<leoprikler>sure, but one would like to keep inputs minimal, e.g. if it only includes gstreamer headers, that's gstreamer only <g_bor>I got much further with my quest on creating an automatic installer. <g_bor>I still have something missing. <g_bor>it goes like icecat-gnuzilla-fixes.patch not found. <g_bor>Do you have any idea how to resolve this? <leoprikler>i found out that dropping gst-plugins breaks clutter-gst <raghav-gururajan>leoprikler I evaded the "no cargs .. libmusicbrainz" error by adding "neon" as input. <leoprikler>g_bor: I'd suggest not building icecat in the installation image if you can avoid it. <leoprikler>but just to be up to date, what is the "mission" of your installer <leoprikler>raghav-gururajan: evince is missing gi, I'm working on putting it in *raghav-gururajan does not know what is 'gi' either xD <g_bor>leoprikler: the problem is in resolve-interface gnu packages gnuzilla. <g_bor>I would not like to build it, it's just the module is pulled in. <g_bor>and the named path is really missing :) <leoprikler>raghav-gururajan: successfully built /gnu/store/4y3qrqg2da1b6z0xhjk5rddgz4znscyi-sushi-3.32.1.drv <g_bor>leoprikler:: I don't know yet. <g_bor>I will have a look at the git log, but first check if a pull fixes it... <leoprikler>gnuzilla should probably be present given its contents <g_bor>leoprikler:it was removed on nov. 16. <g_bor>34369f310319b040d18864ff7435b8ee5f9a26a6 <nckx>g_bor: Do you have a stale Makefile{.in,} lying around? <raghav-gururajan>leoprikler It appears the pkg-def is same? Did you add anything other than 'neon'?? *nckx was summoned by the holy words ‘gjs was a mistake’. <g_bor>nckx: no, this is actually built completely out of tree. <g_bor>This is just a system definition, albeit not a simple one. <nckx>g_bor: And you're certain there's no stray ‘grep -r icecat-gnuzilla-fixes’ in that Guix anywhere? <g_bor>nckx: actually I am not completely sure. What I am doing now it trying it again in a pre-inst-env. Hope that helps. <leoprikler>I amend my previous comment: gjs was a hard mistake. <raghav-gururajan>leoprikler Could you rephrase what you meant in a straight forward way please? I am not able to read between the lines, <leoprikler>current evince does not build gi, so you have to patch that in addition to adding the actual package <brettgilio>civodul: hey, in your email to bandali and I, are you wanting us to start designing the haunt page now (on a different host) and then move it to the dedicated Savannah repo once we have something, or wait for the Savannah repo and then begin working on it? <seepel>Hello all, I just recently installed GuixSD! I noticed that there are a lot of emacs packages in guix itself. Is the recommended way to install emacs packages through guix rather than melpa? <raghav-gururajan>leoprikler Hmm. I get the same error even after using your build arguments. <leoprikler>are you just copypasting my stuff or do you use `git apply`? <pkill9>how is it preferred to send patches in email: as and attachment, as inline (e.g. a message above the patch), or the whole email being a patch (i.e. what git send-email does by default, with the email message in the patch below the "---" line)? <leoprikler>pkill9: git send-email is preferred, but if that does not work for you for any reason, attachments are fine <leoprikler>i.e. you don't have to force your MUA to conform to the format if it involves an unreasonable amount of copying and pasting <civodul>brettgilio: i'm saying you can start writing a page, a we'll host it somewhere <lekzikon>seepel: Both should work, but one of Guix goals, if I'm not mistaken, is to replace all language-specific package managers. <lekzikon>And, by the way, "GuixSD" is a deprecated name, it is now called the Guix System. <pkill9>which repo has the working npm importer? <seepel>Ahhh, Guix System does seem nicer. As far as installing emacs packages do most folks use guix then? <leoprikler>Guix for Emacs isn't broken as opposed to g-sorcery 😉️ <leoprikler>And with each passing day I wonder why I haven't migrated that machine from Gentoo already <seepel>leoprikler: Thanks for the info! <oriansj>I thought guix was going to fall back to pulling from software heritage when downloaded source file's checksums fail; so we don't run into the same old issue of upstream changing the file without bumping the number and all of the guix builds failing <leoprikler>if the link still exists but points to a different file, that's an error <oriansj>but if the file upstream still exists but now has an entirely different checksum when downloaded <seepel>I have come across another question, I am interested in installing Guix System on an old Macbook 2,1 which seems like it is pretty much fully supported by linux-libre. The problem I'm running into is that while it is a 64 bit machine, the UEFI firmware is 32 bit. I found some instructions for debian that I'm not really sure how to adapt: https://wiki.debian.org/InstallingDebianOn/Apple/MacBook/2-1. Does anyone know off hand an easy way <str1ngs>seepel: maybe you can use refind to boot guix <seepel>Ah interesting, I'll investigate. I notice the installer image creates four partitions, the first and last have type microsoft basic data, the second has type EFI, and the third has type HFS. I can't seem to mount the two microsoft basic data partitions, and I assume the HFS partition is the actual system. Is there someplace I could look to understand the boot process? Either the guix manual doesn't explain it, or I'm too dense and <seepel>By the way, I don't mind digging around a lot, the whole reason I'm here is to learn, just not sure where to look next on this one. <seepel>Yes the guix installer. I tried the 32 bit installer and unlike the 64 bit, grub managed to boot, but then it wouldn't boot complaining that it couldn't find bzImage. <str1ngs>the basic boot process for EFI with guix is this. grub efi creates a boot entry for guix to EFI/Guix/grubx64.efi grub load grub config probably from the store. config has location for the linux kernel and initrd. initrd does some setup loading of modules pivot roots to the new root. and then hand off to shepherd . shepherd starts services etc etc <str1ngs>seepel: maybe search to see if refind can maybe boot grub <seepel>Just started reading up an refind, if I understand correctly refind is booted by the stock EFI firmware, and then refind in turn is used to boot the OS? <str1ngs>or potentially there is a way to have refind boot guix but that alot of manual work <str1ngs>refind is a bootloader same as grub. just it's kinda geared more for macs <seepel>Ok, I'll read up and then either be back with the same question again, or a new question because it worked and I found something else I didn't understand :D <seepel>Thanks for the pointers as always! Really psyched that folks here are so friendly. <str1ngs>guix relieas on grub do to the dynamic location nature of where the kernel might be in the store. <seepel>I _think_ if I could figure out how to combine the 32 bit grub with 64 bit system it would also suffice... <str1ngs>one thing that might help is maybe install a foreign distro and see if you can workout boot process with that. they tend to be more static. that way you can resolved hardware/boot issues easier. after that you can use guix to install to a second partion and work through resolving boot issues using the bootable foreign distro <str1ngs>give this expression (origin-uri (package-source gobject-introspection )) $2 = "mirror://gnome/sources/gobject-introspection/1.60/gobject-introspection-1.60.2.tar.xz" is there a guix help function to get the mirrored url? <str1ngs>well this dirty hack did the trick I needed for now guix download $(guile -c "(use-modules (guix packages) (gnu packages glib)) (display (origin-uri (package-source gobject-introspection )))") *raghav-gururajan Yes! Yes! Yes! GNOME-CORE-SHELL is now complete. *raghav-gururajan goes next to finising up GNOME-CORE-UTILITIES. <lispmacs>I made a package definition for liquid-dsp, which installs. however, I realized that the liquid.h header file is getting placed into "include/liquid/liquid.h" rather than "include/liquid.h". How can I fix this? <brettgilio>lispmacs: this kind of stuff is usually defined in a configuration step. But is the path /actually/ incorrect? <lispmacs>brettgilio: the examples that come with the source are showing "#include <liquid.h>" <lispmacs>but I would have to adjust them to #include <liquid/liquid.h> <brettgilio>lispmacs: that may be so, but usually the linker finds those little quirks. Like, the guile c headers are all under include/guile, but you don't type that out in the source code. <brettgilio>So like, that still leaves my question open, is the path wrong? <brettgilio>The easy way to test this is to see where other distros put it <brettgilio>On Arch, it is under /usr/include/liquid/liquid.h <brettgilio>It is almost certainly that the linker just has to know how to find it. Not you specifying it in the source code <lispmacs>brettgilio: hmm, okay, well if everybody else is happy with it, than I won't worry about it <brettgilio>lispmacs: it seems to be where other distributions put it. :) *raghav-gururajan wanted to spend their day-off for good cause and managed to finish 9 patches/packages. This made their day and gonna head to sleep. <lispmacs>he mus have been up very late, referring to himself in the plural <brettgilio>lispmacs: that is the guidelines we use for gnu projects. We say per or their for non-binary representation <brettgilio>raghav-gururajan: you did lots with gnome, I managed to package SWI-Prolog :) <LarryTheCow>Hey, I've got a quick question. Is it possible to use the `--container` flag without linux-libre? <arshin>Hi guix, anyone using rk3399-based SBC (Rockpro64, Firefly)? Interested in what works and what doesn't. <leoprikler>brettgilio: no reason in particular, that's just the order in which I filled in the fields <brettgilio>leoprikler: I would. but it is not necessary, it is just odd. :) or you can wait for me or another committer to fix it, maybe sending a follow up email as a reminder. Idk. <g_bor>I have a problem when building a gexp. <g_bor>I am trying to build it using guix build -f <g_bor>Error is: Throw to key `srfi-34' with args `(#<condition &message [message: "icecat-makeicecat.patch: patch not found"] <g_bor>I don't know if this is a bug or I am missing something. <cbaines>g_bor, does that patch exist in the version of Guix you're using? <g_bor>cbaines: yes, I believe it does <cbaines>g_bor, ok, I wonder what the %patch-path is, that's a parameter in the (gnu packages) module <cbaines>I'd print it out in your script, and then compare the directories it's searching with where the patch is <g_bor>it says in guix describe it is 2e26dd6, from 3rd jan 2020. <g_bor>cbaines: ok, I will try that. <g_bor>cbaines: it is also very suspicious, that the module closure is more than 10 times as big as for gnu package... <joshuaBPMan>Does anybody have an example of configuring guix's vpn service? I'd like to try it, but I'm not certain how to configure it... <pkill9>in the first image, the spam button text is supposed to show that it can't be clicked, but it's a bit naf <pkill9>the second one, the typed text is black-on-dark-grey <g_bor>I have printed that parameter, and it does contain a directory that has the patch file. <cbaines>hmm, but you're still getting the error about the patch not being found I guess? <g_bor>actually I modified the statements to the minimal closure. <cbaines>Also, when debugging odd errors in Guile, I've found it useful to run the code in a REPL, as often you might get a different error <cbaines>So perhaps try just pasting your code in to a REPL and see what happens? <cbaines>ah, I've tried running this locally now <cbaines>I get the same error, and it's occuring within the build environment <cbaines>which could explain why the patch can't be found, as it's probably not been pulled in to the build environment <cbaines>what are you trying to do more generally g_bor ? <cbaines>maybe you'd be better of just putting the guix package in the with-extensions list? <g_bor>I am trying to create an automatic installer for guix, and I was trying to reuse some parts of the partitoning code from the installer. <g_bor>It breaks with a different error, about icecat-gnuzilla-fixes.patch. <g_bor>However the source line pointed to by the error is the same requiring makeicecat.patch <g_bor>I believe this creates some confusion somehow... <g_bor>cbaines: I don't see anything enlightening when using a repl. <g_bor>I have not done this yet, so I am surely missing something. <cbaines>g_bor, I don't think the patch exists inside the build environment <g_bor>cbaines: most probably you are right. <cbaines>I'd think that using guix in the with-extensions list would be better <g_bor>ok, i will show you what I have now, but it still does not work: <g_bor>ok, I will have a look if I can find the needed things there. <g_bor>cbaines: ok, this seems to get me some further. <g_bor>But it is really interesting that having the extension guix and the with-imported modules at the same time does not work <cbaines>with-imported-modules is just aware of the guile module dependencies <cbaines>I don't think it's aware of the patches, so it doesn't pull those in <g_bor>ok, but I still don't know why not having with-extensions guix and the with-imported-modules at the same time work. <g_bor>Now it is clear to me why only with-imported-modules does not work. <cbaines>Maybe the %patch-path doesn't include the patches directory within the guix package <cbaines>it could be that the (gnu packages) module from with-imported-modules takes precedence <g_bor>I am again on my way to make this project a reality :) <g_bor>I believe I will add something to the cookbook one this is done :) <g_bor>btw I have seen that you are working a bit around instrumenting the ci also. <g_bor>It would be really nice to have an overview as which build was schedule on which build build node, when the build started and when did it finish. <g_bor>is this somethign that cuirass knows, or this is only known by the daemon? <cbaines>but I am definately interested in getting that information available (and in to the Guix Data Service) <g_bor>this reprodubility status page is nice. <g_bor>What is the reason there is no data from other architectures? <cbaines>I've only been scraping narinfo files and builds for x86_64-linux so far, but once I've properly configured what build farms build what, then I'll be able to pull in more data <cbaines>The Cuirass patch I send earlier today will hopefully help make it easier to get the data out of Cuirass <jayspeer>Emacs magit does not see git from guix on foreign distro. any tips? <cbaines>jayspeer, that probably means it's missing from the $PATH that Emacs is using <cbaines>Do you source ~/.guix-profile/etc/profile in ~/.profile ? <jayspeer>yeah, git is visible from terminal (more so - in emacs terminal <cbaines>the terminal isn't a great way of testing, as the bash (or shell) startup could be modifying the $PATH <cbaines>jayspeer, what distro and desktop environment are you using? <jayspeer>can I source my bashrc on login? maybe that will help <cbaines>jayspeer, the environment Emacs is running in probably doesn't involve the .bashrc, but I think it might involve the .profile file <jayspeer>cbaines: I edited my .profile, let me get back to you in a minute <jayspeer>I added sourcing .bashrc to .profile and it works now :) thanks for the tip <hexagonal-sun>Does anyone know if there is a way to speed up the building of `profile.drv' after installing local packages? It takes a good few minutes on my machine <hexagonal-sun>Does anyone know if there is a package that provides `ispell'/ <janneke>hexagonal-sun: use (setq ispell-program-name "aspell") <janneke>it's weird that emacs would need this; maybe someone should work on a patch <hexagonal-sun>Yeah, not sure why it uses ispell by default if apspell is the GNU replacement. ***jonsger1 is now known as jonsger
***jonsger1 is now known as jonsger
<grillon>I have installed anthy ibus and ibus-anthy. I can select foreign keyboard but I do not see anthy in the list. in foreign distro when both are installed it works but on guix ibus-anthy seems not visible for ibus <pkill9>i don't know if electron is built from source in any distro <janneke>doesn't electron include/use chromium-browser? <janneke>does electron have npm package dependencies? <pkill9>I hope everyone has a reproducible new year ***ng0_ is now known as ng0
<dadinn>I have a question regarding mapped devices: how do I add a key-file to a LUKS device? This is something I can add in a crypttab file, but there is no such field in the mapped-device data format <dadinn>nckx: is this something in the works, or it is something which will never be added? <nckx>dadinn: Er, neither. It just needs someone motivated to add it. <nckx>*1940s recruiting poster* It could be YOU! <dadinn>nckx: is using an additional cryttab a workaround to this, while the feature is not there? <nckx>dadinn: crypttab is a Debian thing, so no, sorry. <dctrud>dadinn: there's nothing in guix that reads a crypttab AFAIK. On most distros using systemd it is converted into unit files and handled by systemd <Kimapr>every time when i try to run my executable file guix says it doesn't exist <dctrud>Kimapr: where do your executables come from? <dctrud>I mean where were they built / downloaded from? <nckx>Kimapr: We can help you if you're helpful in return. Did you install these using Guix? Something else? <Kimapr>no, i built them using another distro <nckx>Kimapr: Then they depend on that distro's /lib/ld-linux-foo.so, which doesn't exist on Guix System. <nckx>(This is Guix System, right? It has to be.) <nckx>You can provide this file with a single line in your system .scm, but TBH it's probably not worth it because you'll run into other missing library errors. Your best bet is to build the executable with Guix or to use patchelf on the one you have now. <NieDzejkob>dadinn: That's currently on my TODO, I'm hoping to have this done sometime next week (together with extra-initrd like Nix for one-passphrase boot) <nckx>NieDzejkob: Extra initrd? As in, multiple cpios? Also nice. <dctrud>Nice. one passphrase will be great on my machine with 5(!) LUKS devices. <dctrud>NB - I submitted a patch to allow LUKS2 mounts on boot. It is just allowing verison=2 in the LUKS header, since LUKS2 binary header is backward compat r.e. UUID offset. <nckx>dctrud: Ah, that's you 🙂 Yes, I just missed your patch in the first e-mail, it wasn't missing. <rekado>leoprikler: I don’t think anything needs to change with ibus. Gnome uses it internally. You don’t need to install it yourself. <rekado>leoprikler: instead of using ibus-setup and running ibus-daemon manually you are now expected to use Gnome’s Region settings to configure input methods. <rekado>it will list installed ibus input methods. <rekado>this does not work if you still have old ibus state in your home directory, though <dctrud>nckx: the subsequent patch today likely makes that doc patch irrelevant <rekado>I’m using ibus-libpinyin without problems in Gnome after removing all ibus state. <dctrud>Completely different topic - is anyone here using Guix to build containers for Singularity, to run on HPC etc. per the blog/docs? <dctrud>That's how I first came across Guix. I'm not one of the core Singularity developers, so would be interested if people use that / have any thoughts. <dctrud>ughh - I'm *now* one of the core Singularity developers... not not :-) *dctrud gets some coffee on to hopefully improve brain to hand co-ordination <rekado>dctrud: at the MDC we mainly just use Guix, but some of the sysadmins generate Singularity blobs for use on a weird GPU “appliance”. <dctrud>rekado: interesting. I think if you are invested in Guix locally it doesn't make a great deal of sense using Singularity except to distriibute an env to other non-Guix savvy places <rekado>dctrud: welcome to Guix, by the way! It’s good to see a Singularity developer hanging out here :) <dctrud>unless your cluster has a really low limit on inodes on your home... and running out of single files is a necessity... which I've heard of. <dctrud>rekado: thanks for the welcome. I'm committing to using Guix for 2020 and to stop my distro-hopping :-) <dctrud>Well, for personal stuff anyhow. Endless stuff across every disto under the sun for work. <preciouscookie>Hi Guix! My question is a bit offtop. But did anybody manage to get a modern and decent laptop that has no problems running free drivers and firmware? <preciouscookie>I know old ThinkPads are great. But almost nobody delivers such stuff in my country. <dctrud>previouscookie: Most things are okay if Intel graphics (non-discrete) and you are happy to plug in a USB WiFi adapter <rekado>preciouscookie: the Librem laptops are a bit more modern than the Thinkpads, but they are pretty expensive and you can’t run Libreboot on them (they come with coreboot). <dctrud>preciouscookie: Most things are okay if Intel graphics (non-discrete) and you are happy to plug in a USB WiFi adapter <dctrud>(assuming you are OK with BIOS / UEFI non-free) <rekado>on Thinkpads you can’t just change the WiFi card, because of their annoying whitelist. <rekado>there may be some arm laptops that work a little better with only free software, but I don’t know of any. <efraim>Is that the case with the x240/250? <dctrud>The last thing I used that worked straight away (without considerind coreboot/libreboot) was a Dell Inspiron 3000 11" from 2017, which is not that 'decent'. It came with Atheros WiFi. <preciouscookie>dctrud Thanks. Seems like working option. But if there is out of the box solution it would be much better. <preciouscookie>There is E series of ThinkPads and they uses modern AMD cards which also better than Nvidia. Does anybody has experience with such machines? <dctrud>The amdgpu driver required for any sensible performance on modern AMD GPUs won't work without firmware files that aren't provided with linux libre. <dctrud>If you are committed to stock free Guix then that's not likely the way to go. <preciouscookie>I hope one day AMD will open such firmware as well... But now yes, not the best option for Guix. Thank you very much guys! <leoprikler>rekado: I'm pretty sure you need at least ibus-{anthy,libpinyin,...} though, and I'm not quite sure whether it works without ibus itself <Kimapr>and now i can't build things with gcc <Kimapr>ld: cannot find crt1.o: No such file or directory <Kimapr>ld: cannot find crti.o: No such file or directory <Kimapr>collect2: error: ld returned 1 exit status <dctrud>Kimapr: if you have questions around compilation issues it might be better to put more context in a paste at paste.debian.net <dctrud>it's not really possible to get other than a lucky guess without seeing a bit more of the build <Kimapr>i tried several diffent context and output was same <dctrud>by context.. i mean paste the whole output of your compilation, show the commands being invoked etc. <janneke>Kimapr: after installing, it says something like hint: Consider setting the necessary environment variables by running: ...; did you do that? <Kimapr>no... does that cause all the problems? <oriansj>In this particular case, it is a Digital Ocean VM <janneke>guix pull needs at leats 2GB memory, i think <oriansj>with 20GB of Disk, 512MB of RAM and 4GB of swap <oriansj>so can't be memory exhustion unless guix is doing memory locking (doubtful) <civodul>oriansj: apparently it's the build of /gnu/store/3ii6xm9h5qm784aqr7z49xrj20idq2mw-ghostscript-9.27.drv that failed <civodul>if not, you can check the file returned by "guix build --log-file /gnu/store/3ii6xm9h5qm784aqr7z49xrj20idq2mw-ghostscript-9.27.drv" <janneke>so, does ghostscript compile with that memory setup? <Kimapr>i tried to compile simple hello world, an gcc fails <Kimapr>setting GUIX_PROFILE didn't help <oriansj>janneke: well linux treats swap like memory, so it would build just like a setup with 4.5GB of RAM (unless guix or gcc is doing funky memory locking which is unlikely) <Kimapr>i installed it | > guix install binutils <oriansj>Kimapr: what happens if you do: guix environment gcc-toolchain binutils <janneke>Kimapr: if you set GUIX_PROFILE and sourced $GUIX_PROFILE/etc/profile, then what's the value of $LIBRARY_PATH? <oriansj>export LIBRARY_PATH="${HOME}/.guix-profile/lib${LIBRARY_PATH:+:}$LIBRARY_PATH" might have been skipped <oriansj>and possibly export C_INCLUDE_PATH="${HOME}/.guix-profile/include${C_INCLUDE_PATH:+:}$C_INCLUDE_PATH" <janneke>oriansj: yes but please, no...sourcing $GUIX_PROFILE/etc/profile should take care of that <Kimapr>(guix environment gcc-toolchain binutils) did work <oriansj>Kimapr: good it means it is entirely caused by unset environmental variables <janneke>Kimapr: then something went wrong sourcing the environment settings <oriansj>Kimapr: either do sourcing like janneke suggests or adding those 2 exports to your .bashrc will do it <Kimapr>throwing "$GUIX_PROFILE/etc/profile" complains about permissions <janneke>it's printed as advice after you installed it; it's also mentioned as a comment in ~/.guix-profile/etc/profile <Kimapr>so i should put sourcing in my bashrc? <oriansj>Kimapr: if you don't want to remember to do it every time you login <oriansj>leoprikler: I have trust issues with generated files <leoprikler>In Guix System you source them via /etc/profile already, on other systems you should do the same or at least do this in ~/.{,bash_}profile <rekado>leoprikler: yes, you do need to install the actual input methods if you need them. Otherwise ibus will just support simple Xkb input methods. <leoprikler>I thought so. I also haven't got this to working by installing just the methods. <leoprikler>So I currently have ibus + methods, but I use GNOME configuration tools + GTK variable magic <NieDzejkob>you do realize that the shell that's going to execute this is a generated file? <oriansj>NieDzejkob: and I am going to bootstrap that shell from a 357byte binary that I wrote myself <leoprikler>That's fine by me, but *rc is still the wrong file for env vars, those belong to *profile <leoprikler>You'll learn why once you manage to bork `guix environment` with it. <NieDzejkob>oriansj: I am well aware. But before you get there, there isn't really any reason to single out $GUIX_PROFILE/etc/profile <oriansj>NieDzejkob: I don't single out any of it; I don't like generated config files <oriansj>leoprikler: well I do hope for that learning experience one day but I haven't yet hit it <NieDzejkob>I don't see why you are comparing this to bootstrapping <NieDzejkob>bootstrapping is reasonable under a consistent threat model, this is not <oriansj>NieDzejkob: I'm not, bootstrapping is just an extension of my personal preferences not to trust in generated files <oriansj>be they binaries or configuration files <oriansj>I use software (like guix) on my own terms; that is the advantage of free software. To use it however I see fit (and have the right for the community to call me a dumbass for doing something they think is stupid) ***apteryx_ is now known as apteryx
<lispmacs>autotools question: is there some easy built in way to have debugging flags when your run ./configure or make? Or do you ahve to program something like that into your configure.ac? <janneke>lispmacs: what kind of debugging flags? <lispmacs>oh wait, I found out I can do CFLAGS="-ggdb -O0" ./configure <mehlon>oh wait, you're the author of stage0? <oriansj>got lots of help from lots of great people <oriansj>( janneke is really killing the GCC bootstrap bit [down to about 60MB now]) <mehlon>that reminds me, wasn't there that thing "bro pages" a few years ago <oriansj>mehlon: there are also "woman pages" and "fuckit pages" and "IDK pages" <mehlon>it's like man pages but for errors <mehlon>thinking about it, a "bruh moment" would probably a pretty funny name for a kernel panic for my next kernel <oriansj>Well "suck pages" uses your query to mock your ignorance; by looking up the relevent info and then feeding it into a Markov chain along with samples of 4chan's best burns <mehlon>well querying suck pages in duckduckgo did NOT give me what I was looking for <leoprikler>I don't think asking for "4chan's best burns" is going to give you anything of better quality, though. <oriansj>mehlon: well that is a common problem with most program names. You either become so popular you don't need to be listed in search engines or so unpopular that you can't be found either... <oriansj>leoprikler: very very true but not many people out side of IT learn proper googlefu *oriansj or is it duckfu? <mehlon>on guix? umm I think noto fonts should work <alextee[m]>humm weird, i already have it installed but i can't see emoji <alextee[m]>weird, the thumbs up emoji i sent from riot on another PC doesn't show though, but i can see the smileys <alextee[m]>hmm, is there a separate guix package i should install? <leoprikler>did you refresh your font cache after installing? <alextee[m]>i guess, i've had it installed for a while, let me try <leoprikler>alextee[m]: do you have GNOME characters? If so, you can search for the thumbs up there <alextee[m]>leoprikler: that app doesnt open: JS ERROR: Error: Requiring GjsPrivate, version none: Typelib file for namespace 'Gtk', version '3.0' not found <alextee[m]>i think i have gnome installed both in the system and in my user profile maybe that's wrong <leoprikler>yep, gnome should not be installed in the user profile <leoprikler>can you copypaste my thumbs up into gedit and have it work there? <mehlon>ah but it hasn't been packaged yet <oriansj>it is tragic how many web browsers break when you turn of profile guiled optimizations <oriansj>in the case of web browsers it is the downloading and running of a list of arbitrary websites and their contents to determine which way branches should be predicted. <oriansj>leoprikler: exactly or more specifically, what ever websites they think should run fast <leoprikler>for reference, which browsers did you test? just the big ones? <oriansj>leoprikler: yep, most of the small ones have no need for such games <leoprikler>what about the engines themselves, e.g. webkitgtk? <oriansj>leoprikler: I haven't taken that much time to look at that yet <oriansj>but possibly, it is usually denoted by long compile times <leoprikler>be sure to write a paper about your findings when you finalized <oriansj>as GCC can do 5Kloc/second when just doing -O3 <mehlon>I guess we have no choice. let's switch to gopher <oriansj>mehlon: or just say optimizations past -O3 are probably a very bad idea <oriansj>Or did everyone forget the Linux Kernel vulernabilities that occured because of Compiler optimizations? <oriansj>Does anyone honestly believe that similiar things don't occur in things like web browsers? <leoprikler>I don't recall them, because I'm missing news all the time. <oriansj>especially when metrics mean users and few people have the technical desire to audit browsers relative to the number of people paid to audit linux... <leoprikler>oh, I actually remember something related to cache attacks there <janneke>oriansj: don't worry, i think it is safe to assume that something as critical for security, like a web browser, has a small, clean and well-audited code-base <leoprikler>specifically, it was (is?) possible to exploit rowhammer from Javascript using certain browser features and there was a push to disable them <leoprikler>later it was shown that those features are not necessarily needed and browser people just said "fuck it" and put the other features back in <oriansj>leoprikler: because the browser market is a 💩 show <plasma41>leoprikler: Javascript is a cancer. Make the web HyperText again. <mehlon>ah great idea... instead of using javascript we can use a blockchain <mehlon>and instead of HTML we can use... a blockchain <leoprikler>and we program everything using the blockchain programming language BCPL <oriansj>plasma41: I like to think of Javascript as the logical conclusion of an economic model, which is more concerned about making lots of idiots (and advertising companies) happy by mining them for profit <mehlon>we can also serve sites with a blockchain which runs on top of WebAssembly <mehlon>... which is implemented on top of Ethereum <leoprikler>running on top of javascript interpreted in firefox <mehlon>which is distributed with rustcoin <leoprikler>on the one hand, you no langer have to write javascript, which is a great + in and of itself <oriansj>leoprikler: using qemu running windowns 95 inside of the javascript of the original safari on OSX <mehlon>all of which is generated with a guided learning AI <mehlon>running in your ~~~brain~~~!..... on the blockchain <leoprikler>on the other hand, it is some blob from the internet that runs with full permissions <mehlon>the universe is a simulation and it's running in ethereum <oriansj>mehlon: being simulated on a 6502 in Bender's brain <oriansj>mehlon: written in APL running inside of FORTH <mehlon>ahh.. the entire universe can be written in one line of code <oriansj>mehlon: yep here is the first 50 chars of it (((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((((( <oriansj>or in the R way of expressing the solution: 💩 <mehlon>actually the universe probably isn't bootstrappable <mehlon>since you most likely need a universe to make a universe <EuAndreh[m]>Is there a quick fix, or should I bring the discussion to help-guix? <mehlon>I cant quite tell whats failing in there <mehlon>you might need to run with higher verbosity <mehlon>I just tried it and it fails because of some X.509 certification failure <EuAndreh[m]>there's no --verbose flag, how can I increase verbosity? <leoprikler>x509 certification failure sounds more like a server problem imo <brettgilio>Hey all, when I try to do (with-directory-excursion "src/Tests" ...) In a package I'm writing, it fails saying there is no such directory. The directory does exist, though. <mehlon>it seems there's no verbosity flag after all <mehlon>guix builds with containers by default so that might have to do with it <brettgilio>leoprikler: I tried with one set earlier, will try again <leoprikler>perhaps there is also a $package at one point in the tree <leoprikler>try `guix build -K` to look at the directory structure <EuAndreh[m]>I'll send the error on help-guix to further discuss this issue <leoprikler>so the first one runs through but the second does not? <brettgilio>src and packages are at the same level leoprikler <leoprikler>btw. does delete-file-recursively work with paths? <leoprikler>could you do e.g. (for-each delete-files-recursively '("packages/http" ...)) <leoprikler>what would happen if you were to rewrite your recipe in that style? <brettgilio>It is as though it *only* has an issue with that specific directory. <leoprikler>for example, I thought there was also a guile procedure for this <brettgilio>efraim: leoprikler it says "ls: cannot access 'src/Tests': No such file or directory <bandali>have you tried ‘pwd’ and making sure you’re where you expect to be? <brettgilio>Okay, so (invoke "ls" "src") returned a bunch of interesting stuff. <brettgilio>leoprikler: some of them I could. This specific one i could <nckx>brettgilio: Did you receive my PMs? (No answer expected; just curious.) <nckx>My ZNC box ran out of disc space and now it logs only lies. <nckx>'twain't logs that did it. And old ones are compressed, but the algorithm is so inefficient that even compressed files still take up space (boo). <nckx>Or if you're suggesting I delete my IRC logs from 2015: how dare you. <oriansj>nckx: ran out of budget for storage? I know that feel <mehlon>"ran out of storage on a friday night? god you're pathetic" <oriansj>mehlon: hey no shaming of Datahoarders, we have feelings you know <mehlon>it's a reference to that sonic image <oriansj>nckx: well assuming the average IRC channel volume per day at 1MB/day and you log 1024 channels and you spend $129 on a 12TB exernal, you should be good for 36 years <nckx>The real lesson is that I should stop lurking in ~80 channels for years on end. <nckx>Yes, that must be the real lesson, nothing fundamental about my storage habits. Nah. <oriansj>nckx: no, it means save $12 a month for storage and keep buying based on TB/$ <oriansj>Its not like we have a problem for anything *stares at rack* <oriansj>just need some more drives that is all... <nckx>You only have a problem when you start buying tapes. <mehlon>tried making something with GIMP <nckx>mehlon: I what er the halp context? <nckx>Deep-fried to perfection. <oriansj>nckx: no, tapes are when you finally turn the corner on your addiction and stop feeling the need for having to access all of the data all of the time <leoprikler>nckx: You can grep the cloud with network file system ;) <nckx>I have seen tapefolk and their eyes do not say ‘I have turned the corner of anything’. <nckx>leoprikler: Oh cool that sounds both fast *and* efficient thanks. <oriansj>nckx: well a stage0 talk might be nice for FOSDEM but I am unable to attend this year. Anyone want a deep dive so that they can present it on my behalf? <mehlon>nckx: the original was "alone on a friday night? god you're pathetic" <mehlon>based off some comic in a sonic magazine <mehlon>you can spread a cryptominer virus that also stores IPFS data in people's browsers <mehlon>that could give you some more storage <oriansj>mehlon: requiring others to validate one's self is usually a sure sign that the person with that perspective doesn't have much going for themselves and they are just lashing out due to their own feelings of worthlessness <mehlon>I'm guessing youre talking about the sonic image and not the IPFS thing <mehlon>rest assured it is thoroughly ironic and I am not calling anyone pathetic, just in case someone thought that <oriansj>mehlon: level 70, we need you at a 20 <mehlon>well I suppose the joke was not appreciated <oriansj>mehlon: not all jokes are funny to all people <NieDzejkob>leoprikler: "[wasm] is some blob from the internet that runs with full permissions" <- how so? <mehlon>oriansj: that example was actually very funny :D <oriansj>leoprikler: a subset of permissions actually (depending on exploits) <nckx>mehlon: ♥ the quotes around ""funny meme"". <nckx>That is exactly how I picture "funny memes". <mehlon>nckx: to be more accurate it needs three sets of quotes: ""funny" "meme"" <leoprikler>anything that you can do as a user with your browser you can pretty likely also do with Javascript in some form <oriansj>leoprikler: actually more in javascript than what the user can do in their browser <nckx>leoprikler: Isn't davfs the ‘classic’ client-server fs-over-HTTP thing? <nckx>This is going to end in rowhammer and tears people. <oriansj>leoprikler: perhaps pingfs would have been funnier <nckx>Hoist with my own picard. <nckx>oriansj: Thank you for reminding me that exists! <mehlon>nckx: you mean hoisted by your own petard? <oriansj>mehlon: when your RAM manufactor decided cheaping out on DRAM isolation made them too much money, so yay unfixable vunerabilities <leoprikler>mehlon: Rowhammer is an exploit bashed on cache weaknesses, that allows you to flip bits in memory, which could lead up to privilege escalation. <oriansj>mehlon: well Ricker certainly cleaned up with the ladies, so why can't Picard with the dudes? <leoprikler>and you can absolutely rowhammer from javascript <leoprikler>meaning you could do even more so from wasm, since you can't even efficiently scriptblock wasm <mehlon>I still don't quite know what it is but it's great I can row hammertime somehow <oriansj>leoprikler: you can even enter SMM and bypass kernel protections too <nckx>mehlon: I should have known better than to attempt "funny" "memes" of my own here, touché. <mehlon>well to be honest I am not sure what the chat is about but I'm having a great time <nckx>(Wasn't there a Patrick Stewart 'Use the force!’ photo quote with ‘Captain Kirk - Star Wars’ under it? I was obliquely referring to that, which is a recipe for failure on the Net.) <oriansj>mehlon: isn't that the point of chatting? to enjoy one's self and the company of others? <nckx>in my defence, my misremembered version is 105% funnier. <mehlon>I just upload stuff to permanentupload.com since I dont know a good image hosting service <oriansj>well the stock sysadmin version is s/cloud/arse/g <oriansj>and most of the promotional material doesn't even change much in meaning <nckx>Well, stop pushing stuff to either or you'll end up with one pissed-off clown. <mehlon>would you like to mine a bitcoin? <mehlon>thats supposed to be a reference to Saw but it sounds more like Frozen <nckx>And lo, the townpeople's greed did melt the ice. <nckx>That's one nice metaphor for our times. <oriansj>nckx: and the polar icecaps, resulting in the death of millions <oriansj>but atleast Disney made enough money to build a sequal <oriansj>mehlon: actually I think that is frozen 3 *nckx thought we'd managed to scare away mehlon for a moment. <oriansj>perhaps it'll come out before the sequal to: Home Alone Again with the Google Assistant (2018) <oriansj>They just can't stop ramming sequals down the throat of every movie that managed to turn a profit <mehlon>no I just pressed CTRL+Q while trying to press all key combinations on hexchat <mehlon>but... but... ED... is the standard text editor... <oriansj>like does Memento seriously need 3 sequals? <nckx>oriansj: Yeah see the problem with living in an actual dystopian timeline is that I still don't know if you're joking or not. <oriansj>nckx: I mean the year literally started with the Pope bitch slapping someone and a growing risk of WW3 <oriansj>thanks to bomby mick does everything he accused others of doing... <mehlon>understandably slapping someone could lead to global warfare <oriansj>mehlon: I mean the Pope has got some serious pop in that back hand *nckx checks the channel op moderation checklist. <oriansj>a totally justified response is bombing a completely unrelated third party (workd for bush didn't it?) <nckx>Religion; politics… looks good. <mehlon>well clearly he forgot to play his copy of Undertale <nckx>My partner's first words to me in 2020 were ‘all the animals are dead’ and I believed her. It's 4 days later and things have actually got worse. <nckx>Goodbye Australia, you were all right. <oriansj>something something darkside; oh shit the kangaroos are on freaking FIRE!!! <mehlon>all I can think of is "I'll show you what it feels like, now I'm on the darkside" *kmicu needed to double‑check this channel’s name after starting the backlog read. <mehlon>welcome to Guix! Guix is the standard text editor <drakonis>ehem, i have a bit of a problem here, i need to cook up my own guix install image <nckx>oriansj: After decades of absence, wolves have returned to parts Belgium (which is good, mind you). Unfortunately they have developed a taste for kangaroo flesh. Every Belgian, simultaneously: wait why are there so many kangaroos here? Turns out there are, we just hadn't noticed the trend. <nckx>What does this have to do with #guix? About as much as the past half hour, thank you. *kmicu gonna skip today and check in tomorrow. 😺 <nckx>drakonis: Have you read ‘Building the Installation Image’ in the manual? <mehlon>isn't there guix system build iso or sth *oriansj welcomes kmicu to today's insanity <nckx>drakonis, mehlon: That's exactly what you'll find there. <drakonis>i'm on nix right now and the docs are kinda <nckx>kmicu: Have you backed into a clown today? <kmicu>There’s no insanity here. The lates news is that the garlic showed up in the garden. <nckx>That's not alarming and unseasonal at all. (Maybe. I know nothing about garlic.) <drakonis>hmm, perhaps i should produce it on a vm <nckx>drakonis: You should be able to install Guix on Nix using the installation script, then ‘guix environment guix’ and bootstrap your Guix git checkout. <oriansj>also did guix pull --bootstrap 's memory requirements go up again? <drakonis>hmm, does guix container have the ability to execute inside a different root without using the guix store? <nckx>oriansj: And an OOM dmesg (or similar in the .log.bz2) to boot? <nckx>What does the .bz2 log say? <oriansj>ERROR: In procedure dynamic-pointer: Symbol not found: strverscmp <mehlon>hmm I'm trying to find psyced but it's not in Guix? <nckx>mehlon: Added or just submitted for review? <nckx>My git log search for psyced hasn't returned anything yet. <nckx>‘Pattern not found’ so no. <oriansj>oh god: (dynamic-func "strverscmp" (dynamic-link)) <mehlon>all other PSYC-related stuff is already in Guix <mehlon>no one's mentioned psyced since 2016 <nckx>mehlon: You could ask ng0 about the status of Psyced in Guix but I'm pretty sure of the answer you'll get 😉 (‘No thanks, but feel free to bang it into shape yourself.’) <raghav-gururajan>("pygobject" ,python-pygobject-3.34.0) give me unbound variable for "python-pygobject-3.34.0". <nckx>raghav-gururajan: Drop the .0, the variable is named python-pygobject-3.34 (see glib.scm). <raghav-gururajan> ("pygobject" ,python-pygobject) uses 3.28 by default. Shouldn't it use the highest version? <ng0>maybe i have it in my trees somewhere, if that was the question <nckx>ng0: It was meant lightly. <ng0>mehlon: you can also try the gentoo overlay as reference <drakonis>i suppose i can't use guix container to create a container that doesnt use the store, can i? <mehlon>I've been trying to install psyced for days now <ng0>psyced is still working <mehlon>I'm on NixOS and i've tried running gentoo in docker, or just installing it from the tarball <ng0>mostly finished, which is why there are no new releases, or not that often <mehlon>the former took too long to compile git and eselect-repository so I gave up <ng0>as psyc2 has all the attention <mehlon>should I be trying to run psyc2 with secushare instead? <ng0>no, but for the long form answer with a reason try #dev in psyced <ng0>i'm in no good position to type right now :) <ng0>2016.. my unfinished work comes back at me all the time.. :D <mehlon>but for now I can just add your scm file and compile it in guix right? <ng0>p2 is a finished prototype of a project, and ports + ports-wip can theoretically be used with guix <ng0>i haven't used this in a year or more <ng0>sorry, meant "second prototype of a guix deriviate which then moved on to not be a guix deriviate anymore for many reasons" <nckx>raghav-gururajan: At the Scheme level, no, because python-pygobject is just a Scheme variable bound to the older version in glib.scm. Guix doesn't do *any* version comparisons or ‘dependency resolution’. As to why python-pygobject points to an older version: that's unconventional, but probably happened to minimise the number of rebuilds (see ‘guix refresh -l python-pygobject@3.28’ → staging material). BRB… <raghav-gururajan>I am getting "meson.build:49:0: ERROR: Dependency "grilo-plugins-0.3" not found, tried pkgconfig". <str1ngs>raghav-gururajan: maybe grilo-plugins need grilo? <str1ngs>though grilo is a dependency of plugins . possibly the pkgconfig search path is not propergated? <raghav-gururajan>str1ngs I see. Would you be able to provide me with an argument to fix it? <str1ngs>raghav-gururajan: find $(guix build grilo-plugins) has no pc files <str1ngs>maybe investigate why it's not installing the pc files <str1ngs>raghav-gururajan: you can test with something like this. guix environment --ad-hoc libxml2 grilo-plugins grilo pkg-config -- pkg-config --exists grilo-0.3 && echo OK || echo FAIL <str1ngs>take my example with a grain of salt. might be best to enter the build after -K ***daviid is now known as Guest44127
***Guest44127 is now known as daviid
