed insyalls, i am cureebtly doing a manual install of the config file created by the graphical installer <rekado>Blackbeard: I don’t know what you’re comfortable with. If you know JavaScript you could work on styling query input. <dhanvanthri>whenevrr it hangs, i C-c, then rerun guix system init on the config <lfam>Hey dustyweb, are you around? I was wondering if you recommend using DrRacket from the Guix Racket package <lfam>dhanvanthri: Okay, let us know if we can help. It shouldn't be that hard... <rekado>Blackbeard: the idea is to recognize supported query keywords and apply some style to them to give feedback to the user that the keyword is supported. <dhanvanthri>So my question is whether i now have pointers to programss ghat arent actually installed <rekado>so when I type “submitter:rekado” I’d like that to be visually distinguished from any other query text <lfam>dhanvanthri: What do you mean by "pointer"? <lfam>Blackbeard: It works well? <dhanvanthri>lfam: so lets say that the guix sustem init script is downloading pkg1, but it stalls and hangs, when i quit and restart the script, is pkg1 properly installed or not? <lfam>dhanvanthri: Until `guix system init ...` completes successfully, nothing is installed properly <lfam>After it completes, everything you put in your config.scm will be ready <lfam>There is no in-between state <dhanvanthri>lfam: its literally happening right now, gzip says 1kbpa, but the bar is not axtually progrwssing, im about to quit and restart guix sysrem init <dhanvanthri>sorry, insyalling on comp, so typing on phone. feels fuxking trash man <lfam>It sounds like your connection our servers is not that great :/ <dhanvanthri>lfam: Almost every file dls around 1mbps, and then some just stop working in the middle <dhanvanthri>Blackbeard: still been failing to install since last we spoke :'( <Blackbeard>dhanvanthri: if you stop it, you don't need to reboot <lfam>dhanvanthri: Where are you? <dhanvanthri>lfam: Different packaged slow down to zero each time and force failure <lfam>I wish we were still using the CDN... <dhanvanthri>I justvcancelled and reran guix system install ahain, 3rd time for this attemot <dhanvanthri>Atleast it seems to continue where i left off, unlike tge guinstaller <bdju>When was the last time the ISO was updated, and will it be updated again any time soon? I will likely do a fresh install on a new machine soon, so I'm curious. <lfam>It corresponds to the latest release, 1.0.1. So, May 2019 <bdju>I haven't done an install since the graphical installer was added, or BTRFS support. I'm thinking of trying BTRFS for the first time when I do the install. <bdju>Oh wow, so nearly a year old now. <lfam>We are aiming to release again soon <dhanvanthri>Okay, another hang at nss-3.43, i exited and reran guix system install <dhanvanthri>It didnt attempt to download it again, it just skipped to wayland-protoxols-1.17 <dhanvanthri>lfam: So thus is what im asking about, am i just about to experience catastrophic failure when the insyaller is done and my system needs nss? <lfam>Like I said, when `guix system init ...` completes successfully, with an exit code of 0, then the system will be installed <lfam>If your config.scm requires NSS, then NSS will be installed <lfam>It's not like Debian where things can be installed half-way <dhanvanthri>Well i dont know what nss is, its just what im seeing on my sxreen, but tgats super heartening to hear <lfam>NSS is Mozilla's TLS implementation, and nss-certs is their TLS X.509 certificate store <dhanvanthri>I see, that was specifically requested by rhe graphical config wizard, so i guess no meed to worry <dhanvanthri>Well, im still worried about the other hing packages yhat arent explicitly in my config file <dhanvanthri>But i guess the whole idea is that guix is about to just take care of it for me :-) <apteryx>do we have unifdef already available in Guix? <lfam>dhanvanthri: Yes, it should work. But please send a bug report about the download failures to <bug-guix@gnu.org> <dhanvanthri>lfam: i honestly need a snall walkthrough, just email that adress a write ip of the bug? <dhanvanthri>Yo it completed, im fucking exhausted, gonna sleep now, ill repoet back in the morning <dhanvanthri>Im not able to boot into it, libreboot grub hung after entering passphrase <dhanvanthri>I'M IN GUYS, it was the libreboot password lag, it's default is 1MINUTE? Anyway, everything is working <dhanvanthri>However I don't have a default user etc, and root is not pw protected atm... So I'm outta here and ACTUALLY going to be d now. TYVM <brown121407>Hi! Is there a way to access inherited fields in a package definition that uses (inherit base-pkg) ? <dustyweb>lfam: yes drracket from guix package is good and fine :) <lfam>Great dustyweb :) I got a copy of How To Design Programs <dustyweb>lfam: admittedly I ususally use racket-mode in emacs <dustyweb>but I've found drracket to be quite nice <divansantana>Q. Why is there no tor-browser in guix. I would expect it to be there knowing the community. Or is there some tech issue blocking it? <dustyweb>divansantana: there's interest but there are some threads about it... <dustyweb>I don't remember offhand what the challenges were <lfam>divansantana: Ultimately it's considered dangerous to use a distro package of tor-browser <dustyweb>lfam: because it may get out of date too fast? <lfam>divansantana: If one is really trying to stay anonymous, they should not use a different tor-browser from everybody else <lfam>No, because it would be easy to fingerprint <lfam>Combined with the difficulty of building Firefox, nobody has seen the work through to completion <dustyweb>lfam: I'm not sure that makes sense to me <dustyweb>what in particular would make it easier to fingerprint <lfam>I don't know off-hand but something would leak through. Already it's really trivial to identify individual users through browser fingerprinting, and this is such a special case that the risk/reward is just not okay <lfam>Plus, we would never be able to keep it up to date, which would be another way to start de-anonymizing users <Blackbeard>dustyweb: I agree with you, as long as there is no JavaScript running and ads being blocked is really hard to fingerprint <lfam>I know browser users can be fingerprinted by canvas size and available fonts <dustyweb>lfam: I thought that tor blocked system fonts for that reason... <dustyweb>that's why using tor browser bundle is a good idea, they do all those things <lfam>Yes, it's a good idea to use the bundle <lfam>It's not a good idea to make another flavor of it <dustyweb>lfam: I think the desire is to package the tor browser bundle's patches pretty much as is on top of firefox/icecat <lfam>Yes, eventually... in practice we know it's actually really tricky <dustyweb>but currently icecat ships with a tor button thingy <dustyweb>and may confuse users into thinking thye are equivalent <lfam>I mean... we almost removed icecat entirely due to issues with maintaining it. It's really on life-support <lfam>There is not a lot of interest in working on it <dustyweb>maybe we should just switch to tor browser bundle as our base <dustyweb>I mean, I generally think a) the most interesting thing about tor is onion services as a p2p network rather than anonymnity and b) I'm skeptical the anonymnity works for a persistent state actor and c) brwosers are nearly fundamentally broken from a security perspective anyway <lfam>I agree, I don't bother with Tor for those reasons <dustyweb>but still, shipping tor-browser-bundle seems like a good idea <dustyweb>I think tor onion services are interesting <dustyweb>tor as an onion router for the normal internet is not <dustyweb>but I"m interested in using them as a p2p network system <lfam>There's been discussion within Guix and with upstream Tor previously. I think they are worried about 3rd parties creating alternative packages. They are extremely interested in keeping the deployed base totally homogenous <lfam>Even Debian doesn't package it <lfam>They have a thing that just downloads it from Tor <dustyweb>lfam: how about we just have something that unpacks it with sudo right into /gnu/store *dustyweb imagines lfam's face melting in response to that joke <lfam>I knew you couldn't be serious :) <divansantana>lfam: dustyweb: See there is a lot of interest. Thanks for response. Is it easy to run the tor browser downloaded from the site on guix? <dustyweb>divansantana: I've never tried... I'm skeptical it will be, but hey <rekado>divansantana: it probably wouldn’t work without at least patching the binary to use the loader of Guix’s glibc. <rekado>you can try using patchelf on it and see how far you can get. <rekado>Blackbeard: you can see some of the valid query keywords in process-query in (mumi messages) <Blackbeard>rekado: what is messages.go I've noticed guix also has .go files <rekado>it’s a Guile object file containing bytecode. <rekado>nothing to do with the Go language <rekado>Guile is quite a bit older than Go. <apteryx>hmm, our ruby-gherkin package is broken. It won't load at all: require 'gherkin' -> LoadError. That's because it was made for a Ruby more ancient than what we have. <nckx>apteryx: That would sound like a challenge. <rekado>Digit: you can still be excited! <rekado>Digit: we won’t deprecate support for Linux, but it is true that a lot of progress has been made in improving Hurd support for Guix. <nckx>Digit: Only the negative parts were a joke; the positive parts are true! <nckx>(Well, not the time frame, but still.) <alextee[m]>is this not how to match "until the end of the line" ? ".*$" <nckx>alextee[m]: That's exactly what it will do, *including* the end of the line (newline). <nckx>So your replacement line might want to end in "\n". <alextee[m]>yeah, doesn't seem like i'm matching anything though <alextee[m]>i think it's the [ ] , needs to be \\[ \\] probably <nckx>alextee[m]: Whenever your pattern contains ", \, $, (, [, +, or (especially) ‘.’ — stop and think 😉 Some people just escape random things until it ‘works’ and that gives me the heeby-jeebs. <nckx>All of those except " have their own meaning in regex.