<xavierm02>Ok so I found the source of my problem: I used pdflatex from /gnu/store/023xigaka89k984zwlg95ic68nj5a1mb-texlive-union-49435/bin/pdflatex, and this pdflatex isn't aware of amssymb even though it's installed. Installing texlive-bin fixed it. The next question is then where does that textlive-union come from?
<apteryx>this allows writing complicated mcron job as a Guile script, directly in the config file. Using the lambda native mean of mcron fails short in regards to importing modules (hit or miss depending on whether the symbols imported are syntax or procedures)
<xavierm02>rekado_: I tjink my problems were also related to fish and environment variable not being updated but I'm not sure.
<apteryx>(the program-file approach works around this by handing a script to mcron rather than a lambda)
<nckx>apteryx: Oh, nice! Interesting how it ties back to your original 'how do I import modules differently' which I'd kinda forgot…
<kori>nckx: a few pages won't load properly (namely whatsapp, wire, can't tell whats wrong though. wire says the browser is unsupported, i went to install user agent switcher, and icecat said the extension was corrupted, and that was that)
<lfam>Even epiphany is really expensive because it uses webkitgtk
<nckx>gnutec: Installing Guix System on btrfs is as easy as installing it on ext4 (you'll want to chattr +C the database in /var/guix, but that goes for all DBs on btrfs). I don't know whether the installer supports it though.
<nckx>Also, don't go doing things like cross-mounting subvolumes at different places because that will not end in puppies and smiles.
<jfred>oh heck yeah, just got libhandy building as a guix package
<raghavgururajan>nckx Hmm. Iridium's website clearly explains their vision and mission. Glad they mentioned Iridium browser as Free Software instead of so-called open source. So how true their statements are , is a matter of aduit.
<jfred>No problem! I wanna use stuff that depends on it myself so... :P
<jfred>I'm still pretty new to Guix packaging and not at all familiar with Meson/Ninja... took me a bit to figure out that there's a build option to disable installation of glade catalog files
<nckx>dftxbs3e: Yup, and I agree we should eventually follow suit if it doesn't have drawbacks to do so.
<nckx>I'm not sure how modular the build system is (if not it means we'd have to fully build qemu for each, what, 40 packages? that would be quite problematic), and/or if outputs would be a sane solution or not.
<jfred>I'm guessing the real solution is to use native-search-path somehow to let glade pick up on plugins, and set catalogdir so meson doesn't try to shove plugin files into the glade package's dir in the store, but... baby steps
<leungbk>Right now if I run something like `(assoc-ref (json-fetch (string-append "https://rubygems.org/api/v1/gems/" "foo" ".json")) "name")` in my Guile REPL, I get #f. I'm on the latest master branch, but had been messing around with some of the import-related files. I've tried rebuilding but the problem persists. Any idea what might be causing this?
<leungbk>If I `guix pull` and run `guix import` everything's fine, the only problem is when I either experiment with the source or if I'm using `./pre-inst-env`.
<leungbk>And to be clear, this problem persists even when I'm using the REPL with the source unchanged from the current master branch.
<efraim>This is probably related to the guile-json change, you might need to re-bootstrap and configure and make
<leungbk>I had done the re-bootstrap/configure/make-ing, though I didn't do anything specifically related to guile-json; do I need to do something like `./pre-inst-env guix install guile-json` in addition to this?
<leungbk>Or maybe I just have to first `make clean-go`, that would make more sense.
<xavierm02>I'd really prefer to not use emacs. I get that at the moment it has a lot of the knowledgeable users, but the entry cost seems too high, and using it would be like admitting that my dream of making a structural editor someday will never happen
<nixo_>Hello people, I'm again in need of help. I've been working on julia-build-system and it's working. Now I wanted to submit a work in progress to have a small code review before cleaning everything up, but I can't guix-pull the repo. I get ce-9/eval.scm:159:9: Throw to key `srfi-34' with args `(#<condition &message [message: "icecat-makeicecat.patch: patch not found"] 57d3f00>)
<nixo_>What I did is to add my files under my local repo (guix/build-system/julia.scm, guix/build/julia-build-system.scm), adding them to Makefile.am, add the repo to my channels.scm and guix pull. The error isn't related to my code, so I don't how my commit is breaking it .-.
<roptat>nixo_, your work is not related to that icecat-makeicecat.patch at all, right?
<nixo_>roptat: not at all. But setting (commmit "") to be27b8f2e1 (latest master) does work
<roptat>does your copy of the repo have that file?
<roptat>since it means the software also works perfectly fine when called outside of a profile or environment (it can be called from its store path, which most user won't do apart from me :))
<roptat>(I needed an ugly workaround for the same kind of issue with our minetest package: it propagates minetest-data, but my user config uses gexps to refer to minetest's store path, so I had to do something to get the right variable in my environment so the gama data could be found)
<roptat>but I guess it can also look for its data elsewhere in my home directory, so maybe that workaround wasn't really necessary
<roptat>and it's a nice use-case for mods actually :)
<minall>I'm trying to see the functionality of the driver xf86-video-openchrome, any suggestions for a noob?
<efraim>xavierm02: you should look at wrapping the binary
<efraim>Not sure of the best grep option, maybe 'git grep QT_'
<minall>Hello guix!, I'm looking about how does the xf86-video-openchrome driver works, just for researching, and I want to modify something so I understand all the workings, any tips for a newbie?
<shiftlockboom>Hi. I am trying to install Guix on a Libreboot system using the guide. When I do the nano /mnt/etc/config.scm I must have some error coz the guix system init /mnt/etc/config.scm /mnt keeps giving me an erro.
<fps>i wonder how i can debug the failure to log me into any kind of desktop session. the login screen just freezes after logging into e.g. xfce. switching to console and back seems to give me back the login manager, but the session type choice button then has disappeared
<fps>i'll see if i can disable the desktop services and just start x manually from the console with a valid .xsessionrc or .xinitrc
<alexanderbarbosa>Im after no needing to use login managers... just startx.... still have to learn how to... it never finds Xorg...
<nckx>alexanderbarbosa: I used (service special-files-service-type `(("/bin/startx" ,(xorg-start-command)))), it's not user-friendly (and might need some tweaks to load configuration/WM, didn't try that), but it works and displays clients.
<alexanderbarbosa>nckx: of course... I literally ignored that, thinking that is was an issue with my dotfiles ... haha
<nckx>xorg-start-command calls Xorg directly so it doesn't have to ‘find’ anything. Not sure what went wrong there.
<nckx>alexanderbarbosa: I haven't bothered investigating, but its font issues are often not ‘synchronised’ with the rest of the desktop: moji fonts won't work in IceCat, but in everything else, or the opposite. So definitely doing their own special font-findin' thing.
<nckx>Then there was the weird bug of IceCat not rendering numbers — but only numbers! — when its favourite font wasn't installed. Maybe it's fixed now. Weird things like that.
<alexanderbarbosa>nckx: indeed. Guix as package manager under Debian, I always had those issues with icecat. But as I switch to GuixSD everything is fine and easier to fix
<nckx>alexanderbarbosa: That is also my experience, but for some people it's been the other way round, unfortunately.
<nckx>Guix System is generally better supported since it's what all the developers use.
<alexanderbarbosa>damn, icecat and chromium installation are way bigger than emacs ... lol
<vagrantc>there are a few tricky things with running guix on top of a foreign distro indeed.
<keno>I'll file a bug report for that then, thanks for tracking it down!
<keno>perhaps I can trick connmanctl to return online, I'll look into it
<nckx>keno: Thanks! If connmanctl won't budge you can try ‘kicking’ connman with ‘rfkill block all; rfkill unblock all’ if rfkill is installed. But please write your bug report first, if you can, so you can check the current state of your system.
<keno>What would that rfkill stuff be good for? Can you please rephrase?
<nckx>It's the equivalent of turning your wifi on & off again.
<nckx>Toggling a hardware (‘radio frequency kill’) switch would do the same thing but very few machines have those nowadays.
<keno>Oh okay I doubt that will help, I've already turned it on and off again within connmanctl
<keno>I have a hardware switch :) Already tried this one, too
<nckx>Good news for debugging, bad news for your installation completing today. 🙂
<nckx>‘A network is in state 'ready' once it has obtained an IPv4 or IPv6 address or both. The addresses can be acquired via DHCPv4 or DHCPv6, IPv6 address autoconfiguration, IPv4 link local assignment or statically configured for IPv4 or IPv6. A network is in state 'online' if ConnMan has verified connectivity to Internet, i.e. it has managed to look up and connect to ipv4.connman.net or ipv6.connman.net. For all practical purposes 'ready' and 'online' are usuall
<nckx>y equivalent for the intended connectivity experience.’
<nckx>Intel is telling us to treat them as equivalent for our purpose, I think
<nckx>I'll rephrase that: if we do check for proxies and other vagaries, we might want to do it ourselves and not rely on connman.net.
<nckx>Which currently 404s, although that might not affect the probe.
<keno>Yes I think 404 is okay, it just tries to confirm general connection to this domain. And if it achieved to get a 404, it was connected after all
<keno>Even pinging ipv4.connman.net timed out for a while, now it works again
<nckx>I wasn't expecting it to break connman, I am somewhat suprised/disappointed at the boringness of that response. A plain-text ‘It's alive! - connman’ response would consume much less bandwidth and look less dodgy.
<avocadoras>should I set it to GUILE_LOAD_PATH=/usr/share/guile?
<emsyr>Hi. I have big problem to remove gdm for I want to try slim and sddm. Here is my config.scm: https://paste.debian.net/1093638/ . I get an error : service 'xorg-server' provided more than once...
<pkill9>avocadoras: is that after exporting the GUILE_LOAD_PATH and then running guile in the same shell?
<kori>but let's pretend I typed in guix and not guile!
<nckx>avocadoras: I haven't been following the discussion, but (and %load-path doesn't seem to include it anyway) but not that using ‘~’ in an environment variable like that isn't magically expanded to $HOME.
<avocadoras>bit off-topic,this journey started one day when I decided to learn Emacs
<nckx>xavierm02: OK. I guess grub.cfg references aren't considered at all by the garbage collector. I don't see how exactly that happens (I thought the grub.cfg copy in the store would serve as a root), but it happens.
<nckx>xavierm02: Do you have the full file name of the then-missing unifont?