IRC channel logs
2025-05-01.log
back to list of logs
<Tadhgmister>if I have multiple fonts installed how do I specify which should be used as a system default? <sneek>Tadhgmister, you have 1 message! <sneek>Tadhgmister, apteryx says: it was bug#60847 <Tadhgmister>right now my browser is only rendering a specific range of unicode that I have some custom glyphs defined for and not rendering any other text unless I start a shell with `fontconfig` and `font-gnu-freefont`, run `fc-cache -r` and then launch by browser from that shell, any otehr case it seems to just only use my font which doesn't define any <nomike>BTW: I just saw that there is a (define-public clipper) in cpp.scm which is a "Polygon and line clipping and offsetting library" and a (define-public clipper) in bioinformatics.scm which is the "CLIP peak enrichment recognition". <nomike>Is this name-conflict an issue that should be raised? <apteryx>and they share the same name as well? <apteryx>in any case, yes, that's probably worth fixing to use unique variable names, otherwise combining both modules would have one shadow the other <nomike>Yes, definitely a full on name conflict. Also in the package name. <nomike>We should deinitely rename the bioinformatics clipper package, but I'm struggling to come up with a good name for it... <nomike>There is another package with exactly the same name. <apteryx>nomike: maybe, yeolab-clipper, with an explanatory comment <apteryx>or rename the original 'clipper' to polyclipper as done on freebsd, that's the name of the repo <apteryx>we can stick to what users expect when we don't have a good idea <apteryx>(and when there's a conflict our where our naming guidelines fall short) <apteryx>lechner: I've pushed nyacc-2.01 in 748b5858cc9 <apteryx>sneek: later tell partosqq we now have openjdk22 in the repo <nomike>apteryx, I don't think we should rename the cpp.csm/clipper. It is a library which is used by 7 others. The bioinformatic package is not referenced by anything. And it's more likely to have a use-module for cpp in a file than a use bioinformatics. <nomike>I will rename it to clipper-peaks. That is easy to remember and it fits the package synopsis. I'm against including the github user, because that's just arbitrary with no real meaning. <apteryx>one could argue both ways. the beauty of names <apteryx>this metadata appears in virt-manager and Boxes when selecting the Guix System .iso installer for example (or when it is manually searched for). <apteryx>sneek: later tell partosqq we now have openjdk23 and openjdk24 in the repo; time to 'guix pull'! <apteryx>sneek: partosqq see commit d505cb960fd for when openjdk25 comes around; it's easier than attempting to distribute a prebuilt binary (with the exceptions of maintaining custom patches, but there are now none). <Tadhgmister>if I want to run `nix-env -r -i` every time I reconfigure my home config would I put that in an activation service? <hako>nomike: The old name should be preserved as deprecated, see Deprecation Policy in the manual. <nomike>Well, the guix manual says on this, that when renaming a package the old name needs to remain available at least for a year as a deprecation. But in this case IMHO this is not an option, as this would keep the naming conflict in place up until then, <hako>hmmmm, (cpp) clipper has a higher version which will be picked by guix package, so I think it's fine to rename the bioinformatics one. <meaty>Does anyone here have experience deploying servers to the hetzner cloud? what service would I buy for a VPS to run a static site, a mail server, and a calendar server via guix? Also, what is the bootstrap process like--the manual glosses over it, but would we just build an operating system image and send it in to the recovery interface? do hetzner VPSs work with linux-libre? <meaty>Also is there a good way to practice administering a server this way before committing to a vps <hako>The variable can be deprecated <apteryx>anyone knows why virt-manager reboot and shutdown button do not work for guix system? <andreas-e>Hello nomike! I think you have been busy on Guix, thanks for your bug reports! <andreas-e>Hello! ngz, so you ended up using datefudge? <andreas-e>futurile: Not sure if you have time or interest, but if yes, you could review the commits on the games-update branch. I still think we could just push them to master without going through the branch dance. <futurile>andreas-e: Thanks for the reminder - I have the interest, I'll do it over the weekend and push them <ruther>I am wondering... why don't guix records expose 'base' when inheriting? They already can expose stuff like this-package, references to other fields... So I am really puzzled this would be a technical issue. <pastor>Hello. I'm going to submit a patch for adding a new service. This service requires first to fix a package which is broken. Sould this be 2 separate issues or in the same issue I put the 2 commits? <ruther>pastor: I don't think it matters much, but if it's completely broken, probably as a separate patch so that it can be merged first in case of discussion <ruther>pastor: also if it's completely broken, are you sure there isn't a patch for it already? <pastor>ruther: It's not completely broken. Some of the 'libexec' files are not properly wrapped. I doubt anyone noticed since they are not usually required <ruther>pastor: yeah, if it's just that, I would also send just one <old>ehh I just saw on the ML about it now that I can fetch my mails <pastor>old: I didn't know you where active here :) <old>pastor: I hang around yes :p <pastor>Does anyone know which package provides this qt module: `Failed to find required Qt component "Qml".` <pastor>Oh, it's in `qtdeclarative`. Unexpected name <luca>Hi, does anyone have any tips for using a touchscreen laptop on guix? I installed gnome and sway and icecat but scrolling doesn't really seem to work, and in icecat at least touch behaves more like a mouse pointer than a "finger" <ieure>luca, Firefox has gotten better support for touch on desktops lately, I'm not sure which version they landed in. <ieure>luca, There are extensions you can install to make the touchscreen experience better on Firefox-based browsers. <luca>Oh I see. So maybe icecat is relatively old compared to upstream firefox <dstolfa>i think icecat is based on firefox esr, not firefox itself <ieure>luca, IceCat is based on the Firefox ESRs, they're one ESR behind last I checked. <luca>Well newest firefox is much better at this, so the issue must have been fixed upstream already. Thanks! <ieure>luca, Give LibreWolf a whirl, it's based on the non-ESR channel and Guix has the latest release, because I keep it updated. <luca>Well regardless, this is my first touchscreen laptop so I am very much open to touchscreen tips and tricks <ieure>luca, Flip side of that is that it that a web browser is a big package and the non-ESRs update 1-3 times a month. <luca>Oh nice. How did librewolf get added, but not firefox? In terms of licenses <ieure>luca, The Firefox branding is what makes it non-free, LW has their own branding and doesn't restrict you. <ieure>You can get regular Firefox from a third-party Guix channel. Personally, I wouldn't recommend that. <luca>Yeah, I'm aware. But librewolf should do just fine <sham1>Isn't one of the things with Firefox also some of the Mozilla specials that they put in? Which then get stripped by things like icecat <sham1>Which is also why it's incompatible? <ieure>sham1, I'm not really sure about that. I personally am done fighting Mozilla and have zero interest in using Firefox, which has become another piece of user-hostile software. <ieure>I'm happy with LibreWolf for now. I am not happy that Mozilla is self-owning itself into irrelevance and Google domination of the web. <ieure>But I have no power to change that, while I can choose what software I use. <sham1>Yeah. Sadly it's very difficult to even make anything viable in the web space anymore, to try and combat that. Mozilla was probably the best bet, but then... yeah, they drove themselves off the cliff <ieure>Yeah. Hopefully Servo ends up in some sort of usable state. <ieure>Why is `guix pull --list-generations' terribly slow, but `guix system list-generations' and `guix home list-generations' very fast? <ieure>`guix system list-generations' -> 1.5 seconds; `guix home list-generations' -> <1 second; `guix pull --list-generations' -> it's still running. <ieure>I am watching as it grinds for several seconds in between printing the generation info. <ieure>Pull does have more generations, but not 200x more. <ieure>41 pull generations, 28 home generations, 35 system generations. <ieure>But `guix pull --list-generations' is > 200x slower. <ieure>Guess I'll open a bug about it. <jdr>I'm getting a 502 when I try to do a reconfigure, but I have no problems doing a guix pull. Could it be a problem with savannah? if so is there a way to force reconfigure to use a mirror like with --url for pull? <ruther>jdr: reconfigure doesn't really use savannah when unless you've changed guix-service-type's guix field, and if you did that, just make sure to put in the url with codeberg, not savannah in the channels <ruther>jdr: in case you're not changing guix-service-type, can you send backtrace if there is one? <jdr>@ruther I haven't modified that. I'll get the backtrace now. <ruther>jdr: does guix describe currently show you savannah or codeberg url? <jdr>savannah, but I do have other channels like guix-science that are on codeberg <ruther>jdr: that doesn't matter, what matters is guix channel. Yes, it can be because savannah is having issues. I am not sure what the best course of action would be, I think there are two possibilities to workaround this issue. <ruther>or like... there are more, but two are quite easy <ruther>either pull from codeberg, you can pull the same commit if you don't want to update, or use sudo -E instead of sudo (but note that you might end up with root owned files in your home that way) <ruther>the second way will work only if you haven't removed your cached checkouts, but I presume you didn't <jdr>I haven't removed anything, what's the second way? (not that I mind doing the first) <ruther>I've described both the ways in the same message <jdr>Ah, bad reading comprehension. Thanks a lot. I'll try them in order. <ruther>I really think it should be possible to disable the forward check... <ruther>I've seen two diffeerent issues caused by it today <cluelessguixer>Is "nmcli con mod br0 connection.autoconnect yes" enough for peristence across reboots? What about my system config file? <jdr>@ruther pulling from codeberg fixed it. I really appreciate the help. I see that https://libreplanet.org/wiki/Group:Guix/Mirrors says that a migration to codeberg as the primary server is being considered, I think I'd be in favour of at least having it as a fallback. Savannah's great and all, but it doesn't seem to be able to cope with the demands of Guix. <ieure>jdr, The voting process for moving to Codeberg is happening now. <ruther>jdr: the same issue can still happen with codeberg if it went down, I think it's ageneral problem. Sure, moving to codeberg is going to make it less frequent, hopefully. <jdr>@ieure cool, fingers crossed for that <ieure>Yeah, ideally we'd be able to automatically direct users to a mirror instead of having a SPOF. <ruther>ieure: I think that in first place there should be a possibility to disable the forward update check. It's not real to make fallback for every channel a user uses. Why constrain users on updating only if the channels are accessible to check a forward update if the user can just say: yes, it's a forward update <nixfreak_>Hello , trying to figure out why when I run emacs in a guix shell , I stil get doomemacs to come up , even though the I put source ~/.guix-profile before a ran emacs ? <nixfreak_>When I closed geiser-guile repl I saw command it was running doomacs first then -L ~/.guix-profile .. what am I doing wrong <ieure>nixfreak_, Guix System or foreign? <ruther>nixfreak_: why do you expect to not get doomemacs? - What's in your .emacs.d? <nixfreak_>Ruther, I installed emacs using guix ,I thought regular emacs using guix-profile would come up ? <nixfreak_>I'm trying to run a different instance of emacs with guile and the guile libraries I installed with guix <ruther>nixfreak_: yes, regular emacs comes up, loading the user configuration from ~/.emacs.d, if your ~/.emacs.d points to a doomemacs config, you get doomemacs. <zilti>How do I get Emacs 30.1 on Guix? <nixfreak_>so your saying this emacs instance /home/nixfreak/.guix-profile/bin/emacs is symlinked to doomemacs ? so its not a virtualenv ? <ieure>nixfreak_, No, that is not what he's saying. <nixfreak_>Ok so if I want vanilla emacs with geiser-guile repl would it be ~/.guix-profile/bin/emacs-29.4 -Q guile geiser-guile ? <ruther>nixfreak_: yes, -Q means no config, so doomemacs shouldn't be loaded <nixfreak_>right and that works just fine but geiser-guile isn't working <ieure>nixfreak_, `guix shell emacs guile emacs-geiser -- emacs -Q' <ruther>nixfreak_: the emacs has to be in the same profile as emacs. Forgot to say that. Otherwise you don't get the search path for emacs packages <ieure>Mmm, I guess you also need emacs-geiser-guile in there. <nixfreak_>guix package --list-installedemacs 29.4 out/gnu/store/ghjfl16q82z7kx1kra0fs9bw99j5rvpd-emacs-29.4guile 3.0.9 out/gnu/store/2dqynvxbfhw39bq8phw3kbpxcpd5yg0a-guile-3.0.9emacs-geiser 0.31.1out/gnu/store/bqls0jspv31ncipzsj87ykcx89y9lawk-emacs-geiser-0.31.1emacs-geiser-guile0.28.3out/gnu/store/za8ccyp3bxhklhjnbfkd64qavcrh41n4-emacs-geiser-guile-0.28.3guile-goblins <nixfreak_>0.15.1out/gnu/store/q4648183qkl43ky8yp3qqikv63j67dw7-guile-goblins-0.15.1 <ruther>nixfreak_: did you relog or source the profile after installing emacs and packages initially? You won't get the search path until you source the profile. You can use shell of course, as ieure pointed out <ieure>zilti, Emacs 30.1 isn't packaged yet. <ieure>zilti, I believe there's a snapshot in the emacs-team branch. <zilti>ieure, yes, it not being packaged yet is why I'm asking :P I did try the emacs-team branch, but the emacs 30.1 on there fails to even apply the patches, unfortunately <ieure>zilti, Okay, well, the answer to your question then is "wait" or "send a patch." <nixfreak_>yeah weird guix shell: error: emacs: unrecognized option <ieure>What command are you running? <ieure>The precise one you entered. <nixfreak_>guix shell emacs guile emacs-geiser --emacs -Q <ieure>emacs-geiser-guile, like I said before. <nixfreak_>guix shell emacs guile emacs-geiser-guile -- emacs -Q <ruther>you probably need both geiser and geiser guile, but maybe geiser is a propagated dep of geiser-guile, not sure about that <ieure>There is both emacs-geiser (stuff common to Geiser) and emacs-geiser-guile (the Guile-specific Geiser stuff). <ieure>`guix search geiser' if you have doubts about what's available. <nixfreak_>yeah I did , emacs-geiser doesn't work either ok .. ok I'll stop bothering you and do some research <ruther>nixfreak_: so you've tried both geiser guile and geiser, right? <nixfreak_>yeah guix shell emacs guile emacs-geiser-guile emacs-geiser -- emacs -Q , then mX: geiser-guile and doesn't even see geiser <ieure>nixfreak_, Ah, try -q instead of -Q <ieure>nixfreak_, See emacs --help for an explanation. <ieure>nixfreak_, `emacs --help' gives you a very succinct explanation of what they do. <nixfreak_>Ok , I see that now .. still confused though I would of thought install through guix would have its own .emacs file , separate from the actual system emacs ? <ruther>nixfreak_: emacs users .emacs.d by default, if you want separate configs, set the --init-directory to other directory <ruther>nixfreak_: using a container would work if you didn't share ~/.emacs.d, yes. But note that getting GUI programs running in a container is a bit more complicated, you will need to pass through sockets and display environment variables <nixfreak_>Ok , so when you install emacs using guix, guix doesn't create a separate emacs.d directory or init.el file <ruther>nixfreak_: guix isn't making separate environments or anything like that, it's like if you used nix on foreign distro, emacs from nix would also pick up ~/.emacs.d <simendsjo>Anyone using thunderbolt? After I disconnect the dock, the monitors isn't detected anymore. Tried three different docks, so it's somehow tied to my Guix setup. Can someone who have this working share their configuration? <futurile>simendsjo: thunderbolt authorization stuff? <nixfreak_>Wouldn't that have to do with the windows manager or compositor your using ? <simendsjo>futurile: It is authorized. When I first log in, the displays is detected and is working fine. When I disconnect the dock and reconnect it, the displays doesn't work. They state they cannot detect any signals. Restarting the thunderbolt kernel module will detect a single DP monitor of 1024x768, but this will mirror the screen on both these monitors. Tried a lot of "xset dpms force on" et.al. <futurile>simendsjo: oh weird! what are you using for desktop env? (hard to debug I guess) <simendsjo>futurile: StumpWM (thus X). My next step is to go through all logs when I first log in, and all logs when I reconnect the dock to see if I can spot anything strange. But I bet it's a lot of work. <nixfreak_>so not using X, but I do have to manaully put in the monitor requirements in the conf file otherwise it has a default which screws up workspaces. Does StumpWM have a way to add monitor configs ? <nixfreak_>In addition, StumpWM listens for XRandR events and re-configures the heads to match the new monitor configuration. Occasionally StumpWM will miss an XRandR event, use refresh-heads to synchronize the head configuration. <simendsjo>nixfreak_: Not to my knowledge. But I think the strange thing is that it works flawless on login, but then breaks completely when reconnecting. Tried restarting boltd and the thunderbolt kernel module without things getting back on track. I don't mind tweaking a bit, but having to completely log out is too much trouble. <simendsjo>StumpWM "always" miss the mark, so I run refresh-heads after all xrandr change. The problem is that xrandr doesn't see the monitors at all. The usb and minijack works though. <ngz>Is anyone able to stop evaluations on CI? For example, evaluation 2055748 is just wasting bandwidth and could be removed. <ruther>ieure: I think the reason guix pull --list-generations takes so long is because of the news, it checks for news between the commits, from the local checkout. It would even pull in the checkout if the commits weren't available <ieure>ruther, Interesting hypothesis! Like it's doing an O(n) search over channel news for every generation? <ieure>Tired of fighting Savannah, switched my stuff over to use the Codeberg mirror for real. <luca>more like thosr scrapers are fightinh savannah <the_tubular>Â I wonder how far this enshitification will take us ...