<lispmacs>does anybody here happen to use a USB-UART converted which works with Guix out of the box? <fnstudio>hi :) i'm progressing in my journey of gradually migrating my system to guix, one bit at the time; i now seem to have encountered a problem with i3lock <fnstudio>which i think it's a known (configuration) problem <lispmacs>joshuaBP`: USB adapter for connecting to the serial RX/TX pins on an arduino or similar board <fnstudio>basically, i3lock won't accept my password <lispmacs>joshuaBP`: I think maybe it is usually called USB-TTL <fnstudio>i suppose this has to do with i3lock having to interact with pam <fnstudio>anyone knows how this should be dealt with when using guix on top of a host system? <lispmacs>joshuaBP`: most arduino boards of course already have that built into the board, but some other boards I have also have UART pins <lispmacs>for screaming 9600 bytes/sec data transfer speed! <joshuaBP`>I just got my linode running guix. That's pretty awesome. <nefix>what's the difference between inputs and native-inputs? <PotentialUser-72>Hi everyone. Is there a workaround the wifi freezing up at installation? There was no option to do wired or wireless before that. <PotentialUser-72>I also tried the previous version of the installer and it would not connect to the wifi as well eventhough my laptop's wifi works with Trisquel 8 <leoprikler>nefix: native-inputs is for stuff, that doesn't run on the target machine <joshuaBP`>PotentialUser-72: hmmm. I've not encountered that issue...would trying the manual installation help? <nefix>leoprikler: and propagated-inputs? <joshuaBP`>also does anyone have an example nginx server via a let's encrypt certificate? <leoprikler>get "propagated" into your profile/environment if the package is built for one <nefix>I see. So native are build only, inputs are runtime and propagated are in your profile, right? <leoprikler>e.g. if you install foo and it propagates bar, you now have a reference to bar in your profile as you would with traditional package managers <leoprikler>yes, but we don't use "build only", because guix is actually smarter than you in figuring out what persists beyond build ;) <joshuaBP`>PotentialUser-72: hmmm. Can you try switching to a different virtual terminal at that point? <joshuaBP`>When I was first getting started with guix, I had to install manually. I installed a bare bones set up at first. <joshuaBP`>Then I reconfigured it to be more graphical. <PotentialUser-72>hmmm installing manually would be great but I would like to go with graphical if possible :-) <joshuaBP`>PotentialUser-72: For me, I had to do manual installation first. I couldn't get the graphical install to work for me on my macbook. <joshuaBP`>PotentialUser-72: did you get any error messages when you switched to those virtual terminals? <PotentialUser-72>I'm not really great with manual installations therefore the graphical one <joshuaBP`>PotentialUser-72: I gotcha. The graphical installer is just sometimes not the best. Guix devs try to make it so, but it does still have issues. one option might be for you to try the manual installer. it's really not that hard. <PotentialUser-72>Maybe I'll just wait for the next version of the installer :-) Thanks for the discussion guys. ***catonano_ is now known as catonano
<pkill9>network-manager works better for me than connman, lol <str1ngs> I've never installed guix from the installer. I've always just used guix system init :) <feloadr>is there a way to have guix build the grub config but not run grub install (loading it via configfile directive from other distro's grub installed to mbr)? <brettgilio>feloadr: I believe guix system build will build it <feloadr>yes, but the guix system recongifure command fails out when it tries to run grub-install (apparently it doesn't like to install directly to a btrfs partition) <feloadr>so i'm wondering if there's a flag in my bootloader-configuration or something to get it to just not run the install part <mroh>make a (bootloader (bootloader-configuration (target "doesnt exists"))) config entry, so grub would fail (but reconfigure still works). <feloadr>mroh, do i need to leave the (bootloader grub-bootloader ...) part in there to get it to write the grub.cfg? <mroh>grub fails, but still generates a grub.cfg <mroh>ah, yes, that need to be in there <apteryx>feloadr: there's guix system --no-grub <apteryx>feloadr: I guess the no-grub wouldn't build the grub config, so nevermind <feloadr>ah ok, yeah i figure it's best to have each distro manage its own grub menus <feloadr>mroh, so i see a grub failure still and guix system reconfigure doesn't exit success (returns 1).. am i missing something still or are you saying i should just expect an error and everything else is probably fine? <mroh>feloadr: yes to both: expect an error, but everything is fine ;) <mroh>hmm, maybe we could add something like: "dont throw an error, if target is #f" or so... <Kimapr[m]>anyone knows how to use emacs-matrix-client properly? i'm having hard time trying to use it <reepca>IIRC you can use M-v to scroll up and it'll trigger loading older stuff <Kimapr[m]>but why doesn't it load stuff while i browse with cursor? <str1ngs>Kimapr[m]: because if it were to load then you would lose your place. there could though be a feature in emacs-matrix-client to follow on message or something. <reepca>So I'm trying to build ungoogled-chromium and currently ld is using 11.3 GB of resident memory, 19.4 GB of virtual memory, and my system is crawling. Who on earth designed it to require so much memory? <str1ngs>you could potentially lower the cores. but 4 is not that high. could be how chromium does linking. <str1ngs>guix weather ungoogled-chromium shows 0% so maybe it's an issue with the build. <reepca>it's now succeeded for me, it was stuck at 100% for quite some time <str1ngs>that last linking phase takes sometime. and I create the qtwebengine package which is based on chromium that was one of the more intense parts. <reepca>was it thrashing (constantly swapping) on your system too during that phase? Or is it just always take a long time? <reepca>also is there a way to use password authentication with 'guix copy'? I get "guix copy:error: SSH authentication failed for <my-ip-address>: Access denied for 'publickey'. Authentication that can continue: publickey,password <Kimapr[m]>str1ngs: can't it load the messages and then move the cursor back to the line it was on before the loading? <terpri>another example of a free program distributed as appimage: koreader for desktop gnu/linux <terpri>i'm going to try packaging flatpak-builder (useful anyway) and then see if simply wrapping appimages in flatpak containers "works" <guix-vits>terpri: why not `guix pack`? There were slides somewhere.. like "it's better, because.." did not remember well. <terpri>guix-vits, this is for running unpackaged programs, rather than distributing guix-packaged ones <terpri>it's miserably difficult to get an appimage running under guix (as an appimage is a statically-linked binary that uses FUSE to mount a filesystem stuffed inside the binary, but expects the contents to be run on an FHS system with "common libraries" preinstalled"...) <terpri>better to have proper packages of course, but zulip and koreader are two examples of free programs that are packaged with appimage <terpri>...and don't have guix packages yet, probably more free programs i don't know of and some non-free programs as well <terpri>presently the extension relies on google for speech-to-text, but should use the (free) mozilla deepspeech engine for local STT at some point <terpri>i'd like to see a TTS extension based on mozilla common voice too, the lack of TTS extensions has been a hurdle for getting some friends to switch from chrome to firefox/icecat <terpri>(i assume the chrome extensions just send everything to google...) <terpri>note that mozilla common voice subprojects are both free and do *local processing* for TTS and STT, iiuc <terpri>and i renamed riot -> element (new name is not great imho, but whatever) <terpri>for that matter, i should add all the packages bitrotting in my ~/guix folder <terpri>it's pretty bad, just because it's sooo generic <terpri>and doesn't quite roll off the tongue in conversation (vs. "signal me about that", "hit me up on riot", etc.) <terpri>but if the name is unpopular enough, maybe a client with a cooler name will become dominant and we won't have another electron app to maintain :P *guix-vits "GlugGlug is a vacant name now? Hmm.." <terpri>there are so many cooler matrix-related math words to choose from! eigen{value,vector}, tensor, manifold, names of mathematicians... <pkill9>why is it better to use icecat instead of chromium? <terpri>pkill9, if you're referring to my earlier comments, i'm attempting to get people to switch simply from chrome to firefox, for obvious reasons <terpri>probably no freedom-related reason to prefer icecat to ungoogled-chromium, though i have personal reasons to prefer firefox (check the changelogs ;)) and mozilla is much more aligned with free-software values than google <guix-vits>terpri: Did they released the Pocket sources? *guix-vits i'll be quiet, promise. <terpri>guix-vits, not yet, i was just looking into that the other day as a friend is setting up wallabag on hcoop.net <terpri>"We’re still a small team and we do not have the bandwidth to spend full time effort on open sourcing our historical server code, but as we’ve developed new services we’ve been working on open sourcing. With the code we are open sourcing it’s more complicated than just adding a license and making the Github repo public. We need to conduct security audits, legal/IP audits, we need to review 10 years <terpri>We had plans coming into 2020 to do a bit more, but our progress this year has been slowed due to COVID-19 and the need to adjust our priorities." <terpri>they're releasing bits and pieces, but very slowly <kmicu>pkill9: it depends on the definition of ‘better’ but using Chrome (Blink) derivatives puts more power in hands of Google to shape the web. <kmicu>(which they do and that was the intention for providing a browser ‘for free’) <terpri>my *guess* is that they honestly intend to release it as free software, but are probably a very small team spending most of their time just keeping the centralized service going for 250 million+ firefox users <terpri>(which suggests that mozilla is simply not allocating enough resources to the pocket service) <terpri>(that's my honest guess based on being somewhat familiar with mozilla and also with shoestring-budget hosting services, e.g. i've volunteered for hcoop.net for more than a decade in both technical and administrative capacities -- it's a coop of ~100 members essentially run by three volunteer sysadmins...who are generally too busy with day-to-day tasks to take on large projects. fortunately all of *our* software has <terpri>been free from the beginning :)) <terpri>(although even there we've made minor mistakes despite being 100% freedom-focused from the start, e.g. i forgot to put a license on the emacs mode for our service configuration DSL, but it's higher priority for me to package our custom tools for guix than to email the other contributor to that unintentionally-nonfree program...) <terpri>busy with both day-to-day administrative tasks for the coop and full-time jobs as professional sysadmins, i should say (hi jackhill :)) <terpri>by the way, have i mentioned that i packaged smlnj for guix, which was one of the *very first* guix packages i wrote, and not exactly trivial? :P WIP (possibly bitrotted): https://paste.debian.net/1159728/ <terpri>mlton as well, which relies on smlnj for bootstrapping <terpri>i relied heavily on the debian build scripts, but had to figure out some "fun" bits like the exec.c patch on my own <terpri>NieDzejkob, i'd have to double-check, i don't think it's 100% bootstrapped from C (or if it is, some of the C is auto-generated) <guix-vits>BTW, what is preferred: make patches, or use (substitute*? In the cases like the above package. <guix-vits>I see that on line 96 substitute is easy to read, as the same change is performed on multiple files. I withdraw my question :) <terpri>a matter of taste, really. in the guix repo, patches appear to be used mostly for big changes, or for diffs takes from elsewhere (e.g. security bugfixes) <NieDzejkob>terpri: it seems that there's also a WIP smlnj on the mailinglist, you might want to send your work there, since it seems that one had some issues that still had to be resolved? <terpri>ah, right, it's a small C runtime, and most of the system is written in SML...some of which is precompiled with SML/NJ itself (along with the source, but still precompiled) <terpri>to bootstrap from plain C, one would have to port the SML/NJ-specific Compilation Manager system to another SML implementation, i think <terpri>as well as whatever extensions the bootstrap compiler requires <terpri>NieDzejkob, sure, i'll take a look <terpri>(n.b. that smlnj.scm i pasted is more than two years old...) <terpri>another important ml implemntation is mlton, which does whole-program optimization. but in practice it can be bootstrapped either by itself...or sml/nj <terpri>(polyml -- a very simple, C-based sml -- is partially supported for bootstrapping, but the build was broken last time i checked, again due to non-standardized build systems) <nefix>hello again! Is it possible to run services inside a guix environment? e.g. have an isolated environment (even a container!) running postgres or libvirt <rekado>janneke: I don’t know the current status of the childhurds. I took some time off to sleep, so I’m not up-to-date on everything. <nefix>guix-vits: what I mean is: I run guix environment -m manifest.scm and the services are already started <nefix>(repeating since you seemed to be disconnected) guix-vits: what I mean is: I run guix environment -m manifest.scm and the services are already started <guix-vits>nefix: should be possible. IDK, but: make an shell script, and try to execute it with `guix environment ... -- ./script.sh`. And script will invoke the needed commands. <janneke>rekado: good, i'm also still slumbering, i'll let it rest until "someone" who's also sleeping wakes up again ;-) <nefix>what would be nice is to have herd inside the container, or something like that <nefix>to be able to run the same services that are defined by Guix, and not using scripts <leoprikler>all guix commands are also exported as functions, so if you need any of that, just pull it in <nefix>leoprikler: I've tried o do so, but I haven't succeeded <nefix>I've successfully imported herd and started it, but I can't figure out what's actually the libvirt service to run inside '(start <herd service>)' <nefix>Also, starting herd seems to block the "start" of the shell <leoprikler>For the libvirt service, I think there is one implemented for the OS config <leoprikler>you obviously can't use that on a foreign distro, but you might want to have it as a reference <nefix>also, why wouldn't it run in a foregin distro? <leoprikler>guix services are not directly shepherd services <nefix>oh, that's what I'm missing! <nefix>So, what are the differences then? <leoprikler>guix services *instantiate* shepherd services (if they extend them) <leoprikler>so you have to take a look at the actual shepherd code that in /path/to/guix/gnu/services/virtualization.scm <nefix>ok, let's see if I can manage to do it! :D <fnstudio>hi all, once `guix pull` terminates, i'm served this message that recommends setting `GUIX_PROFILE="${HOME}/.config/guix/current"` (instead of the existing `GUIX_PROFILE="${HOME}/.guix-profile"`) <fnstudio>however, if i do so that completely breaks my system, well that's a tad dramatic... i need to revert it back as no binary is found <fnstudio>leoprikler: `guix pull --list-generations` gives me 2 results <leoprikler>okay, did you modify any of your ~/.{bash_,z,}profile or ~/.*rc as a result? <fnstudio>leoprikler: well yes, in order for guix to work, as part of the setup / early configuration <leoprikler>revert whatever change you made and spawn a new shell <terpri>NieDzejkob, that may well be true, it's been a long time since i looked at polyml in any detail <terpri>basically all i remember is that it was the only popular SML i knew of that could be easily bootstrapped (to an extent) within guix <terpri>and has more c/c++ runtime stuff (in terms of loc) than actual sml code if cloc is to be believed <fnstudio>leoprikler: sorry for the delay; this is the PATH after commenting out any guix-related change in my `.bash_profile`: `/usr/local/bin:/usr/bin:/bin:/usr/local/games:/usr/games` <fnstudio>leoprikler: yeah, sorry, i should have mentioned it! <leoprikler>you now want to update your .bashrc with the hint, first setting GUIX_PROFILE to ~/.config/guix, then to ~/.guix-profile <fnstudio>hrm, sorry, i'm a bit lost; i suppose there's no point in simply setting the env var twice, i suppose that means: 0) set the var to .config/guix... 1) source it 2) set it to .guix-profile... 3) source it again? <leoprikler>source $GUIX_PROFILE/etc/profile on 1) and 3), but yes <fnstudio>leoprikler: and is it `.bashrc` or rather `.bash_profile` by the way <fnstudio>leoprikler: np, thanks so much, brb hopefully with good news <fnstudio>leoprikler: yay, that seems to have fixed it; i killed X, logged out and logged in again, and everything seems to be as expected in terms of my apps - all guix apps are visible <fnstudio>is there an ultimate test to see whether it's picking up the right generation? <fnstudio>`which guix` gives me `/home/user/.config/guix/current/bin/guix` <leoprikler>you don't want to install guix in your user profile <fnstudio>hm... does that mean i should `guix uninstall guix` <fnstudio>i still get some warning/hint when launching `guix pull --list-generations` with regard to `guile: warning: failed to install locale; hint: Consider installing the `glibc-utf8-locales' or `glibc-locales' ...` <fnstudio>but i guess i can look it up on the internet <fnstudio>first, and then come back if i couldn't fix it by myself <fnstudio>right before the other vars in `.bash_profile` <fnstudio>(actually, before or after doesn't change things) <str1ngs>fnstudio do you install guix with guix-install.sh <fnstudio>str1ngs: oh well, that was a couple of weeks ago already, but i think i followed the official tutorial pretty closely <fnstudio>str1ngs: i can definitely look that up in my history though <str1ngs>fnstudio: so you followed the binary install and you did'nt use guile-install.sh? <rekahsoft>What is the expected behavior when a profile has packages whose store paths have conflicts? Eg. a profile that contains 'python-wrapper' and 'python2'? <fnstudio>str1ngs: sorry for the delay, i wanted to double-check that in my history; yep, i did use the binary as explained in the binary installation guide online <rekahsoft>When I run 'guix environment --ad-hoc python-wrapper python2' it appears that what python I get depends on the order I pass the list of packages. I would more expect an error or at least a warning? <str1ngs>fnstudio: in your profile check you have glibc-locales installed. with guix package -l <str1ngs>rekado: it normally will mention if there is a conflict <fnstudio>str1ngs: yes, i do - both glibc-locales and glibc-utf8-locales as a matter of fact <str1ngs>fnstudio: okay you only need glibc-locales. now get the guix commit with guix describe. then switch to root with sudo su - . and do guix pull --commit=<hash> . this assumes you have not pulled as root. check as root with guix describe just encase. <str1ngs>fnstudio: also make sure GUIX_LOCPATH is set with env | grep GUIzX <str1ngs>fnstudio: check GUIX_LOCPATH first. I assumed it was set already. <leoprikler>rekahsoft: there is actually a reason for this, as the files affected usually tend to be caches <leoprikler>there is a conflict resolver, whose inner workings you see if you turn up verbosity high enough <str1ngs>fnstudio: let me know if all that is okay. then I'll explain how to install glibc-locales to the root's guix profile. <fnstudio>str1ngs: cool, working on it right now, thanks; when you mention the env var, is it when i do the thing as root or when i move back to user? <str1ngs>fnstudio: do it as your user. for the root profile should not matter. the guix-daemon.service will set GUIX_LOCPATH <str1ngs>we'll follow up on that once you have pulled as root with the same commit as your user. <fnstudio>str1ngs: ok, so that's now done, i mean the `guix pull --commit ...` that i launched as root <str1ngs>fnstudio: and as root does guix describe look right? <fnstudio>str1ngs: well no, it doesn't look very good; it says `guix describe: error: failed to determine origin` <str1ngs>fnstudio: okay use ~/.config/guix/current/bin/guix describe <fnstudio>str1ngs: (sorry, i suppose that's `~` as in `/root`, not as in `/home/user`?) <str1ngs>you could do ~root/.config/guix/current/bin/guix describe <str1ngs>though it might be easier in root's .bashrc to add ~/.config/guix/current/bin to beginning PATH. I'm just making sure you are using the right guix. <fnstudio>str1ngs: ok, so, that does list a generation (i suppose this means that guix does see glibc-locales as installed?); but i still get the warning `guile: warning: failed to install locale` <fnstudio>to be clear: `~/.config/guix/current/bin/guix describe glibc-locales` lists a matching generation as well as giving me the warning <str1ngs>fnstudio: can you paste the output of guix describe as root? <fnstudio>str1ngs: sure thing (any preferred / official paste service here?) <fnstudio>str1ngs: right, sorry, that's noted for next time <str1ngs>fnstudio: it's just guix describe. it does not take a package name. but that's okay. <str1ngs>fnstudio: now as root do ~/.config/guix/current/bin/guix install -p ~/.config/guix/current glibc-locales. after that run guix describe and it should glibc-locales. <str1ngs>fnstudio: also add ~/.config/guix/current/bin to front of PATH in .bashrc for root. saves extra typing :P <str1ngs>depends what you prefer. guix-install.sh uses /etc/profile.d/guix.sh which is better. <str1ngs>but you can use .bashrc or .bash_profile <str1ngs>fnstudio: did you install glibc-locales to root's guix profile? you can check with guix describe <fnstudio>str1ngs: yep `guix describe` now gives `glibc-locales 2.31` <str1ngs>fnstudio: okay that's fine. now as your user restart the guix-daemon. with sudo systemctl restart guix-daemon. this assumes you use systemd <str1ngs>does guix still complain about locales? <str1ngs>what does localectl list-locales output? <str1ngs>you can paste in here if it's just a couple of line. yes as user <fnstudio>C.UTF-8 and en_GB.UTF-8 (on different lines) <str1ngs>and localectl status the system locale line? <NieDzejkob>BTW, C.UTF-8 is a debian invention and does not exist on Guix <str1ngs>there is such a thing as C locales. meaning none I think? <str1ngs>generally C.UTF-8 is for a machine not a human. so useful in a a build environment. and when AI becomes sentient I guess. <str1ngs>fnstudio: okay looks good. now normally I don't recommend this. but I find this will probably resolve the locales now. if you reboot it should be okay now. <fnstudio>happy to refactor my locale settings if that helps btw <str1ngs>nope all looks good. but I would reboot at this point <fnstudio>str1ngs: so it was all a matter of turning it off and on again? (lol, i'm kidding! doing that now, brb) <str1ngs>I'm not sure why rebooting helps here but it does. strange as it seems <str1ngs>most of this stuff relates to foreign distro btw <fnstudio>listen... i'm in a bit of a transition phase, with a long term goal of trying and switching to guix system... it's not the end of the world if i get that warning <fnstudio>you've been extremely helpful and kind already <str1ngs>okay try this get the guix-daemon PID using. ps aux | grep 'guix-daemon' . should be the second column. then paste the output of sudo cat /proc/$PID/environ | tr '\0' '\n' <str1ngs>fnstudio: seems okay. and echo $GUIX_LOCPATH is still set? <euandreh>Does "guix deploy" envision superseding Terraform? <euandreh>like, does it want to implement things like creating infrastructure in a provider's API, managing IPs, DNS, etc? <fnstudio>str1ngs: yes, still set (sorry for the delay) <NieDzejkob>euandreh: I recall some digital ocean-specific facilities for creating the server itself, but I only heard of it in passing... <fnstudio>str1ngs: i might have some connectivity issues in the next hour, if i'm not responsive i'll try and get in touch again later/tomorrow <str1ngs>fnstudio: no worries. I'm not sure why the locales are not working for you. normally once I check everything we did it should work. on guix system this is less problematic. seems to happen more on foreign distros <euandreh>NieDzejkob: yes, I saw even some discussions in the mailing list about it, and that caused me to wonder <euandreh>kmicu: Is there a discussion about it somewhere? <kmicu>euandreh: no discussions about that yet because deploy is fresh. <euandreh>dftxbs3e: I mean, there's no mention of any tool on the manual, and the few commits I did were handwritten xD <dftxbs3e>euandreh, I was wondering if yasnippet could do this <dftxbs3e>Also, I don't see the point in describing the diff in the commit message <euandreh>dftxbs3e: the manual points to the GNU standard on commit messages, which has some reasoning behind it <dftxbs3e>I would think the commit message rather be used to explain why the diff was made <euandreh>I kind of agree, but I didn't contribute or searched the logs enough to have any strong opinions on it <dftxbs3e>I have read the GNU guidelines for commit messages <joshuaBPMan>guix deploy: error: failed to deploy locke-lamora: no matching pattern <nckx>dftxbs3e: Manually, but it's not a real Guix commit or commit message. See the branch. <Kimapr[m]>Is that okay that flatpak tries to mkdir(/var/lib/flatpak) when installing apps? That makes it require root <joshuaBPMan>Hey guys, I'm having some weird issues when running the shepherd...I'm getting an error like this: <joshuaBPMan>;;; In procedure load-thunk-from-memory: incompatible bytecode version <joshuaBPMan>I think I had guile 2.2.7 installed a moment ago, and now I'm running guile 3.0, which seems to make lots of issues. <joshuaBPMan>Just running shepherd as a regular user and I get a ton of thos incompatible bytecode errors. <joshuaBPMan>user_oreloznog: I'm not super familiar with flatpak...but if it makes it require root, then probably not...? <efraim>joshuaBPMan: I have guile-3.0.4 installed, it's been like that for more than a month <joshuaBPMan>I'm going to reconfigure and see if that fixes my errors. <efraim>Kimapr[m]: you can try using flatpaks with the --user flag <joshuaBPMan>efraim: also thanks for you guix video that you posted. <efraim>joshuaBPMan: I think there is some incompatability between the bytecode with 3.0.2 and 3.0.4 <efraim>I'm glad you liked it :) I forgot after a while I was still recording and thought I was just talking to myself <joshuaBPMan>efraim: also, I guess I'm using guile 3.0.2. How do you have 3.0.4 installed? <efraim>I added 'guile' to my manifest, but IIRC guix has 3.0.2 embedded somewhere <leoprikler>I'm kinda dumb rn, how do I match the start of a filename in #:include-regexp <leoprikler>I'm trying "^pattern.*(...)", but that doesn't work. <leoprikler>Meanwhile having it without the leading ^ includes stuff, that has a different prefix <joshuaBPMan>user_oreloznog: the nix-service may already have it packaged. <user_oreloznog>Hmm, i don't understand, I think you want ask to another person ;) <joshuaBPMan>wow, so I can't even start the shepherd as a regular user. <nckx>I'm running Wayofthefutureland and all my X apps get the throwback ✖ cursor. Is that expected? <nckx>Only for the default; the Black Hand Of Selection shows up fine when I mouse over some things. <nckx>It's very... hm. ‘Modern.’ <joshuaBPMan>nckx: oh. I thought it was a specific cool something...like wayfire... <nckx>The X app in question is this very HexChat (waves). <joshuaBPMan>nckx: I do have that weird X thing in qterminal too. <nckx>It's perfectly usable but I must be missing something somewhere. <joshuaBPMan>not in firefox though...I'm pretty sure firefox is using X. *nckx fires up the IceCat. <nckx>Indeed, you're right, I get the Black Arrow of Sanity. <nckx>Although interestingly(?), the hand cursor is different in HexChat (when hovering over a nick) and IceCat (when hovering over a link). <nckx>HexChat's is black and even more oldskool than IceCat's white one. <nckx>I'd make a screenshot but I'm sure that would be yet another quest under Wayland. <nckx>I'm a bit surprised my old gnome-screenshot binding didn't work. Maybe it needs enablement. <guix-vits>nckx: the binding is in ~/.config/sway/config? <nckx>I just copied my old i3/config. Print Screen is still bound to gnome-screenshot. It launches but is unable to capture any pixels, which I find surprising. <guix-vits>Ah. Honestly i did had some troubles with the default i3wm config. <guix-vits>but the gnome uses Wayland by default, in another hand. <kmicu>Whoa, nckx embraces the future? So modern, so secure. 😹 <nckx>So janky. Much jank. Very jank. <nckx>guix-vits: Yeah, it does (I've been told), so we must be building/configuring/wrapping gnome-screenshot poorly. <nckx>kmicu: If the future means that rotating my screen 90o (oh, right, my compose key is broken too) makes windows fall off the edge of the screen: boy am I living in the future. *kmicu now expects a quality (service wayland-makeitso-type) <nckx>I mean, I *know* X is a stinking pile of rotting hacks and skeletons knee-high behind every closet, but at least the bones have had time to settle and you can sit quite comfortably on top of them. <kmicu>Compose key must work so it’s a misconfiguration of an early adopter. You will fix those because you are our Wayland pathfinder. <nckx>Wayland, eh, not so much. <kmicu>Dear Early Adopter, just fix the bugs for me, m’kay. Sincerly, Late Majority. <nckx>kmicu: Meh ☹ I need xmodmap (because my compose key is my Fn key) and the first answer I found was ‘no xmodmap for (s)way(land), what you actually want is to create your own custom keyboard layout file and load that’. <nckx>Well, no, I don't, actually, want that. <nckx>I want a one-line to plonk in a dotfile. Kthxxx. <kmicu>Anything what is possible with xmodmap is possible with xbk after a decade of reading its source code and documentation. <nckx>I'm sure XKB is Turing-complete so sure. <nckx>And I'm just ranting because everything has changed and my decades of X wart knowledge is now worthless 🙂 I'm actually quite enjoying diving into this stuff. <kmicu>So many unique X bugs triggered by non‑parenting XMonad. I will miss that. *kmicu rechecks whether xinput devs finished that ‘friendly’ xkb layer. <nckx>kmicu: Ooh, is there a Wayland xmonad yet? <nckx>Fun story: I switched from xmonad to i3 two years ago because ‘Wayland was coming and I didn't want to be forced to transition on their time’. <nckx>Used X for two more years so that decision did not age well. <butterypancake>is anyone having trouble lately? I did a sudo guix system reconfigure just the other day and then my computer wouldn't boot. It forced me to reinstall the OS. In hindsight I think the only reason my system rollback didn't work was because I didn't mount my boot partition correctly in a recovery environment :P <nckx>butterypancake: ‘...computer wouldn't boot...had to reinstall the OS‘: ‘having trouble’. Are you by any chance English? But no, no such trouble, and I pull/reconfigure several times a week. <dftxbs3e>nckx, is it possible to skip reproducible builds for powerpc64le-linux-gnu at first to at least get some more contributors and users on board? <dftxbs3e>there's a non-trivial issue with that on big endian, and I suspect it'll happen on little endian too <dftxbs3e>I'm exhausted working on this without much engagement from the wider community <dftxbs3e>it's an enormous task for someone so remote from Emacs, GNU Guile or Scheme like me <nckx>I'm sorry to hear that. I'm not sure which reproducibility requirement or discussion you're talking about. I wasn't even aware that all other bootstrap binaries were, in fact, reproducible. <nckx>dftxbs3e: Is there anywhere I can read more about the current state & problem? (Ideally not ‘the past few months of IRC logs’... :-/ ) <butterypancake>nckx: I'll be honest, I'm not sure what your comment about being English means... are those colloquialisms? I'm an English Canadian who doesn't interact with non-Canadians very often :P <nckx>You have a knack for understatement, was all. <butterypancake>oh I see :P I mean I have a knack for getting myself into digital trouble so it was an average saturday for me <dftxbs3e>nckx, there was emails on guix-devel but I got little endian to compile since then, so we should be able to make use of that OSUOSL VM.. just needs someone with knowledge of the CI system to actually make that work on ppc64le <kmicu>butterypancake: it would be helpful to see the diff between configs used for reconfiguration. Otherwise it could be anything. <butterypancake>I tried creating a boot media from the latest commit but that wouldn't boot either. I had to use the 1.1.0 boot media <dftxbs3e>janneke did it for GNU Hurd, maybe they can help a bit <butterypancake>the reconfigure was identical to my last reconfigure. I just wanted the newer kernel :P <dftxbs3e>nckx, search for "Helping with powerpc64-linux" in guix-devel archives <nckx>blackbeard[m]: Yes in that I'm @nckx@anticapitalist.party but no in that I don't produce content at all. <dftxbs3e>nckx, I need help upstreaming the changes badly <kmicu>butterypancake: I think then rollback should work and let you boot into previous working kernel. 🤷 <dftxbs3e>I'm not sure to totally get the commit message guidelines, they're vague and I couldnt find any program to actually test if a commit message *is* compliant <nckx>So is my understanding correct that the bootstrap binaries you've worked on can't be built reproducibly, and won't be merged until then? <dftxbs3e>It would be useful to get commit access to push changes on a branch <butterypancake>kmicu: ya, except I foolishly mounted my boot partition to /boot instead of /boot/efi. I only realized that after the roolback didn't work so I started reinstalling the OS. So I formated the drive, then read the part about mounting to /boot/efi :P you live you learn I guess <dftxbs3e>so it's more visible to the wider GNU Guix community <joshuaBP`>dftxbs3e: just 'cause I'm curious...what's the difficulty in porting to power? <dftxbs3e>nckx, marusich thought it was better to get reproducible binaries but I assumed that was a requirement <dftxbs3e>It indeed makes sense for it to be a requirement, otherwise all the reproducibility of GNU Guix is moot <butterypancake>what still runs powerpc processors? I thought they where outdated. The only machine I know that uses them is the Wii. <joshuaBP`>butterypancake: raptor computing is selling power pc machines. <dftxbs3e>joshuaBP`, GNU Guix being historically very x86 focused so inherits many x86-isms <nckx>butterypancake: POWER9 is a thing. 🙂 <kmicu>butterypancake: was that filesystem layout change between reconfigurations? <joshuaBP`>They are expensive, but they are the only recent machines that one can buy that are potentially endorsed by the FSF. <butterypancake>kmicu: no, I couldn't boot at all so I booted from a guix bootable media, mounted my drives (wrong), then did a chroot, and a rollback <joshuaBP`>dftxbs3e: oh! I didn't know they were certified yet! Cool! <kmicu>butterypancake: you were not able to boot from the boot menu into previous working kernel (guix system revision)? <butterypancake>So what's so great about powerpc or power9 or whatever? Is it just that we don't have an intel management engine? The cpu's aren't open though right? <dftxbs3e>joshuaBP`, but porting anything that big to a new platform is always an adventure, GNU Guix or else <joshuaBP`>butterypancake: it is VERY open artetechure (sorry for that misspelling) <joshuaBP`>butterypancake: and it's potentially as fast or faster than x86. <dftxbs3e>butterypancake, the whole system boots without any proprietary blobs <butterypancake>so the firmware is open so it's better than intel/amd but the cpu isn't open so it's worse than risc-v? Is that a fair statement? <joshuaBP`>dftxbs3e: hmmm. I figured it would be, but it's sad that that is the case. <dftxbs3e>butterypancake, no because RISC-V chips arent open either, the PowerISA is royalty-free just like RISC-V <butterypancake>dftxbs3e: risc-v isn't open hardware? wtf? than why do I care. Oh no, research time I guess <joshuaBP`>RISC-V chips also have you sign a NDA if you want to develop new standards. <kmicu>butterypancake: that could be a serious bug then. Do you keep your system config in a version control by any chance? <dftxbs3e>butterypancake, it isnt quite how it works, you have the ISA that you are free to implement without paying fees to anyone but individual manufacturers are free to give or not give you the necessary bits to manufacture the chip yourself - which is unlikely you'll do anyway because it's very expensive <kmicu>butterypancake: thers’s ISA and its implementations. We can have open ISA and at the same time libre and proprietary implementations. <dftxbs3e>"Open Hardware" doesnt require chip designs to be public domain or something <dftxbs3e>There's libre implementations of PowerISA like RISC-V but they are not performant, and for the case of RISC-V, no chip is actually really performant <dftxbs3e>The IBM POWER9 designs arent published, but all the firmware is, and PowerISA is royalty-free and tons of spec documents available freely <butterypancake>dftxbs3e: well true open hardware doesn't exist because people don't public domain their chips. I'd still like to use a chip that's public domain though. Is there a good term for open hardware where everything is published? (preferably even the manufacturing process but that's probably going too far) <joshuaBP`>dftxbs3e: what is firmware? is it binary data? binary code? <butterypancake>kmicu: the latest commit worked perfectly a week ago. This two days ago, I did another reconfigure because I really wanted the new kernel (no changes) and then computer go boom. <joshuaBP`>butterypancake: yeah, I reconfigured today. Went from linux 5.4 -> 5.7.14...I've got no issues. <dftxbs3e>joshuaBP`, well, it's what your hardware peripherals use as software to work, that includes bootloader because CPU is a "peripheral" and it needs it to work, I would sya <butterypancake>joshuaBP`: it's a binary that tells the computer how to use hardware (like a wifi card or processor). That binary is sometimes given by the manufactured (bad) and sometimes we can recreate the binary because we have source code (good) <dftxbs3e>the boot sequence on IBM POWER9 is comprised of skiboot, hostboot and petitboot <joshuaBP`>hmmm. thanks for the info. hopefully the 3-d printed laptop EOAM 68..? works. but I got to admit, that is a terrible name for a product and standard. <joshuaBP`>dftxbs3e: are there that many boot sequences on intel and amd chips? <dftxbs3e>joshuaBP`, who knows, it's proprietary so they may have implemented different boot sequences over time <dftxbs3e>The boot sequence has changed heavily between IBM POWER8 and IBM POWER9 <joshuaBP`>dftxbs3e: I didn't know they have different boot sequences...it power 9 open as much as power 8? <dftxbs3e>IBM POEWR9 is the only high performance chip that is that open <dftxbs3e>This level of openness otherwise only exists in either ancient hardware or embedded space <nckx>joshuaBP`: Probably more (and who knows what). ‘How many’ just depends on how you count/name things. <joshuaBP`>dftxbs3e: I can try to help you do some porting and testing of things, but I am probably way out of my league. I'm struggling getting guix deploy and shepherd user services to work. <joshuaBP`>heck, half of the time I can't even get irc to work well. my real nickname is joshuaBPMan, I've no idea why irc keeps pushing me into the current nick that I have. <dftxbs3e>joshuaBP`, I can give you access to a dev machine, I'm new to all this as well <dftxbs3e>I'm stuck on the first step, getting GNU Guix to work on my machine, then I'll struggle with using it :P <nckx>dftxbs3e: Is your most recent (git?) work public? <dftxbs3e>not everything is committed but what's not is really super super WIP <dftxbs3e>nckx, I was also stuck on this thing about gcc.. <dftxbs3e>Will dig what I attempted because it's been a month or so <dftxbs3e>I don't have AC, just a silly ventilator blowing full speed towards my face <nckx>Heh. You are describing me. <dftxbs3e>I'm so exhausted by this combination of things, hotness, and software complexity, since few weeks <nckx>I've asked Savannah peeps about branch-level push access (they must get that question all the time) but I suspect the answer will be no. That or ‘provide your own hook and we'll install it’, and I've never run a git server in my life. <dftxbs3e>nckx, I'm probably trustworthy enough but yes.. At least restrict master/core-updates/staging but anyways it requires to add my GPG key which I wont do <reepca>dftxbs3e: yikes... here I am feeling uncomfortable because our AC stopped working and it's 83 degrees Fahrenheit inside... *nckx (A mere 28.3333333333 degrees C.) <reepca>in my defense, I sweat very easily <dftxbs3e>This hotness will last until at least until next Wednesday (!!!)\ <dftxbs3e>the Talos II machine running full speed in the room and producing heat doesnt even matter anymore <dftxbs3e>nckx, I have a tricky question for you, can I/ <kmicu>So reepca is from US, Myanmar or Liberia. <kmicu>Europe created OCaml and Guix. Europe is blessed. <kmicu>Soon Ludo will be able to power Guix build farm with power from ITER’s fusion. <dftxbs3e>The code comments say the code is bad already <dftxbs3e>I'm trying to make the thing conditional, it wasnt before, which wouldnt be upstream-able <reepca>"So we did some statistical analysis, and it turns out 90% of americans live on ancient indian burial grounds. Legally speaking, that means you have the right to be called the Most Cursed Nation". <nckx>dftxbs3e: Try changing '("--enable-plugin" → `("--enable-plugin", and ,(if (string=? → ,@(if (string=? <joshuaBP`>dftxbs3e: I'm game to try to help out. Just ping me at some point. I've got to go to work now. <nckx>If I'm still seeing/thinking straight you also need to remove both '()s around "--with-long-double-128" and "". <nckx>dftxbs3e: Are we *still* stuck on this? I added that option in January :-/ (without the fancy & correct if, though). <nckx>And I know I wasn't the first. <dftxbs3e>nckx, At some point just that change didnt make things work, but now it does, that's one of the first patches, but many things have been done on top of non-upstreamed changes <dftxbs3e>Because I'm wasting my time patching something that moves <dftxbs3e>nckx, I tried that already and it does not work and the target variable well it's an attempt of mine too but it should contain the target system, not sure how I can make it contain that, (%current-target-system) or something? (that doesnt work, returns #f) <nckx>OK, I'll take a proper look it this later, I'm at ~10% attention here now. <dftxbs3e>Also I'm kind of a perfectionist and re-do things over and over until patches are as small as possible <dftxbs3e>nckx, making the thing quasiquote is messing with the other unquote stuff inside <nckx>Yeah, I didn't realise how messy the code was until I scrolled up. The comment isn't lying. I will take another look at this, I promise, but meanwhile it's 7:30 on a Sunday and I have to work for work and the sun has landed on earth :-/ Later. <dftxbs3e>nckx, See you later, thanks for paying attention after my desperate call for help. <dftxbs3e>nckx, Found another way, there's a gcc-configure-flags-for-triplet function above <butterypancake>no-one has looked at my patch that's been up for a week and I'm sad... It's not a hard patch to review. It's one line. and my other patch depends on it... <dftxbs3e>(I can't do much, I'm not an authorized committer, but still) <butterypancake>I mean no because no package uses that flag (yet) but It might cause a rebuild of every emacs package? I'm not sure how that works <butterypancake>the annoying part is that the flag is actually implemented. Someone else wrote code for the flag. They just never made sure it got passed into the build system <butterypancake>so, I just made a new boot media that now works. I do think at one point in time in the last few days the repo was in a bad state that broke my computer, but I think it's fine now, and I might even hazard another system reconfigure (although I'm very hesitant) <butterypancake>what do you guys think of making a guix-chroot script that works a lot like the arch-chroot script? It'd be so you can use an existing guix system from a recovery environment so you could easily roll-back a guix install that you can't currently boot to. It'd do stuff like mount /proc and /dev and start the guix-daemon <butterypancake>hopefully it'd do enough so a guix system roll-back would just work. So it'd source the root profile and everything <vagrantc>you'll also need the system profiles to be activated <butterypancake>I'm certainly not knowledgable to make this script then :P I'm not sure how this all works <butterypancake>maybe I'll throw a patch and hope the reviewers roast it until it actually works :P <vagrantc>overall, it sounds useful ... have had a similarly hard time chrooting into a guix system <nckx>If you do, consider making it a proper package so we can package it in Guix (and include it on the live medium) 😉 <nckx>Or a guix system chroot subcommand. <nckx>Whatever you think fits best. <nckx>(I like the sound of ☝ by the way.) *nckx will stop telling others to work now and do it myself. <nckx>It = my work, not it = the script, sorry, it's too damn hot. <nckx>Motivated? What's the opposite? Oh, right, that. <butterypancake>making it a guix system chroot command sounds like it should be compatible with the container commands. I haven't used the containers, is there already a command for getting a shell in one? <nckx>It's 8 *pm* here, sorry if that wasn't clear, but it feels like noon. <butterypancake>i feel like having it a script would make more sense than a sub command. It'd be more familiar for people coming from arch and then it wouldn't get mixed up with the container commands <butterypancake>the arch-chroot script is called arch-chroot.in. Why the .in? that seems odd to me <nckx>butterypancake: I don't understand: why would it get ‘mixed up’? <nckx>What Arch users are used to doesn't really matter. <butterypancake>nckx: I'd like a shell in my container. I'd like to chroot into it. Lemme use guix system chroot to go into the container. idk, that's what I'd think. <nckx>butterypancake: So the ‘guix system’ part of the command doesn't make it clear? I'm sceptical 😉 <nckx>butterypancake: So how would this work on Arch? <nckx>And I've never used a container. <butterypancake>well the arch install media has an arch-chroot script. They have the script in a package called arch-install-scripts which is full of handy script to have on an installer. This script isn't super useful when your not installing anyways so putting it in a package instead of the guix binary seems right <butterypancake>idk, maybe I'm wrong. This entire thing is a little above my pay grade <nckx>Is it just me or are you using ‘script’ in a specific... meaning? To me, it seems like an irrelevant implementation detail. It would probably be Guile, sure, but the users wouldn't need to know. <nckx>They just type ‘guix{ ,-}chroot’ and off they pop. <butterypancake>cuz I wanted to do it in shell because I wanna just copy the arch-chroot shell file. guile is scary and the documentation sucks <nckx>Anyway, as much as I like the subcommand idea, 99% of the time it's going to be used from the live medium. I guess. Although I've certainly chrooted a Guix System from another (real) Guix System so it makes total sense to me 🤷 <butterypancake>leoprikler: functional is fine, but I never know where to go when I need to look up the documenation for a function <butterypancake>well then I have to figure out how to get a proper repl but I can never figure out what modules to import anyways <nckx>guile without readline is pretty sharty, but once you install guile-readline it's eminently usable. <butterypancake>I use emacs and I just want to be able to use the geiser function to describe the symbol at point, but the gesier repl never know what I'm looking on <leoprikler>you probably want to paste your use-modules into the geiser repl <leoprikler>e.g. try ,use (srfi srfi-1) and then describe "partition" at point <butterypancake>maybe it's hidden in the back or something. Would not recommend those reads for someone learning guile though <nckx>joshuaBPMan: Ta. Cool nick bud. <nckx>Did you find a forgotten ‘fallback nick’ setting in your client? ‘IRC’ (Freenode) wouldn't cut off a random piece like that. <nckx>butterypancake: I don't really disagree with you, but what *would* you recommend? <leoprikler>I'm already confused w.r.t. what you expect from geiser. <butterypancake>nckx: someone to update those damn manuals :P they really should be what I recommend <mroh>butterypancake: also, C-c C-b (geiser-eval-buffer) often helps to make things "known" for geiser. <leoprikler>I'm pretty sure it's not the only autocompletion tool for Emacs, that relies on knowledge being stored in some accessible buffer (here the geiser repl) <nckx>I haven't run Geiser this year; let's give mroh's tip a whiz. <butterypancake>in a package description, how do I format a reference to another package? <nckx>butterypancake: Can you be more specific? Or link to an example? <butterypancake>"this package gains extra functionality if you also install @code{other-package}" <joshuaBPMan>I just added in a section about installing a guix system on a linode. <nckx>Thanks, I saw that. Does dustyweb not have commit access? <nckx>(‘A Linode’ will never sound right to me but I guess that's the punt.) <joshuaBPMan>nckx: He does. I just submitted an updated patch today...he's been waiting for me to format his email for about 2 weeks now. So it's really silly of me to say that I'm waiting on him. <joshuaBPMan>Just to be clear everyone in this room loves Chris Webber. <joshuaBPMan>nckx: what were you joking about...? someone needed to update the manuals? <joshuaBPMan>I actually realized today, that even though I have guix system on a linode...if an update screws up the linode, then I cannot roll back the guix-y way. I'll have to make periodic updates. <nckx>joshuaBPMan: Which line? I don't think I was joking about the manuals just now. <nckx>joshuaBPMan: So why can't you load GRUB? <dustyweb>joshuaBPMan: ah, I need to catch up and commit it :) <nckx>joshuaBPMan: Ah, no (serial) console at all? Never mind then. <joshuaBPMan>hey hey! we've been talking about you! all good things all good things. :) <joshuaBPMan>take you time. I dragged my feet making that patch anyway. <butterypancake>joshuaBPMan: I was actually talking about the guile and geiser manuals. The guix and guix-cookbook manuals are quite good in my opinion <nckx>Even the most budget host I ever used (Scaleway) supported a serial tty that I could teach Guix's GRUB to use. <dustyweb>joshuaBPMan: yeah you can find it on the web gui... somewhere ;) <nckx>butterypancake: Thanks for the compliment 😉 <dustyweb>joshuaBPMan: I'll try to get to your patch soon, or if anyone else wants to, also welcome ;) *dustyweb gets back to outlining another FOSS & Crafts episode <joshuaBPMan>I'll have to checkout your FOSS and crafts podcast. I liked the last podcast you did too. <fnstudio>str1ngs: thanks for your help earlier by the way, sorry i had to go afk <dustyweb>joshuaBPMan: thanks :) I do hope to cover Guix on FOSS & Crafts at some point too but we'll have to think of a thematic hook <dustyweb>if you're looking for a friendly casual conversation channel, #fossandcrafts has been fairly nice to hang out in (and guix is frequently under recent discussion) <dustyweb>I do hope I can get more guix hacking time again soon... so much going on tho! <PotentialUser-86>Hi, I have a question: I installed emacs-vterm but it seems that emacs does not recognize the dependancy libvterm. The guix documentation make it sound like it should happen automatically. Does anyone know how to figure out why it doesn't? <dustyweb>PotentialUser-86: did you restart emacs by any chance? <dustyweb>PotentialUser-86: hm, another possibility is that the emacs lisp profile paths aren't set up in your environment <mroh>PotentialUser-86: guix system or foreign distro? <mroh>PotentialUser-86: you get a runtime error after `M-x vterm` or you cant build it? <PotentialUser-86>same happens also with emacs-pdf-tools that also depend on external package <mroh>hmm, something wrong with your clock/time? <dustyweb>PotentialUser-86: did you just start installing emacs packages? <dustyweb>I ask that because unfortunately sometimes when emacs starts adding new custom paths from your configuration, the "easiest" way is to log out and log in again <dustyweb>after you install the first few, it'll have it set up by default <dustyweb>try logging out and in with your entire user <fnstudio>str1ngs: lol you know what, i did a couple more attempts (with minor changes) around the locale problem and now... it works! so at least you know all your help paved the way to success :) <dustyweb>PotentialUser-86: not the most fun solution, but possibly the easiest <dustyweb>you shouldn't have to do it again after the first itme <dustyweb>there's also a command to list suggested environment variables but I don't remmber <fnstudio>minor attempts = i relaunched `guix install glibc-locales` again, both from root and as a normal user, and this time it did it <dustyweb>PotentialUser-86: lucky for you I am procrastinating on other tasks ;) <nckx>I'm testing your other patches though. <nckx>mroh: How do you test these, by the way? <janneke>dftxbs3e: as long as you're working on a new port that's not official, you can more or less set your own rules <janneke>if you believe that postponing to achieve reproducibility helps, that's okay <nckx>janneke: I believe they want to postpone reproducibility to achieve n>1 workers, so as not to burn out. <nckx>mroh: Worth a try, but no :-p I'm already squeezing this in. <nckx>mroh: It looks OK. I'll test & merge. Just not now. <janneke>also, ime, it often helps to let a problem "marinate" while working on other stuff <janneke>doing n==1 development can be exhausting <joshuaBPMan>janneke: I personally don't know how you did all that hurd development. That sounds like a ton of work! And super awesome! <janneke>joshuaBPMan: while i did a lot of the work, i didn't do it alone <mroh>nckx: you mean, my commit msg should have included "[inputs]: added gtk+"? <NieDzejkob>mroh: Yeah. Also, definitely don't make the reader look up a bug number to understand why a commit is there <NieDzejkob>you can reference a bug#, but this should be an additional note, not all the info there is <NieDzejkob>nckx: re 4cafdce2102e3ebc7b3a6152464a62a454b6be45: wouldn't that attempt to call the git package as a procedure? (aka you're missing a quasiquote or a call to list) <NieDzejkob>is there a texinfo command that allows you to define and explain a term in a package definition? <nckx>NieDzejkob: Without looking, I'm sure you're right, I'm hopeless at writing code outside of a ‘computer can say no’ situation. <fnstudio>ouch more locale problems here, guix on top of alien distro; i've recently installed glibc-locales and glibc-utf8-locales so as to get rid of the "guile: warning: failed to install locale" warning <fnstudio>the warning has gone but in my attempt i must have misconfigured something else <fnstudio>launching a terminal produces this error: `urxvt: default locale unavailable, check LC_* and LANG variables. Continuing` <fnstudio>any idea on what i may have misconfigured? <nckx>NieDzejkob: Thanks, fixed. And @dfn is exactly what you want. <nckx>Double-check how it renders, though. Pragmatics trump semantics :-/ <terpri_>wait, why would i want to source env vars from both ~/.config/guix and ~/.guix-profile? are they used for different things? <terpri_>also nckx, xkb on its own is not so bad, actually ;) simple changes can be made pretty easily, it's just poorly documented <terpri_>...and tbh i don't know why the documentation is so bad (no xkeycaps equivalent -- even though it would be easier to implement properly with xkb! -- no xmodmap equivalent for trivial changes, no good tutorials i've run across, etc., etc.) ***terpri_ is now known as terpri
<terpri>let me paste a simple example, for "lisp-dvorak" (dvorak has '<>' on the qwerty 'we', and parens on the same number keys as qwerty, i figured it might be more ergonomic to swap them) <krusnus>Is there any way to get firefox sync and lockwise working in icecat? <terpri>krusnus, dunno if they just hid the UI or stripped out the code, let me take a look (lots of services.sync settings in about:prefs...) <joshuaBPMan>It says I'm synced, but I'm not sure that it is actually syncing passwords... all the time <joshuaBPMan>Also note, that running firefox means that you are trusting a pre-built binary. Guix cannot build firefox from source without great difficulty. <joshuaBPMan>but I trust the people behind mozilla, so I'm ok with using firefox. <terpri>in icecat changelog: "Put Firefox Sync back." hmm <joshuaBPMan>I personally have had just better luck with firefox. Icecat, the last time I used it, would get really laggy. And sluggish. <joshuaBPMan>Firefox still occassionally mis-behaves, but not as much <krusnus>joshuaBPMan: thanks I'll look into it! <fnstudio>is there an equivalent of what in debian is called `locale-gen` in guix (when run on top of a foreign distro)? <terpri>joshuaBPMan, it's not really that hard, the main difficulty is running a post-ESR version because it requires the latest & greatest version of rust... <terpri>i have standalone packages (for firefox development) but only for now-ESR versions :( <joshuaBPMan>terpri: I guess it's not hard, it just takes a lot of compile time right? The rust bootstrap bath is a C++ compiling an old rust, which in turn slowly compiles later rust compilers.... <terpri>joshuaBPMan, one issue is that icecat is based on firefox ESR, which lags several versions behind plain firefox, missing out on performance improvements often <terpri>krusnus, about:config, type in the setting name and set the value as desired, both of those options are predefined in (my) icecat <terpri>(sometimes you have to define new settings via the right-click menu, but not in this case i think. if they were missing from about:config, you do right click -> new -> boolean for both <krusnus>terpri: thanks for all the help it seem to be woking now! if i have more problems i'll look into the nix service thing :) <krusnus>and thanks joshuaBPMan aswell of course :p <terpri>joshuaBPMan, also about:sync-log shows you ... well, just a filesystem log directory actually, but there is a WebExtension to show exactly what is stored on the sync server, hard to find the right keywords for it though :( (i'm sure it's free software) <terpri>there is also now an official firefox flatpak package available from flathub.org as of...july 20 <terpri>(caveat: flathub.org also distributes nonfree software, and of course with plain firefox you're on your own for turning off anything you consider to be an anti-feature) <terpri>(there was a flatpak of at least Developer Edition (~= beta) firefox before, but it was unofficial; this one is sanctioned by mozilla) <terpri>(and naturally you could build the flatpak recipe if you don't trust flathub.org) <joshuaBPMan>terpri: have you tried using any flatpaks yet? What's your opinion of it? <joshuaBPMan>terpri: does the flatpak build script depend on a binary of rust? Also I only have 8GB of ram on an old laptop...I wonder what firefox's build requirements are <joshuaBPMan>hmmm, apparently 4GB of RAM with 4GB swap is the minimum needed to build firefox <kamil_>joshuaBPMan: I have. No, programs distributed through Flathub, Flatpak's official remote are pre-compiled. <joshuaBPMan>kamil_ I guess I tend to trust the people behind flatpak are not bundling back doors in the software I run. <kamil_>joshuaBPMan: In regards to the Firefox flatpak, it is directly distributed by Mozilla Foundation <kamil_>I meant that it is published by them <joshuaBPMan>yup. I admire and support the guix view of being able to build software is important, but I just can't get icecat to work reliable for me. (admittedly, I haven't tried it in about 2-6 months). <kamil_>joshuaBPMan: Yes, it does. You'll also need to explicitly install xdg-desktop-portal and xdg-desktop-portal-gtk and, depending on your DE/WM of choice, install dbus. <kamil_>You'll also need to append 'dbus-run-session' when starting a WM/DE from a TTY <kamil_>AFAIK it's a wrapper that starts a dbus user session for the duration of time that a program is running for <joshuaBPMan>I'm using sway...I think it run dbus for me...but I don't really know though. <kamil_>joshuaBPMan: You'll certainly need it then. `dbus-run-session sway` will do the trick. Without it, saving files from within flatpaks won't work. <kamil_>For Sway elogind is needed on Guix. Dbus will be automatically pulled in as a depedency of the elogind service. Yes, it'll automatically start on boot. No, you'll still need to launch a separate dbus session that runs in the user mode -- the one that is started by Guix runs in the system mode, which somehow matters to Flatpak. <joshuaBPMan>Also what is dbus? It's linux way for inter-process communication right? <terpri>joshuaBPMan, i use the "unofficial" fedora(?) flatpaks for sites that don't work with icecat (UA sniffing etc.). no issues IME <kamil_>joshuaBPMan: I'd say it's more of Freedesktop way for doing that, but yes, I think so <kamil_>joshuaBPMan: One more thing. Since you're using Sway, you might also want to give Firefox flatpak permission to wayland sockets and set env variable MOZ_ENABLE_WAYLAND=1 to run it natively under Wayland. (flatpak override --help for instructions on how to) <terpri>it's unfortunate that icecat and mozilla haven't been able to collaborate more closely, e.g. why not provide an option that filters out nonfree extensions (license is part of extension metadata)...instead of making users go to addons.mozilla.org directly and check licenses themselves 