IRC channel logs

2020-05-10.log

back to list of logs

<cbaines>anadon, looking some more, just removing the header was suggested, and seemingly OK https://github.com/repology/repology-updater/issues/218#issuecomment-525966018
<cbaines>I wonder why that hasn't happened though. I'll send an email to the Guix issue...
<anadon>I just checked it and the incorrect header is still there
<anadon>It looks like once that header gets removed, then everything with repology will get fixed.
<bdju>Failure on tex again. I'm not actually sure I can update without removing packages or something... Not sure what's going on. Couldn't free more than 16GB doing GCs (where's the rest of the space being taken? / and /home are separate...), and that 16GB wasn't enough apparently.
<lle-bout>bricewge: problems with network on ci.guix.gnu.org solved for me
<rekado>the search on issues.guix.gnu.org yields poor results when searching for packages because Xapian splits words at hyphens
<rekado>so it doesn’t see “emacs-exwm-next” as one phrase
<rekado>you can find it with “emacs AND exwm AND next”
<rekado>or with “emacs NEAR exwm NEAR next”
<rekado>clearly that’s not great
<lfam>It's really annoying when using mu in general
<lfam>I find myself using the web-based interface to my email account for code searching
<rekado>I wonder why phrase search doesn’t yield the expected result
<PotentialUser-25>Hi. I'm getting a ghostscript.DRV error when I install guix. How do I resolve that?
***modula2 is now known as defaultxr
<lfam>PotentialUser-25: What is the error?
<PotentialUser-25>It says guix system: error: opening file ghostscript 9.27.DRV no such file or directory
<rekado>PotentialUser-25: can you tell us more about how you got there?
<PotentialUser-25>It happens after I chose either lxde or I3 to install on graphical install
<rekado>this sounds like a corrupted install medium
<rekado>have you verified the signature after downloading the installer image?
<rekado>this also catches a corrupt download
<PotentialUser-25>Yes. Should I redownload the install file then?
<lfam>No
<lfam>How did you create the install USB or DVD or whatever you are using?
<lle-bout>hey, any idea why I have buggy characters in Emacs like this? https://framapic.org/IbkD9tTH2Ymy/d1PeqemJksd9.png
<lle-bout>I use standard DejaVu-Sans-Mono 11 font
<PotentialUser-25>I followed the instructions on the manual
<lfam>Which ones? For USB? Or DVD? Give us something to help you with
<PotentialUser-25>I followed the one for usb
<lfam>Did you change the placeholder text "system" appropriately?
<lfam>And did you make sure to sync?
<PotentialUser-25>Yes I did sync
<PotentialUser-25>I ensured that the file I if and the location for of is correct
<lfam>I would make the USB again and try again
<leoprikler>lle-bout: you should have a monospace serif font there
<lfam>Sounds like it didn't get imaged fully
<PotentialUser-25>OK. Thanks.
<leoprikler>it's probably missing
<lle-bout>leoprikler: hmm, is that DejaVu-SansSerif-Mono ?
<leoprikler>in my emacs the default value for that is Monospace Serif
<leoprikler>you probably want DejaVu-Serif-...
<leoprikler>and the face you want to customize is named "Fixed Pitch Serif"
<leoprikler>btw. has anyone figured out, why Epiphany can't find fonts other than sans, serif and monospace?
<lle-bout>leoprikler: sorry had to restart emacs to try
<lle-bout>leoprikler: even with the default font I have that issue..
<lle-bout>it used to work before
<lle-bout>I don't understand what happened
<leoprikler>yeah, core-updates can break a lot
<lle-bout>I don't have core-updates, I'm on master
<leoprikler>so, what is the value of your Fixed Pitch Serif font rn?
<lle-bout>leoprikler: how can I check this?
<leoprikler>specifically the font family
<leoprikler>M-x describe-face
<lle-bout>Font: #<font-object -PfEd-DejaVu Sans Mono-normal-normal-normal-*-13-*-*-*-m-0-iso10646-1>
<lle-bout> Fontset: -PfEd-DejaVu Sans Mono-normal-normal-normal-*-13-*-*-*-m-0-fontset-startup
<lle-bout>it's default, apparently
<lle-bout>.emacs doesnt have anything about fonts right now
<lle-bout>I have spacemacs-dark theme if that helps
<leoprikler>Okay, do M-x customize-face RET fixed-pitch-serif RET
<leoprikler>Not sure if spacemacs sets that already, but user overrides always take priority over themes
<lfam>lle-bout: I think that the changes in core-updates did break fonts to some degree: https://debbugs.gnu.org/cgi/bugreport.cgi?bug=41132
<xelxebar>mbakke: Hm. So grafts are completely orthogonal to builds. I guess this makes sense, because if native-inputs could also be grafted, then there'd be a notion of grafted vs. non-grafted substitutes.
<lle-bout>lfam: but I don't use core-updates, I'm on master, and it's been a while I have this issue, few months probably, never bothered to solve it until now, I'll check leoprikler instructions wait
<lfam>Oh lle-bout. Core-updates was added to master yesterday
<lle-bout>leoprikler: it says Monospace Serif as font family, also 'STANDARD'
<leoprikler>okay, now try changing that to DejaVu Sans Mono
<leoprikler>then set for session
<lle-bout>leoprikler: that worked it seems, it shows the text now
<lle-bout>what exactly is a Serif, Sans or Mono font?
<leoprikler>Good, now change the "Sans" to "Serif"
<lle-bout>(btw poor irc user with nick 'mono' getting pinged all the time)
<lle-bout>leoprikler: that works too
<leoprikler>good, then set the latter for the future
<leoprikler>serif fonts have those thin lines everywhere, sans fonts don't
<lle-bout>leoprikler: cool! can I add back my font line in .emacs?
<leoprikler>emacs does that for you, if you click that button AFAIK
<leoprikler>I have the following preamble in my .emacs/init.el:
<leoprikler>(setq custom-file (locate-user-emacs-file "custom.el"))
<leoprikler>(load custom-file)
<lle-bout>leoprikler: okay thanks a lot! I don't understand Emacs fully yet but that helped me understand this part better.
<lle-bout>I've uncommented my modifications and then used the wizard to set fixed-pitch-serif family and it modified things nicely.
<lle-bout>well, looks like emacs-guix has a bug, you can't list services anymore
<bdju>
***catonano_ is now known as catonano
***jonsger1 is now known as jonsger
<lle-bout>hey, I have another question, how can I set default keyboard layout with ssdm,wayland,sway?
<lle-bout>sddm*
<wxie>hi
<ryanprior>Hey wxie :)
<wxie>ryanprior: Will command "sudo -E ./pre-inst-env guix-daemon --build-users-group=guixbuild" take long? There is no output for me after 10 hours.
<lle-bout>wxie: it's a service so it doesnt stop
<lle-bout>it listens on a socket
<lle-bout>put it in the background
<lle-bout>GNU Guix later uses it to manage builds
<wxie>lle-bout: ok. I will put it to the background. Thanks. Save me time.
<lle-bout>wxie: you can use & character after the command in a shell
<ryanprior>aaaah I hope you weren't waiting patiently for 10 hours to see it exit!
<ryanprior>Maybe we should make the guix-daemon print some happy little message on start-up?
<wxie>ryanprior: I just went to sleep, and it's still running in the morning.
<ryanprior>Okay that's good =D
<ryanprior>Seems like a start-up message on stderr would be friendly
<wxie>ryanprior: What's the [env] in the shell prompt? Is it important?
<ryanprior>That's telling you that you're inside a Guix environment (I see that when I use --container)
<raghavgururajan>Hello Guix!
<raghavgururajan>ryanprior: Hey, I wanted to convey my apologies. I thought I could help, but sort of occupied with some other things.
<wxie>ryanprior: Do I need stay in the container? Some commands are not working under [env].
<ryanprior>Hey raghavgururajan no worries, you haven't been keeping me waiting! I have lots of other things going as well :)
<ryanprior>wxie: you can go ahead and exit the container, if you need it again you can just create it again.
<raghavgururajan>ryanprior: I am glad to hear that.
<wxie>ryanprior: ok. Thanks. I now can begin my guix development journey.
<ryanprior>Generally you create a container (such as by using `guix environment --container -m manifest.scm` with a manifest file) to perform some specific task that requires the environment the container provides.
<raghavgururajan>Folks! What is the difference between (dbus-root-service-type) and (dbus-service)?
<ryanprior>The first time you create a given environment it takes a long time, so you might be cautious about doing it often- but every time after the first, it'll take under a second to create the environment container, because it's already done the work.
<wxie>Allright.
<ryanprior>wxie: so glad to hear that! If you want to pair on creating a guix package feel free to ping me, I enjoy helping people get started =D
<wxie>Glad to hear this, ryanprior .
<FennecCode>Hey, so I've been having a problem where I can't run binaries by doing "./binary". It'll claim that there's "No such file or directory". Does anyone have any ideas? 🤔
<ryanprior>FennecCode: does it work if you use the full path?
<lle-bout>FennecCode: is that an emoji at the end of your message? if so, then if anyone knows how to make Emacs ERC support emojis I'd love to make it work!
<ryanprior>lle-bout: https://github.com/iqbalansari/emacs-emojify/
<FennecCode>ryanprior: It doesn't work with the full path either 😕
<ryanprior>FennecCode: can you execute other binaries in other places? such as /usr/bin?
<FennecCode>lle-bout: Yes, and the package ryanprior posted is the one I use 😊
<lle-bout>FennecCode,ryanprior: thanks a lot
<FennecCode>I can execute binaries from packages I have installed. Like, I can execute binaries that are in "/home/fenneccode/.guix-profile/bin/"
<ryanprior>Okay that's a good sign. Where is the binary you're trying to execute?
<FennecCode>Under Documents somewhere. It's a C program I'm developing on locally.
<ryanprior>So like /home/fennec/Documents/mycprogram/bin/a.out ?
<FennecCode>Yea, exactly 🙂
<ryanprior>What do you get from `ls -l home/fennec/Documents/mycprogram/bin/` then?
<FennecCode>"-rwxr-xr-x 1 fennec users 80552 May 9 17:55 a.out"
<FennecCode>Oh, I'm a dingus. 😛 I think it's just a matter of it being a different GCC version or something. 😛 I recompiled now that I have my dev environment all set up, and now it's working fine 😛
<ryanprior>🚀
<ecbrown`>egg-cellent
<FennecCode>🥚
*ecbrown` records an observation: guix pull might profit from a bit of feedback when it goes to visit a substitute server (which may be down!)
<lle-bout>FennecCode: hey, can you post an emoji to try?
<FennecCode>🙀
<lle-bout>yay!
<FennecCode>🌵
<FennecCode>Cool! 😊
<lle-bout>thanks it works!
<lle-bout>but it uses images, not actual fonts?
<lle-bout>I wonder if you can use noto emojis?
<FennecCode>Dope! I really love emojis, so I'm glad to see people in this channel embrace them rather than "oh... we don't do that here" 😛
<lle-bout>emojis are standard UTF-8 so it doesnt bother me
<FennecCode>Exactly 😊
<lle-bout>but it looks like emacs-emojify is a bit gross in its implementation because it uses images and not fonts
<FennecCode>I think I tried to set it up once, but between being monochrome and squished they didn't really do it for me.
<lle-bout>how do you type emojis then?
<FennecCode>I think it's on the C end of Emacs to support color fonts at this point. I'm not sure what can be done besides embedding pictures in order to get emoji to work, unfortunately. 😕
<KE0VVT>OMG, I want color emoji in Emacs!
<FennecCode>I use a separate package called company-emoji that adds auto-complete based on shortcodes 🙂
<lle-bout>oh nice!
<FennecCode>Right! I would kill for color emojis in Emacs 😊
<lle-bout>how to enable by default emojify-mode in ERC?
<lle-bout>and company-emoji too?
<FennecCode>I have company enabled globally, and I just have it in my company backends list
<KE0VVT>And I wish Fedora's emoji helper still worked. It provided a viable replacement for the Windows+: universal emoji input box in Windows.
<FennecCode>`(add-to-list 'company-backends 'company-emoji)'
<FennecCode>And then I use global-emojify-mode for displaying the emojis 🙂
<lle-bout>FennecCode: OK! will try
<ryanprior>Railwaycat's fork of Emacs (https://github.com/railwaycat/homebrew-emacsmacport) has smooth kinetic scrolling with pixel-resolution and color emojis
<mroh>lle-bout: try something like (with-eval-after-load 'erc (add-hook 'erc-join-hook 'emojify-mode))
<FennecCode>Gosh, that would be awesome. 😊 It kills me when things don't support emoji shortcodes. I essentially have all of them memorized at this point, and I use them everywhere 😅🙂
<FennecCode>ryanprior: Ohhh~ 🤤
<lle-bout>FennecCode: how to enable company globally? I don't know much about company
<FennecCode>Too bad I'm not on Mac 😛
<FennecCode>`(global-company-mode 1)'
<lle-bout>mroh: thanks but I'll enable globally instead!
<KE0VVT>ryanprior: All that fork does is reverse the removal of Mac-only support for color emoji.
<ryanprior>Are you sure that's all it does? If so, why doesn't upstream support smooth pixel scrolling on Mac?
<KE0VVT>ryanprior: Support was removed upstream for color emoji, because it only worked in Mac, and they did not want Mac to have a better experience than free users.
<lle-bout>another thing that's been bothering me a lot, how to always have ERC use the whole buffer?
<KE0VVT>Don't want people to start saying "If you want color emoji in Emacs, install it on a Mac. It works there."
<KE0VVT>lle-bout: I thought it always uses the whole buffer.
<lle-bout>KE0VVT: well it's stuck at the top left end for me!
<ryanprior>Maintaining code in upstream Emacs takes effort, and if the effort only benefits users of non-free platforms, it's good for that to be undertaken in forks like Railwaycat's.
<KE0VVT>ryanprior: It took no effort to maintain it. It was an unintentional side-effect of macOS. They intentionally inhibited it, so Emacs would not work best on macOS.
<lle-bout>KE0VVT: GNU Emacs like many other GNU projects are political projects so it makes sense to make moves like this. Technology always has been political.
<KE0VVT>lle-bout: I was not speaking negatively of the action. I support it.
<ryanprior>That's frustrating & didn't know it was quite like that. Emoji is a strange, political feature, so much so that the desire for the latest emoji is one of the biggest motivations driving Android phone manufacturers to upgrade to the latest OS. (because Google won't release emoji updates for old OSes)
<lle-bout>KE0VVT: That's good. I just wanted to clarify for any lurker
<KE0VVT>lle-bout: I was just highlighting that it was a political decision, and the forkers are not doing anything clever.
<lle-bout>Emojis I think makes chatting more emotional and friendly, it puts more friendliness and makes people happier because they suddenly can express their feelings
<ryanprior>elementary OS has color emoji, including in its terminal. I wonder what it would take to roll that out into Emacs.
<lle-bout>It would be amazing to have flawless Emoji support everywhere in the GNU/Linux landscape
<apteryx>any ideas what this could be: ;;; WARNING: loading compiled file /run/current-system/profile/lib/guile/2.2/site-ccache/git/tag.go failed: ;; In procedure load-thunk-from-memory: incompatible bytecode kind; and it leads to an actual problem when attempting 'guix environment ...'
<apteryx>seems I need to update my system profile, as it currently runs Guile 2.2 which mismatches my user profile?
<lle-bout>hmm maybe, I remember seeing those errors without any impact
<lle-bout>:smile:
<apteryx>that's weird though that a different guile version in the system profile vs a user profile would cause failures... In my mind they should be independant of each other.
<apteryx>independent*
<ryanprior>If you use a pure environment they should not interact, but in an impure environment you could have some interaction.
<apteryx>guix environment --pure equally fails
<apteryx>it has to do with the GUILE_LOAD_PATH
<apteryx>it contains a mix of 2.2 and 3.0 Guile modules
<apteryx>and /run/current-system/profile/share/guile/site/2.2 is in there
<ryanprior>What's setting GUILE_LOAD_PATH? Anything other than the environment profile?
<apteryx>I think the system profile is merged with the user profile (I don't recall where this happens in the code).
<ryanprior>I used to have some path setting stuff in my .bashrc that was screwing up my pure environments, could it be something like that?
<apteryx>otherwise you wouldn't be able to user system installed applications
<ryanprior>Oh interesting, that sounds problematic.
<ryanprior>Why would you want to use system installed applicatoins in a pure environment though?
<apteryx>the failure occurs before the environment has a chance to be set up
<apteryx>it reminds me of the issue of having a single PYTHONPATH for Python 2 and Python 3 stuff -- the moment you make the mistake of mixing both in a single profile, all hell (may) break loose.
<KE0VVT>I want to play with Guix on the spare computer named scruffy, but I'm too tired to shlep it downstairs.
<raghavgururajan>Folks! What is the difference between (dbus-root-service-type) and (dbus-service)?
<apteryx>ah, the Guile 3 vs Guile 2 GUILE_LOAD_PATH issue can be fixed by just sourcing ~/.guix-profile/etc/profile
<apteryx>(or login out then back in)
<apteryx>well, the GUILE_LOAD_PATH is still a mix of Guile 2 and 3, but doesn't break things given that the guile 2.2 stuff appears at the end.
<lafrenierejm>In the middle of a new install of Guix System and I'm getting the error "users: unbound variable" for this config file: https://paste.mozilla.org/m1N2AhNU
<sneek>Welcome back lafrenierejm, you have 1 message!
<sneek>lafrenierejm, daviid says: get-keyword is defined/exported in/from (oop goops), so you may either (use-modules (oop goops)) or 'snarf it' if you don't want to use the all (oop goops) module, just bind the following to a 'local' name: (@ (oop goops) get-keyword)
<raghavgururajan>Is there a build-system for pascal applications?
<ryanprior>raghavgururajan: all the build systems in master are here: https://github.com/guix-mirror/guix/tree/master/guix/build-system
<ryanprior>Looks like no Pascal.
<lafrenierejm>What build system(s) does Pascal even use?
<lafrenierejm>Found the error in my config. There was an extra paren immediately following the file-systems argument.
<ryanprior>Darn those parens!
<lle-bout>well I gave up on display managers with sway and it works great!
<lle-bout>I was worried that ALT-F<n> could allow to espace swaylock but it doesnt!
<lle-bout>ALT-F<n> doesnt work at all after starting sway.
<lle-bout>Which is OK!
<lle-bout>One issue I see is that dmenu depends on X11, so I can't do `xwayland disable` just yet.
<lafrenierejm>lle-bout: Surely there's a wayland alternative to dmenu?
<bandali>they quit, but yes, there's https://hg.sr.ht/~scoopta/wofi
<lafrenierejm>Does anyone have experience using a RAID1 array for their EFI system partition?
<lafrenierejm>bandali: Yeah, I noticed they had dropped. :)
<lafrenierejm>Figured they might look for their nick in the logs, though.
<bandali>lafrenierejm, ah, true; i keep forgetting this channel is publicly logged :-)
<efraim>hmm, looks like there's a bug with efl-1.24 on guix :(
*rekado updates issues.guix.gnu.org with improved search and a few other fixes
<leoprikler>wow, I actually see issues on the start page again
***apteryx is now known as Guest2620
***apteryx_ is now known as apteryx
<efraim>well it wasn't fixed by removing ibus and glib support
<efraim>I guess I should try with rebuilding enlightenment with my modified efl
<rekado>huh?
<rekado>leoprikler: issues have been on the start page since a long time
<leoprikler>not for me at least
<rekado>(I still haven’t updated, actually. Still waiting for the database lock.)
<rekado>long time ago they used to be fetched asynchronously from debbugs, and that was slow
<rekado>since a few months the issue lists are prepared on the server, so every client gets them when they request the page.
<leoprikler>I remember being involved in a discussion about just that quite recently.
<leoprikler>maybe in April or so
<xelxebar>Putting together a texlive package, how does one decide between the names texlive-latex-foo vs texlive-foo? Is there a particular convention?
<rekado>the new convention is texlive-foo
<rekado>where foo is the name in the texlive.tlpdb
<rekado>leoprikler: this was changed on March 30: https://git.elephly.net/gitweb.cgi?p=software/mumi.git;a=commit;h=ed096349eefeff5d5571eceb853a4a5f6eda7260
*rekado re-indexes the mail database to update the search
<rekado>raghavgururajan: about issue #41138: I think the severity is not actually “important” (no more than most other bugs), and I preferred the old descriptive title.
<rekado>please open a separate bug for separate issues.
<raghavgururajan>rekado: Sure thing!
<efraim>rekado: not really a supported platform, but icedtea-1 builds with your bootstrap path on powerpc-linux :)
<rekado>woo, that’s great!
<bricewge>lle-bout: It seems to have bee resolved on my side too. But I guess it'll not be the end of it...
<raghavgururajan>bricewge: Just following up on #40994.
<bricewge>raghavgururajan: Working on wireguard ATM, will get to you later in the day
<raghavgururajan>bricewge: Cool!
*rekado tries to bring back fibers in mumi
<civodul>yay for the core-updates merge! \o/
<civodul>thanks mbakke, lfam, & everyone!
<rekado>issues.guix.gnu.org now shows “forgotten” issues on the front page.
<rekado>I hope this will help us reduce our patch review / bug triage backlog
<efraim>👍
<rekado>I also fixed phrase searching and changed the way xapian interprets the query, so search results should now be much more relevant
<civodul>rekado: woow, nice!
<civodul>it's smart to highlight forgotten issues
<civodul>we should link to it from https://guix.gnu.org/contribute/ and from the manual more prominently
<civodul>it used to be that the front page of issues. wasn't helpful, but now it gives a very good overview
*rekado still fiddles with the search
*janneke tries adding <hurd-menu-entry>
<mothacehe>hey guix!
<civodul>hey comrades!
<civodul>janneke: today i want to finally fix the vm-image issue that you found
<janneke>civodul: great! i played with your "debug-diff" and "found" something possibly interesting before i gave up/paused
<civodul>ah!
<janneke>i changed #$linux => #+linux because for hurd, we cannot run in a cross vm; we use a native linux vm
<civodul>ok
<janneke>(an alternative would be to add expression->derivation-in-hurd-vm
<civodul>that rings a bell
<janneke>unless we change that, we'll try to build linux for i586-pc-gnu
<janneke>...BUT together with that change, i also changed $loader => $+loader
<bricewge>rekado: That's a great feature! The search result being in a chronological order is also helps greatly to bring up forgotten issues.
<janneke>and ...here i'm losing the trail... it could be that we must only have parts of loader native, and others cross?
<jonsger>does someone know whats wrong with git fetch over https:
<jonsger> git fetch origin
<jonsger>fatal: unable to find remote helper for 'https
<cbaines>jonsger, I think I ran in to that recently. Installing openssl possibly helped.
<jonsger>not for me
<cbaines>jonsger, hmm, maybe nss-certs as well. Make sure that the environment of your shell is up to date with your profile.
<jonsger>cbaines: how can I do that?
<cbaines>source ~/.guix-profile/etc/profile will do it
<efraim>I have a reproducer for the glibc-bootstrap-system-2.2.5.patch not found
<efraim>guile -c '((use-modules (guix scripts environment)) (guix-environment "--ad-hoc" "fontconfig" "--" "fc-cache" "-frv"))'
<jonsger>cbaines: thx
<cbaines>jonsger, did it worK?
<efraim>guile-git on powerpc-linux is failing one of the tests
<efraim>testing it now with guile2.2
<jonsger>cbaines: not really
***jonsger1 is now known as jonsger
<efraim>failed with guile2.2 also
<jonsger>its working over git:// and ssh://
<olivuser>hello guix!
<olivuser>I have a general question, is it possible to be logged in, on several machines, as the same user?
<pinoaffe>olivuser: could you be more specific?
<olivuser>pinoaffe, I am logged in as olivuser on my desktop pc. I am going to leave said desktop pc to go outside, and would like to continue using that nickname when I am on my phone's irc client
<pinoaffe>olivuser: with IRC that is not possible out of the box, some users opt to use some slight modifications / postfixes, for example olivuser_
<pinoaffe>others host one central client (or "bouncer") and connect to it from several devices
<olivuser>when you mean host, do you mean that it is necessary to have one machine running at all times?
<pinoaffe>olivuser: pretty much
<olivuser>so if I am interested in having something like this "bouncer" would be a good keyword to start searching the net
<pinoaffe>I used to run weechat on a VM and connect to it from a local device, but nowadays I use matrix, which supports "bridging" irc to matrix
<dutchie>matrix is probably the easiest way. if you have an always-on machine, znc is a popular choice (and what I use)
<olivuser>dutchie, okay nice. I plan to use a raspi as a server, so I could run znc on that machine as well
<dutchie>though matrix is its own chat service that can access IRC, not a specifically designed bouncer
<olivuser>pinoaffe, that sounds nice but I dont understand how matrix can act as a bouncer
<dutchie>yes, that'd work, and i'm sure lots of people do it that way
<dutchie>olivuser: it has a bridge to all of freenode
<dutchie>so you join #freenode_#guix:matrix.org (iirc that's the right format) and it creates an irc user for you
<olivuser>on every machine?
<dutchie>from your individual matrix account
<pinoaffe>olivuser: matrix is a federated chat service of its own that allows representing external chats inside itself, a program that does that is called a bridge and acts as a bouncer for you
<olivuser>hm. sorry but I dont understand why, in matrix, I would not have to have several accounts, which I prefer to avoid.
<pinoaffe>olivuser: because the bridge acts as a single IRC client, and relays all messages to/from matrix
<olivuser>and the bridge is the same across devices?
<pinoaffe>olivuser: the bridge is run centrally, on a server
<olivuser>pinoaffe, so the baseline situation is the same as it would be for a bouncer, right?
<olivuser>that is, I would, in both cases, have to have one central machine that is always on?
<olivuser>(sorry if that had been written some lines earlier...)
<pinoaffe>olivuser: or you would need access to a bridge that is always on
<pinoaffe>there are a couple of publicly accessible ones
<olivuser>pinoaffe, and if I were to have "my own" bridge, I could use said bridge for all bridging purposes related to matrix, right?
<olivuser>(though in theory such possibility should be existent for publicly accessible ones as well...)
<pinoaffe>olivuser: for all irc<->matrix bridging purposes, there are other bridges for bridging to telegram, whatsapp, etc
<olivuser>pinoaffe, yeah thats why I ask, but would that not be possible as well if I had "my own" bridge-hosting device?
<olivuser>I mean, could I not host all bridges I desire if I host one bridge for irc?
<pinoaffe>olivuser: yes
<olivuser>pinoaffe, thanks a bunch, that has helped me a fair bit. last question: I reckon that if I go on the matrix webpage, I am going to find good resources to get me started on how to set stuff up, right?
<pinoaffe>yes
<olivuser>pinoaffe, thanks a lot for your time :)
<jonsger>rekado: oh thanks for your mailto: link improvement on issues!
<efraim>I use quassel and run quasselcore on a server and connect using quasselclient or quasseldroid from PCs or my phone
***janneke_ is now known as janneke
<lafrenierejm> https://paste.mozilla.org/z0Y80nhq "grub-installl... exited with status 1"
<lafrenierejm>^ Getting that error when attempting a fresh install of Guix System. any ideas?
<cbaines>lafrenierejm, is that from a command you ran, if so, which command?
<cbaines>ah, I see the efibootmgr command is failing... I don't really know much about EFI
<lafrenierejm>cbaines: It's occuring when running `guix system init /mnt/etc/config.scm /mnt`.
<ecbrown>lafrenierejm: any chance you are on an macbook and selected the wrong boot mode when holding down Option
<lafrenierejm>ecbrown: Nah, this is on a desktop.
<ecbrown>ok, sometimes i see this when i boot to the wrong mode, need UEFI instead of BIOS
<ecbrown>(legacy)
<lafrenierejm>The main weirdness I'm introducing is attempting to use a RAID1 array across two (identical) 256M partitions on two identical disks.
<lafrenierejm>That is, I'm attempting to use a RAID1 array for the EFI system partition.
<ecbrown>oh
<ecbrown>i've always seen that split into two, and that stuff on the first disk only
<ecbrown>i.e. two partitions
<ecbrown>sda2 participates in raid
<lafrenierejm>ecbrown: Yeah, that's what I'm doing now.
<lafrenierejm>ecbrown: Ostensibly it _can_ be done if one sticks the RAID array information at the end of the partition (which I did), but it's definitely going to be an uncommon case.
<ecbrown>lafrenierejm: about a week ago i said fuck it and went zfs
<ecbrown>whatever gets it to boot, then do everything in zfs-land
<lafrenierejm>ecbrown: Are you using zfs for your root, then?
<ecbrown>yes
<lafrenierejm>Nice! How are you liking it?
<ecbrown>it is killer
<lafrenierejm>Mind sharing your system config?
<ecbrown>it's a free install of a system that ought not be recommended (a foreign distro)
<ecbrown>today i'm going to test out Guix' zfs on a proper Guix system
<lafrenierejm>Ah. Okay.
<lafrenierejm>Best of luck!
<ecbrown>i have a suspicion that somewhere i have to load a kernel module, and not sure yet how to do that (not investigated in documentation yet)
<ecbrown>you too, cheers
<lafrenierejm>ecbrown: There should be a `kernel-arguments` argument in `operating-system` that should be useful.
<lafrenierejm>I should be in chat most of the day EDT. Would be interested to hear how it goes. :)
<ecbrown>cool, thanks. will certainly be crowing when it works :-)
<raghavgururajan>ecbrown: We have new service called "kernel-module-loader-service-type". Check it out at http://guix.gnu.org/manual/en/html_node/Linux-Services.html#Kernel-Module-Loader-Service
<ecbrown>raghavgururajan: thanks for the tip! i will check it out for sure.
<raghavgururajan>ecbrown: Cool!
<fishinthecalcula>Hi everybody! Do you think it would be safe to serve a small PHP application through Apache using the store's path as webroot?
<raghavgururajan>fishinthecalcula, I do not know about safety, but it's better *not to* mess with store unnecessarily. May use read-only symlinks inside you homedir.
<bricewge>ecbrown : If you want to use ZFS as root on Guix System it won't work. I discussed with lonzo1 about this a few days ago http://logs.guix.gnu.org/guix/2020-05-09.log#083448
<pkill9>i don't see anything wrong with it, as long as nothing needs to be written
<pkill9>and if something does need to be written, it'll jus tfail
<pkill9>just fail*
<raghavgururajan>Unless executed as root?
<pkill9>nah, root can't write to the store either
<pkill9>it gets remounted as read-only
<raghavgururajan>Wow, I did not know that.
<raghavgururajan>Wait what
<raghavgururajan>I did try nuking the store. Also, some folks have manually edited the store.
<pkill9>raghavgururajan: it gets mounted read-only on guix system, maybe those people were using it on a foreign distro?
<raghavgururajan>There is no super above superuser (root)?
<ecbrown>bricewge: thanks for the link. i will check it out. no, just want to test zfs package in guix.
<pkill9>no there isn't
<pkill9>you could get around the store being mounted read only
<pkill9>not sure specifically, but you can mount it again as read-write or something
<bricewge><pkill9 "you could get around the store b"> ecbrown : It's broken you need to remove the line with DEPMOD
<raghavgururajan>I am confused. If its super-readonly, then how guix writes on it?
<raghavgururajan>Maximum privilege is root.
<ecbrown>bricewge: ok, i had suggested that "kmod" be included to efraim
<pkill9>the guix daemon is started without the store being mounted read-only, then I think there's a service that mounts it read-only
<pkill9>so all users including root 'see' only a readonly version of the store
<pkill9>but the guix daemon 'sees' the original 'real' version of it being read/write
<pkill9>although since you can restart the guix daemon safely, I'm not sure
<raghavgururajan>Just checked gnu/store's ownership info. User: roor and Group: guixbuild.
*raghavgururajan is tempted to manually edit/delete a file inside the store
<pkill9>if you run `mount`, you'll see "/dev/<disk> on /gnu/store type ext4 (ro,relatime)"
<pkill9>i assume you just run "umount /gnu/store" as root if you want to manually write to it
<pkill9>but your Guix manufacturer's warranty will be void if you do that
<raghavgururajan>Haha.
*ecbrown recollects hacking the installer script to install to existing /gnu mount point
<raghavgururajan>Not gonna do that on my daily driver.
<pkill9>:P
<bricewge>ecbrown Yes probably all package using modprobe should get it from /proc/sys/kernel instead of kmod.
<raghavgururajan>`sudo rm -r /gnu/store`
<buffet>ecbrown, yeah, i hate how it just dies when /gnu exists
<raghavgururajan>That is my dream command.
<buffet>like /gnu/store, sure, but /gnu?
<buffet>also, i got yet another issue (sorry) im trying to `guix pull` as a user, but i get `guix pull: error: mkdir: Permission denied`. anything i can do?
<pkill9>buffet: try running `strace -o /tmp/guix-pull.log guix pull`, it might show you what directory it's trying to create
<ecbrown>usually i see that stuff when i screw up e.g. ~/.cache/guix permissions when running sudo
<buffet>~.cache/guix/checkouts huh
<buffet>well
<buffet>owned by root yeah
<buffet>seems to work now
<buffet>thansk
<ecbrown>no problem
<pkill9>yea if you run `sudo -E guix ...`, it uses your user's ~/.cache/guix and so ends up setting ownership of that directory, or the subdirectory it's working iwth, to root
<buffet>alright i see
<rndd>hi everyone! i want to install openjdk-10 but package openjdk installs version 12. how to specify version 10?
<cbaines>rndd, guix install openjdk@10 should do it
<rndd>cbaines: oh thanks, where find info about this syntax?
<cbaines>rndd, there's some documentation here https://guix.gnu.org/manual/en/html_node/Invoking-guix-package.html
<cbaines>rndd, specifically he bit about -i package
<rndd>cbaines: thanks
*rekado updated issues.guix.gnu.org to mumi 0.0.1: no more debbugs lookups!
<civodul>rekado: oh, so everything's stored locally and there's an async rsync?
<rekado>yes
<rekado>I haven’t been able to bring back fibers yet, because I’m not sure how to make it work with guile-sqlite
<rekado>but that would be the next big performance improvement
*pkill9 want sto add a gitea service
<raghavgururajan>nckx: Is this fingerprint (F5BC 5534 C36F 0087 B39D 36EF 1C9D C4FE B9DB 7C4B) correct?
<cbaines>rekado, I've been learning more about the fibers web server and sqlite
<cbaines>at least not how to do it :)
<cbaines>well, and a few things to make it somewhat work
<civodul>rekado: it works well already performance wise
<raghavgururajan>rekado: Nice! It saves lot of time. :-)
<jonsger>rekado: thanks for all your work on mumi, berlin and the years as maintainer!
<raghavgururajan>pkill9 +1
<raghavgururajan>If installed applications wants to use dbus, should I use (service dbus-root-service-type) or (dbus-service)?
<pkill9>this is a sick joke: the Gitea source code is hosted on Github
<lafrenierejm>pkill9: lol
<pkill9>oh, they are working on it: https://github.com/go-gitea/gitea/issues/1029
<raghavgururajan>I aware that the latter is included in %desktop-services, but I currently have %base-services with other stuffs and would like to enable dbus.
<raghavgururajan>pkill9: Hahah!
<raghavgururajan>pkill9: They gonna be meta-giteaed!
<raghavgururajan>*They're
<lafrenierejm>Does anyone know why `mdadm ... /dev/md/root /dev/sda2 /dev/sdb2` from a Guix System USB installer would not result in any devices in /dev/md?
<joshuaBPMan>well it looks like I was a little slow in packaging sway1.4
<joshuaBPMan>Looks like someone beat me to it.
<bricewge>raghavgururajan: Just extend your existing dbus-root-service-type with simple-service: “(simple-service 'dconf-dbus dbus-root-service-type (list dconf))”
<raghavgururajan>bricewge: Thanks! But why dconf?
<bricewge>raghavgururajan: It's just a line from my config, I need it for virt-manager to save it's preference.
<raghavgururajan>Oh, got it, never mind.
<raghavgururajan>Cool!
<janneke>hmm, where is our /proc/config.gz ?
<jonsger>has someone an example how to modify network-manager in %desktop-services
<apteryx>jonsger: here I extend the base network-manager-service-type config to add the openvpn plugin: https://paste.debian.net/1146008/
<jonsger>apteryx: I'm more interested how to "append"/add it to the (services)
<jonsger>because network-manager is already in %desktop-services, this makes problems
<lfam>Sounds like a cookbook recipe in progress!
<jonsger> https://paste.opensuse.org/view/raw/51635386 doesnt work
<reepca>jonsger: ahh, I see your problem. modify-services returns a modified list of services
<apteryx>jonsger: you put my snippet in a (modify-services %desktop-services ...)
<reepca>which in that paste then gets cons'ed onto %desktop-services
<jonsger>without modify-services i get: "guix system: error: =>: bad use of '=>' syntactic keyword"
<reepca>so the result is that your services list looks like (dnsmasq i3lock ... set-xorg-configuration (desktop-service1 desktop-service2 ... modified-service ... desktop-service-n) desktop-service1 desktop-service2 ... desktop-service-n)
<reepca>you need to delete the final %desktop-services
<reepca>that way everything up till the (modify-services ...) gets consed onto the modified list of services
<jonsger>so modify-services is actual a modify-and-return-services function
<reepca>in scheme most things are functional by default; if it actually mutated the list it'd be called modify-services!
<reepca>but I can see why that would be confusing
<apteryx>lfam: I like the idea of adding such snippets to the cookbook
<jonsger>ah. tbh. I found all this packages/services stuff in config.scm very boilerplatish
<lfam>I think we should give a lot of examples of service customization, because it's really too hard for new users who don't know Scheme, but I think most users want to customize at least one thing
<lfam>And, we should make sure to keep them up to date like the rest of the manual
<apteryx>jonsger: I'm curious, why? It seems concise and to the point to me. But I understand it takes some time to tame new syntax.
<reepca>I guess it could be called "modified-services", but then it could be mistaken for a variable or field name
<apteryx>anyone familiar with Perl? I'm loosing hairs over why a C module it builds doesn't properly link with zlib
<apteryx>(packaging perl-perlio-gzip)
<reepca>rekado: out of curiosity, are your issues with using guile-sqlite3 with fibers related to blocking?
<apteryx>it just doesn't link at all to it (althoug I can coax 'perl Makefile.PL' into finding zlib, by providing the 'LIBS=-L/gnu/store...-zlib/lib/libz.so' variable).
<jonsger>apteryx: https://paste.opensuse.org/view/raw/85894519 would be lean and user focused
<apteryx>looks like you want wisp syntax (a pythonesque way to write scheme)
<apteryx>jonsger: I think not having to specify a configuration object is how it used to be (we'd work with services directly instead of service-types). I don't recall what the rational was to go to the route that was chosen, but I can guess uniformity and composability were pluses.
<jonsger>apteryx: my approach would more to leave the complex config system like it is and make an easy language or so which gets translated then
<apteryx>sounds like a fun project to learn Scheme macros
<reepca>jonsger: one difficulty with an implicit modify-services like that is that there are some service-types where it's fine to have multiple instances (non-extensible ones, in particular). How can it tell whether the user wants to modify the configuration of something that's in %desktop-services or create another instance of it?
<apteryx>ah... -L must be used to specify the base path of something to look for using -l. It works now.
<apteryx>seems MakeMaker tries to be smart but doesn't look in LIBRARY_PATH
<rekado>what do y’all think about extending modify-services to allow deletion and addition of services?
<rekado>much like modify-phases
<efraim>It'd be nicer than passing around the remove snippet and bring it more inline with modify-phases
<rekado>one could argue that it’s “wrong”, but then modify-phases would also be wrong and we should go back to acons’ing phases
<rekado>I really don’t like the remove snippet
<rekado>it looks like there’s a lot of potential there, but … all we do is delete one matching service.
<rekado>the power of lambda! … oh, we’re done.
<reepca>does deletion delete one instance or all instances?
<rekado>uhm, … yes
<rekado>:) I don’t know
<rekado>what does our “remove” snippet do?
<reepca>SRFI-1 remove removes all matching elements
<rekado>to me the only problem with and extended modify-services is that currently we don’t have any verbs
<rekado>we match on service types directly
<rekado>there’s no “change”, so we can’t just add “add” or “delete”.
<reepca>so a strange thing just happened
<reepca>I tried running lsblk and got this message: "bash: /home/reepca/.guix-profile/bin/lsblk: No such file or directory"
<reepca>I then ran 'which lsblk' and it reported /run/current-system/profile/bin/lsblk
<reepca>according to guix package --list-generations I've never had util-linux installed in my profile
<reepca>I have no idea why bash tried executing it there
<rekado>did it only happen once?
<rekado>did “type lsblk” show the same location?
<reepca>'type lsblk' => "lsblk is hashed (/home/reepca/.guix-profile/bin/lsblk)"
<reepca>does --list-generations consider propagated-inputs?
<reepca>huh, seems the last update switched it to a util-linux:lib dependency.
<reepca>and that, folks, is why you hash -f after upgrading
<reepca>I was just a bit freaked out because I had just run 'sudo rm -rf /var/cache/fontconfig' and gotten a "it's busy" error message, and was wondering if it was bind-mounted to the store or something
<lfam>Font trouble?
<reepca>aye, went to look at the graph in the "Service Composition" node of the manual, and the names in the rendered graph nodes are all boxes :-/
<lfam>Since the core-updates merge?
<lfam>I reported a bug about font problems I noticed since then but it was a bad bug report :( Unclear and confusing
<reepca>I'm updated to that point, and I don't remember it happening before then
<reepca>I hadn't run fc-cache at that point, and I'm not sure if emacs caches the rendered image
<lfam>It seemes to me that the hard-coded dependency of fontconfig on gs-fonts is not working anymore
<lfam>s/seemes/seems
<reepca>time to spin up another emacs to check I guess
<mbakke>but there were no changes to fontconfig, nor gs-fonts
<lfam>I know, and yet it doesn't work
<lfam>So maybe it never worked, and we should remove the dependency
<lfam>Either way, something changed
<calher>OK, got scruffy working totally. Now, to download and write Guix image.
<reepca>can confirm it still happens in a freshly-opened emacs
<lfam>reepca: Are you on Guix System? Or another distro?
<reepca>lfam: Guix System
<BlackMug>Hello
<BlackMug>im trying to install guix as offline mode , but i cant skip that in the installation process
<BlackMug>any work around?
<mbakke>BlackMug: the installer requires internet access because it does not contain everything needed for offline installation
<mbakke>would be nice to have a "jumbo" image that could be used offline, but alas
<BlackMug>well yeah thats just sad not having this already hehe
<reepca>if you can hook up the disk you want it installed on to a device that has guix running on it (can be foreign distro), you can use guix init directly from there
<BlackMug>you mean running guix with different distro?
<reepca>aye. For example, my current install of Guix System was installed from guix running on Ubuntu. I just plugged in the second hard drive, wrote a config.scm, and used 'guix init' to install to it.
<BlackMug>ah i see, i know i can do that or running it in a VM but would be nice that Guix having offline installation so as IP,Gateway... configurations before fully installation
<BlackMug>specially for static IPs/Proxied, you cant get internet without manually configuring that
<reepca>hm, it should be possible to ctrl+alt+f2 in the installer to hop over to a terminal where you can set that up
<BlackMug>ah then thats great , same as linux files right? resolv.config ... or something different in need to read
<reepca>good question, I don't actually know.
<BlackMug>np
<BlackMug>wish guix has complete iso same as other gnu/linux distros
<BlackMug>anyway thank you for the help will come when new versions appears
<BlackMug>hope this can be addressed as an issue
<BlackMug>cya!
<mbakke>BlackMug: can you file a bug about it so it does not get forgotten?
<reepca>I wonder how much space it would take for the transitive sources of every single package we have...
<GLAT-agent>Hi.
*GLAT-agent is selling cheap GNU/Linux licenses.
<GLAT-agent>Will you add glat-client package to Guix?
<KE0VVT>GLAT-agent: What does GLAT stand for? In Esperanto, GLAT is the word for GLBT.
<drakonis> https://notabug.org/GLAT/howtotell KE0VVT
<GLAT-agent>KE0VVT: GNU/Linux Activation Technologies
<GLAT-agent>Yes, NotABug admin defaced our repositories
<GLAT-agent>$99 for single computer license
<KE0VVT>GLAT-agent: To be honest, it would be nice if GNU/Linux users really did give $99 to the project whenever they installed.
<GLAT-agent>$49 for each copy for 25 or more computers
<GLAT-agent>We are also selling Key Server licenses
<GLAT-agent>Most of users here are running non-genuine GNU/Linux copies.
<civodul>GLAT-agent: i think this is the wrong channel
<GLAT-agent>civodul: why?
<civodul>this channel is about Guix, please be on topic
<GLAT-agent>civodul: if you want, I will send you Certificate of Authenticity with GuixSD logo
<GLAT-agent>civodul: but you have to purchase a license first
<bandali>GLAT-agent, please take your trolling elsewhere
<KE0VVT>Oh, I thought this was another channel. Sorry for humoring this troll.
<GLAT-agent>> Will you add glat-client package to Guix?
***ChanServ sets mode: +o civodul
***GLAT-agent was kicked by civodul (Kicked by civodul)
<KE0VVT>Grandan dankon.
<civodul>nedankinde ;-)
<reepca>wait, I just realized that the graph with the box-characters in the manual was a png, not a pdf. So the rendering issue must have happened when it was generated.
<civodul>janneke: i just reported a bug for profiles vs. build hooks vs. cross-compilation
<civodul>it's annoying
<civodul>reepca: which rendering issue?
<reepca>civodul: 8.17.1 "Service Composition" in the manual
<reepca>in fact all the graph images have boxes instead of characters
<civodul>you mean the on-line copy? https://guix.gnu.org/manual/en/html_node/Service-Composition.html
<reepca>no, the online copy works fine. But the one I've got has boxes.
<civodul>or Info?
<reepca>info, aye
<civodul>ah
<civodul>oooh indeed
<dftxbs3e>jonsger, efraim, nckx: btw here's a complete explanation of the 128-bit double problem on POWER : https://www.talospace.com/2018/12/the-saga-of-power-isa-128-bit-long.html
<civodul>reepca: should check the log of guix-manual.rdv
<reepca>I can't seem to find any *.rdv files in the repository
<ngz>Hello. I'm trying to update X-Moto package, but I'm encountering a SDL error (<https://paste.debian.net/1146055/>). In particular, I use sdl-union, and the configure script properly finds it, but the error is related to a standalone SDL-ttf library. Do you have any idea about how to solve this?
<jetomit>ngz: the tennix package has a workaround in the 'fix-include phase that might be relevant
<jonsger>dftxbs3e: thx I have read that a while ago
<jonsger>but to my understanding the problem for ppc64le is our to old bootstrap gcc compiler and not the glibc itself
<jonsger>on glibc master seems to quite some activity regarding 128-bit floats on power9 :)
<ngz>jetomit: Thank you. I know that workaround… I wrote it :) But I'm not sure we can use CPATH anymore. Besides, there seems to be a deeper problem, in sdl-union.
<civodul>reepca: like so: guix gc -R $(guix gc --derivers $(readlink -f ~/.config/guix/current))|grep guix-manual.drv
<civodul> https://ci.guix.gnu.org/log/qs44rw1qf3a8q4x2gpli66z3gpinxiwa-guix-manual
<civodul>hmm
<jetomit>ngz: oh ok :)
<reepca>ah... I think I should have realized s/rdv/drv a lot sooner
<civodul>oops, sorry
<dftxbs3e>how could one track where exactly this kind of error is happening? "ice-9/eval.scm:293:34: Wrong type to apply: #<package gcc-cross-boot0-wrapped@7.5.0 gnu/packages/commencement.scm:3328 3fff7acaf630>"
<dftxbs3e>the line doesnt help
<dftxbs3e>that's during GNU Guix's compile process
<dftxbs3e>it would be very constructive for any of the most experience GNU Guix maintainers to make a live stream of them hacking on GNU Guix just to gather their workflow and usage of tools
<dftxbs3e>live stream or recording
<rekado>dftxbs3e: what did you do to get this error?
<dftxbs3e>rekado, I did nothing in particuliar, this is a result of a complex git merge
<dftxbs3e>particular *
<rekado>what command gave you the error?
<dftxbs3e>"that's during GNU Guix's compile process" - so "make"
<rekado>ok
<rekado>I’m guessing that something expects a monadic procedure but you somehow managed to give it a package value.
<rekado>is there any more output?
<rekado>and what’s the resulting diff?
<dftxbs3e>rekado, there's no more output
<dftxbs3e>how could you tell: "something expects a monadic procedure but you somehow managed to give it a package value" - from that error?
<dftxbs3e>And right, I can try diff-ing with upstream master, that's good strategy'