IRC channel logs


back to list of logs

<nckx>vagrantc: ‘Up to’ is great. ‘Up to 100% bug-free.’
<vagrantc>nckx: as opposed to previously being 86% :P
<Guest28>Does guix system image not work with (file-system (type 'btrfs))?  If I look at the image with gparted, it says that it is ext4
<bjc>oh, poetry is known to be broken
<nckx>Guest28: It completely replaces the file-systems field, so yes (it successfully ignores it! :) and no (it won't be btrfs).
<nckx>ACTION away, g'nite.
<Guest28>nckx: Ah okay.  So I would need to install the system and afterwards reconfigure the system?
<Guest28>sneek: later tell nckx  Ah okay.  So I would need to install the system and afterwards reconfigure the system?
<sneek>Got it.
<Guest28>Good Night
<bjc>is anyone using emacs with guix-managed treesitter libraries? its unable to find the .so files for me. i assume i have to tweak some setting to tell it where to look, but i can't figure out how
<bjc>d'oh. as soon as i say that, i see the variable in NEWS
<yewscion>Hey Guix, I've recently set up a new box with Debian as the main distribution, and I am unable to use many host programs—including firefox, gnome-settings, and the gnome desktop environment in general(all icons disappear, as does the desktop background and the cursor)—once I've installed any Guix program that sets $GDK_PIXBUF_MODULE_FILE in the profile. I think it's related to On most of
<yewscion> my machines, I solely use GNU Guix programs regardless of the base distro (though most of the time I use Guix as the distro entirely). But on this one, I want to be able to share the machine with people as well as make some use of Debian's apt packages. Is this a known issue, or did I happen to do something wrong somewhere?
<apteryx>anyone interested in a telephony or qt team?
<lissobone>yewscion: Firefox is not exactly Free Software, that's why it's not included in Guix. But you can install GNU IceCat, a Free version!
<singpolyma>Firefox is Free Software but it is not FSDG compatible
<yewscion>lissobone, singpolyma: I appreciate that, and use GNU Icecat as my browser of choice. However, what I'm trying to do is to /not/ break Debian by installing a package in my profile: Stock packages on foreign distros shouldn't be negatively impacted by the use of Guix.
<lissobone>The only way it might mess up a foreign distro, I think, is by messing up environment variables.
<yewscion>Indeed. In this case, specifically, it is messing up $GDK_PIXBUF_MODULE_FILE, which Debian (as it is a rather conservative distro, update-wise) needs to find a dynamic library Guix no longer ships.
<yewscion>Upstream, the functionality is apparently now linked statically (which is great for the future, but a breaking change currently).
<yewscion>If You have a machine with a foreign distro handy, You can see what I am talking about by installing gdk-pixbuf in Your profile and logging out and back in: Gnome can't find icons or backgrounds, gnome-settings and tweaks cannot start, and other applications—those installed as stock on Debian, not any I have a particular attachment to, but still—will also fail to load if they use that library.
<yewscion>Unfortunately, this is /also/ the case if You install any package that sets that env variable, not just the library itself.
<lilyp>yewscion: why should you set $GDK_PIXBUF_MODULE_FILE in the profile? Is there some weird propagation or are you installing things that you'd normally just want in a shell?
<yewscion>lilyp: I am not trying to set it; It was set automatically by a package I had installed.
<lilyp>which one?
<dcunit3d>has anyone had any recent problems with the git-repo package? i'm needing to set up a python virtualenv (based on the guix python3 package) in order to use git-repo on my guix machine.
<dcunit3d>i'm not sure if it's been reported anywhere yet. it's likely that either something in the package was recently bumped or needs to be bumped. i mainly use git-repo to sync large collections of git repositories, the XML for which I collect using graphql
<HiltonChain[m]>podiki: I have some patches relevant to mesa-updates (I can find them in the output of "guix graph mesa"), can you have a look at them? (They are #64228 #64637 #64638 #64639)
<yewscion>lilyp: pitivi is one such package. I was silly when debugging this on my machine, and did not keep logs. Sorry for the delay.
<lilyp>yewsicon yep, that's a bug, see inkscape for how things should be done
<yewscion>lilyp: Alright, thanks for the confirmation! I'll read through the definition for inkscape, and if no one else submits the patch once I have one ready I'll submit one.
<HiltonChain[m]>I got "error: tcc: unbound variable" when adding "#:use-module (gnu packages engineering)" to (gnu packages wm) and then invoking "make", can anyone reproduce it?
<lilyp>HiltonChain[m]: looks like you're trying to create a circle
<HiltonChain[m]>lilyp: Looks like engineering->gnome->music->wm ...
<devmsv_web>good mirning everyone is anyone experiencing problems with emacs-next an eshell tramp? I can't cd /ssh:user@host:~/ and ru any probram
<devmsv_web>this is what I get:
<lilyp>without having seen the paste, there was a similar complaint recently
<lilyp>can you time-machine back to 29.0.91 and see if it still happens there?
<lilyp>if not, file a bug; probably one of our substitute*s broke
<ncf>does the guix module system support sum types? where can i find documentation about it?
<ncf>(by guix module system i mean whatever you call the analogue of nixos' module system)
<ncf> it even typed
<mfg[m]>ncf: not sure what the module system of nixos is (i don't know nixos) but the modules in guix are regular guile modules, so you may find the information you seek in the guile scheme manual in the section about modules?
<devmsv_web>I have filled a bug report as this is happening also with an "old" guix system with emacs 29.0.91 which I don't use much.  In emacs 28.2 everything works right.
<devmsv_web>lilyp: would it help if I do a git bisect building package with guix from emacs 29 to emacs 28.2?
<ncf>mfg[m]: so a nixos module would define a configuration option like with a type like string; this option would be reflected in the manual along with its type, and when the user tries to set = 3 they would get an explicit type error. is that not how guix does things?
<hiecaq[m]>Hey Guix, in the guix-devel list "A Forum for Guix Users" thread I read about the proposal of bridging the IRC to Matrix, but aren't we already bridged? I'm sending this message from Matrix.
<hiecaq[m]>That thread makes me confused.
<devmsv_web>hiecaq[m] I read you, so I guess yes, there is a Matrix bridge
<hiecaq[m]>devmsv_web Good to hear. I thought the bridge had been broken.
<mfg[m]>Yes, does bridge matrix to IRC. That is true for every IRC channel there
<mfg[m]>As I understand the thread the problem is more that mailing lists are old technology and most newcomers aren't used to it
<hiecaq[m]>Yeah, I can see the point, but I quite like mailing lists, because they gives us the freedom of choosing the front-end. For example I read them in mu4e, and do some quick filtering to skip those that I'm not really interested.
<mfg[m]>That is an advantage yes, but also a disadvantage. If you are new and don't know what you want, what should you choose then? A forum has a static structure where you don't have to choose that much.
<mfg[m]>If anything at all
<zamfofex>A potential solution would be to have a Web interface to the mailing list that allows posting.
<mfg[m]>yeah, the reply feature of mumi is deactivated because of bugs iirc
<minima>hi, i'm using emacs+geiser to edit some guix files from a repository checkout (where i've previously ran configure+make, so all .scm files have their respective .go companions); i've followed the docs and added the checkout folder to geiser-guile-load-path; still M-. (geiser-edit-symbol-at-point) and C-c C-d C-d (geiser-doc-symbol-at-point) don't seem to work as expected
<minima>M-. seems to redirect me to the object definition only when the definition is in the current file
<minima>C-c C-d C-d doesn't seem to find any docstring at all
<minima>also, i'd love this to work with all guile symbols/variables, not only with guix objects - do i have to add anything else to my geiser-guile-load-path?
<jpoiret>janneke: Did you *have* to update the guix package twice?
<jpoiret>i believe it shouldn't be needed anymore because of current-guix
<HiltonChain[m]>minima: You may try guix-devel-mode from the package emacs-guix, it provides a function guix-devel-use-module (or "C-c . u" in the mode).
<HiltonChain[m]>Evaling the define-module part with geiser-eval-last-sexp (or "C-x C-e") also works.
<minima>hi HiltonChain[m] thanks, trying that now
<minima>hm geiser-eval-last-exp against the module gives me an error, so perhaps that's the root of my problems
<HiltonChain[m]>What's the error message?
<minima>so the workflow should be: setting things up as per the docs, then opening a guix source file, making sure that geiser guile gets started, then 'C-x C-e'-ing the define-module part at the top - if i get it correctly
<janneke>jpoiret: it's not strictly necessary, but it's very handy until we have guix pull actually working
<janneke>guix system init bare-hurd.scm /hurd => gives us previous guix
<minima>HiltonChain[m]: here's the error:
<minima>i've been trying to edit a random file from the checkout, just to see whether any of the go-to-definition functions works
<minima>maybe i need to start emacs from within the guix development environment... (i.e. the guix shell with the various dev dependencies)
<HiltonChain[m]>I think that's not necessary...
<HiltonChain[m]>Can you jump to a symbol within the module? (e.g. The first exported symbol in gnu/packages.scm is search-patch, so try "M-." on the #:export line?)
<minima>yeah, that works indeed!
<minima>oh wait, now things seem to work as expected
<minima>C-c C-d C-d works for guile symbols too, apparently
<minima>and M-. works across files in the guix checkout
<minima>(i.e. i'm being redirected to the file where a given symbol is defined)
<minima>so that seems to be all i need... now i need to see how to make this happen consistently
<minima>i suppose the guix-devel-use-module was the key...
<minima>i'll play with this a bit more but many thanks HiltonChain[m]!!
<jpoiret>janneke: yes but why twice?
<jpoiret>oh, you mean you run that from the install image directly, without locally building guix on the vm?
<jpoiret>well, not install image but rather image
<janneke>jpoiret: yes, that's the use case
<janneke>of course you can clone guix and use ./pre-inst-env ...
<janneke>but it was handy to just run `guix build ...' in a fresh vm/image
<nckx>hiecaq[m]: The Matrix-to-Libera bridge will be disabled before the end of the month. The room will be bridged to this channel but this hasn't yet happened due to bugs and a complete lack of support on the EMS side.
<sneek>Welcome back nckx, you have 1 message!
<sneek>nckx, Guest28 says: Ah okay.  So I would need to install the system and afterwards reconfigure the system?
<nckx>podiki[m]: Hi! Why does 09e73683 top-level import (guix build utils)?
<nckx>ACTION builds.
<podiki[m]>nckx: that was the Godot update? Probably a mistake, I saw a warning after pulling later too. I can take a look later if you don't want to make the fix first
<nckx>I'm going to make sure it doesn't break ‘guix pull’ or whatever and will. Thanks!
<podiki[m]>Hilton Chain: thanks, a quick look at the updates seem good, might have to be on next mesa update as I hope we can push the current one with substitutes very soon. If we do another rebuild I'll try to include
<podiki[m]>nckx: thanks!
<HiltonChain[m]>Thank you!
<podiki[m]>Thank you for the patches!
<nckx>Thank you both!
<HiltonChain[m]>Hi nckx, can you apply #63847 to core-updates? :)
<nckx>Looks like it.
<nckx>HiltonChain[m]: In future, please write patches for the intended branch, not master.
<HiltonChain[m]>Thank you! I didn't realize that, sorry...
<nckx>It just means it'll have to wait until this evening :)
<ulfvonbelow>anyone noticed that python-pyside-6 seems to be missing its egg info stuff? When any other python package requires it it complains that it doesn't find any candidate versions.
<ngz>sneek: later tell rekado Do you remember why "artistic2" license in texlive importer is bound to gpl3+ instead of artistic2.0?
<sneek>Got it.
<HiltonChain[m]>Could #64361 be reviewed as well? It's a trivial change but fixes a reported build failure.
<mirai>HiltonChain[m]: "Honor tests? flag" is what's I tend to see but other than the bikeshed, LGTM
<mirai>unrelated, but core-updates isn't a thing anymore
<mirai>or did it get reinstated?
<nckx>I heard something to that effect as well.
<nckx>Weird that it's still mentioned in the manual, twice.
<nckx>If it's really dead, I shall remove these references.
<mirai>nckx: do cherrypick your patch out before removing it though
<nckx>I meant the manual references, not (yet) the branch.
<nckx>By the way, we seem to be leaving behind a *lot* of dangling objects on now. I guess this is due to more large rebases? They've bumped the git gc timer from two weeks to daily(!). If this breaks anything, it's probably a bug on our side, but be mindful.
<geri>just curious - has anyone managed/bothered to make their guix home/system config async/use multiple cpu threads/whatever it's called properly?
<mirai>geri: wdym?
<geri>like one of the pros of funcitonal programming is that it's eas(y/ier) to use multiple cpu cores at the same time
<geri>i wonder if anyone has actually done that in guix
<mirai>you mean make guix (the program) use multiple threads?
<ulfvonbelow>the configuration-specifying code is usually not exactly resource intensive - it's the process of building / substituting / computing derivations that takes some time
<geri>ulfvonbelow: i feel like it too, but it'd be fun to hear if anyone actually has done it
<geri>mirai: that's also an interesting idea
<geri>it probably does already tho
<geri>for building at least
<ulfvonbelow>builds can happen in parallel in accordance with --max-jobs and the number of build users available. I'm not sure whether any attempt at parallelizing shoveling derivations to the daemon has been done
<geri>aighty, thanks
<ulfvonbelow>especially since AFAIK we currently only use a single store connection at a time
<nckx>matrixbridge: Why you no work >:(
<geri>cause it's saturday
<geri>why would it work
<nckx>‘Could not request link. Error M_AS_UNKNOWN: An internal error occurred’
<nckx>Ya borken bot.
<geri>why does this phrase sound so funny
<nckx>matrixbridge: Hellooo.
<nckx>matrixbridge: botsnack.
<nckx>I could do this all day but I'll stop. The problem is, there's no official word on this ‘internal error’, and the only rumour floating around is ‘I retried eleventy times and it worked!’ which is probably doing wonders for the system load.
<nckx>It's pretty laggy but that might just be Element being Element.
<geri>i love consistency of software :)
<oldfashionedcow>nckx: try a diff client maybe
<oldfashionedcow>i remember i liked nheko whyen i tried matrix
<nckx>Will it show the hosted extensions that does? I guess I'll find out.
<oldfashionedcow>i think so :P
<nckx>(The other common response is ‘I gave up trying to use the EMS-hosted one and spun up my own’, which (1) hahaha no (2) the only on in Guix appears to be matterbridge, and I don't want to write a service today.)
<oldfashionedcow>nckx: oh you are hosting your own?
<nckx>I am not.
<nckx>I would like to keep it that way.
<nckx>I'm not well-versed in matrix, but I like the presumed discoverability that gives us, FWIW.
<nckx>That said, if anyone's a Matrix expert, please enlighten me as to what best to do.
<nckx>ACTION hides the cluehammer, already has a headache.
<oldfashionedcow>nckx: whats the main issue with the bridge?
<nckx>The current one?
<nckx>Or the one I'm trying to enable?
<nckx>If the former, it's going bye-bye:
<nckx>If the latter, I have no idea, the above ‘error’ ‘message’ is all I get.
<nckx>If ‘why not host our own’: we're not historically great at sysadministry as it is, and I would much prefer that the ‘help it is broke’ cries that have come through the Matrix bridge since its inception not become our problem.
<itd>nckx: ask, maybe? (looks like that worked for )
<nckx>A search of ‘public rooms’ for that name returns no results?
<AndrewYu> gives me the UltraProprietary vibe...
<nckx>And people say IRC is complicated.
<nckx>Both as a user and an admin I'm not a fan, but I'd like to capture the significant audience it brings here.
<nckx>nheko say: [2023-07-15 18:34:29.534] [db] [error] Restoring secret 'nheko.47DEQpj8HBSa+/TImW+5JCeuQeRkm5NMpJWZG3hSuFU=.pickle_secret' failed (7): The name org.freedesktop.secrets was not provided by any .service files
<nckx>ACTION did mention the headache already, right?
<zamfofex>IRC is fairly simple, but tools and clients surrounding IRC have been tradionally “scary” and made it difficult use for some people.
<nckx>Installing keepassxc as suggested makes no difference. C'mon people.
<zamfofex>AndrewYu: The only thing missing is a “pricing” link. 😅
<itd>nckx: enabled secrets in the settings already?
<nckx>I've never used it, so no.
<nckx>Didn't help, but felt good.
<nckx>Something else (promoting appservice on a whim, totally undocumented) did seem to help though.
<nckx>nckx`: Are you my mummy?
<nckx>I didn't get my own message but both ends are getting flooded with users from the other.
<nckx>So that's… good.
<nckx>ACTION welcomes our redpilled overlords.
<zamfofex>I feel like I have yet to hear a single positive thing about Matrix! 😄 But it does seem to be working.
<nckx>I still don't like it. Thanks to everyone who helped me log into nheko, though. I didn't and won't use it, but its slightly different UI gave me the appservice idea.
<mfg[m]>Well, i actually enjoy using matrix. I'm mostly using the cli client gomuks, which works good enough.
<mfg[m]>and with all the bridges i can use literally every other service there :)
<mfg[m]>it was annoying to setup tho, took me a weekend
<AndrewYu>weechat-matrix looovvvesss crashing randomly
<nckx>I'm missing a HexChat-tier graphical client. Nheko ain't it.
<nckx>(The standard answer to ‘I would like my chats not to consume a CPU core and gig of RAM’ seems to be ‘TUI clients exists’, which isn't great.)
<mfg[m]>what against tui clients?
<nckx>Too few pixels.
<nckx>Also, monospace fonts.
<mfg[m]>yeah tru, good point. I guess i'm used to that 👀
<nckx>It's a valid preference, just not mine.
<nckx>So mfg[m]'s message show up in the #guix IRC channel, but my replies don't make it to
<nckx>…but then how did mfg[m] see my messages, mfg[m]?
<mfg[m]>there is a channel on I didn't know that
<mfg[m]>it's actually
<nckx>I just bridged to the IRC channel.
<mfg[m]>so libera already bridges to matrix
<nckx>Not for long.
<zamfofex>I think they are removing that feature.
<nckx`>ACTION sighs.
<zamfofex>I can see the “Test.” and “sigh”.
<nckx>I'm surprised this isn't bigger news, but then I'm rather biased:
<nckx>zamfofex: That's my Matrix-native account. Messages sent here, on IRC, don't arrive.
<AndrewYu>Were there wallops for it?
<nckx>Dunno, I wasn't very active until recently.
<nckx>I'll wait until the extremely rate-limited joins are complete, then try again.
<mirai>Someone™ needs to write _and maintain_ a HexChat-like client (in Guile)
<Guest28>Does browserpass-native require any special treatment?  It I installed the extension and the host native package but it says "no such native application"
<mfg[m]>does guile even have good bindings to a windowing toolkit like gtk?
<nckx`> chatters: #guix IRC messages are being delivered here, but Matrix messages aren't making it back there. So this room is going to be confusing at least until the Matrix <-> IRC sync is complete. Hopefully, at that point, everything will just work.™
<nckx`>ACTION .oO Famous last words…
<mirai>mfg[m]: I think there's GTK bindings for guile
<mfg[m]>Hm, i can't find anything but this:
<mfg[m]>and this seems to be abandoned for a long time
<zamfofex>mfg[m]: There is a link at the top of that page to
<mfg[m]>Classic, just reading the first two sentences and not the third...
<Guest28>There is also g-golf
<nckx>Hi Guest28. I didn't respond to your sneek message because I didn't understand the full context.
<mfg[m]>Ah yes, g-golf seems to be the most recent and actually maintained version.
<Guest28>nckx: Sure, no problem.  Context:
<nckx>Sure, but I don't understand the wider context/goal implicit in your ‘so’.
<Guest28>Basically I wanted to use BTRFS on my Raspberry Pi 4.
<nckx>But then I don't use VM images, really.
<nckx>Reconfiguring won't reformat anything, if that's what you hoped.
<nckx>It'll just fail to boot if you force it to mount as btrfs.
<Guest28>Yes, that is what I thought.  Well, how would I use BTRFS on my Pi?  There is no arm installation image
<nckx>ACTION doesn't know, is piless.
<Guest28>Sometimes I question myself if a Intel NUC would be so much simpler to use as a low power server
<nckx>OK, my IRC-sent messages are arriving on the Matrix side now. I think we're set.
<nckx>Not the same price class, but almost certainly.
<nckx>Then again, people keep offering me free *Pis, they apparently grow on trees, so I understand the need.
<zamfofex>nckx: I don’t think Guest28 wants to reconfigure the system, but rather generate a system image anew.
<Guest28>Free Pis? Those are really expense nowadays
<nckx>zamfofex: Oh, maybe. Only ‘install the system and afterwards reconfigure the system’ did not read that way to me.
<nckx>ACTION should have accepted them then.
<zamfofex>nckx: Was that not a question? It seems the original goal (from the log linked) is to generate a btrfs image, and they were wondering if the solution would involve reconfiguring.
<Guest28>Okay, so it would be possible to use BTRFS instead of ext4 for my Pi?
<lfam>Building linux-libre 6.4 on bordeaux for arm64 is failing consistently with these errors:
<sneek>lfam, you have 2 messages!
<sneek>lfam, rekado says: right above that error is “hexdump: No such file or directory”. Could it be that it tries to extract a number using hexdump, fails to do so, so there’s an empty variable somewhere, and so it passes the wrong arguments to “truncate”, so that the file name is in the position where a number would be expected?
<sneek>lfam, rekado says: truncate would be called with “-s SIZE FILENAME”, so my guess is that SIZE is the empty string due to an earlier error.
<lfam>Any ideas?
<lfam>A reply to my question
<Guest28>technically it is possible to convert ext4 to btrfs but I want it to directly install on a fresh btrfs.
<zamfofex>Guest28: Can you share your system configuration?
<Guest28> it is basically this
<lfam>Guest28: Have you tried changing the type of the filesystem mounted on "/" to "btrfs" ?
<lfam>Oh, I see the earlier context that there is some issue with `guix system image`
<lfam>There's currently no way to use `guix system image` and not get an ext4 filesystem
<Guest28>Yes, and the thing is that the Pi is an ARM CPU.  For ARM is no installation image and that is the reason I need to create a system image to flash it to the SD card.  I guess my only option is to convert ext4 to btrfs but on the other hand I am not really experienced in Guix and still a beginner.
<lfam>Yes, that's the simplest option
<nckx>What I've done on (full, UEFI, PC-style) aarch64 machines is boot another OS like Debian, then install Guix from there using the standard manual procedure.
<pjals>isnt there a thing in matrix to do exactly this but with a more client-integrated ui?
<nckx>I don't know enough about Alternative Bootloaders to know if this is an option here.
<pjals>yea, tombstone should work well enough for this
<nckx>But I don't want to do that yet.
<nckx>It's pretty permanent.
<oldfashionedcow>okay guix is TRULY facisnating
<oldfashionedcow>I sort of want to try implementing something similar into portage
<nckx>(Also, IRC peeps won't have a clue what this is about. :)
<oldfashionedcow>nckx: :D i'm an IRC peep
<nckx>B—b—but USE flags.
<lfam>Guest28: Note that if you do convert your ext4 filesystem to btrfs, you'll also have to adjust your config.scm accordingly
<nckx>oldfashionedcow: We are the best peeps.
<Guest28>nckx: Why is there an at symbol before your name? That wasn't the case yesterday.
<nckx>I have ascended, but thanks for the reminder.
<nckx>Guest28: ‘@’ denotes a ‘channel op[erator]’. Basically an admin.
<oldfashionedcow>voice means that when the channel admin "moderates" the channel, you can still talk
<oldfashionedcow>you can see info about this with
<Guest28>Ah. Wondered why the at symbol is omitted on mentioning your name.
<oldfashionedcow>/msg chanserv flags #guix
<nckx>Libera culture is not to walk around with your badge out, but I needed Matrix to respect my authority to do all this bridge stuff.
<nckx>It has nothing to do with the ‘@bob: Hi!’ addressing convention in some other places.
<oldfashionedcow>nckx: :D well see here is the thing. Atm I have a powerful desktop (dual 16 core xeon's and 256 gb of ecc ddr4 ram) which most of the time i actually use ssh'd with things such as tmux, vim and the like. I am FIERCLY loyal to gentoo (like extremly lmao), but I think if i look for a laptop that is actually libre, I might just use gnu guix on my laptop
<oldfashionedcow>truly facinating
<zamfofex>oldfashionedcow: What did you find interesting about it?
<oldfashionedcow>zamfofex: It seems like nixos, which I like as a concept, but implemented "correctly". Also I am so intrigued by the compiler magic
<pjals>btw, is my nick still pjals?
<pjals>(or pjals[m])
<oldfashionedcow>that thing there
<oldfashionedcow>pjals: pjals!
<nckx>oldfashionedcow: I used Gentoo for many years. It was nice.
<nckx>Then I cheated on it with Exherbo and now look where I am.
<Guest28>oldfashionedcow: How long does it take four your machine to compile gcc or llvm?
<oldfashionedcow>Guest28: here lemme grab the times
<oldfashionedcow>note this is with lto
<oldfashionedcow>(flto, i use clang)
<oldfashionedcow>Guest28: 2023-06-11T03:02:17 >>> sys-devel/llvm-16.0.5: 10′42″
<oldfashionedcow>10 minutes
<Guest28>This takes 5 hrs on my machine
<nckx>I was slowly reinventing NixOS (poorly; in bash) so as soon as I found out NixOS actually existed, I was lost forever.
<oldfashionedcow>Guest28: LMAO
<oldfashionedcow>nckx: that is one of the greatest things i've read today#
<Guest28>it is an old i7 from 2014.  only 4 cores and 8 logically :D
<Guest28>is it 16 logical or physically? what clock rate run they
<oldfashionedcow>Guest28: so its actually 16 logical per cpu
<oldfashionedcow>I compile with -j64
<Guest28>It is a workstation or server computer isn't it?
<oldfashionedcow>yea a workstation
<oldfashionedcow>though you can rack mount ti
<Guest28>Are you not afraid of your electricity bill?
<oldfashionedcow>Guest28: :P not covered by me ;)
<oldfashionedcow>also it doesn't use too much power
<oldfashionedcow>And its not brand new either, its an hp z640
<oldfashionedcow>I bought it used for about £700
<oldfashionedcow>took quite a bit of saving
<mfg[m]>haha i still have a z800 here :D
<oldfashionedcow>mfg[m]: niceeeeeee!
<mfg[m]>i mean yeah it's really old and gets pretty loud, but hey it still works
<Guest28>I want a workstation like that, too :(
<oldfashionedcow>mfg[m]: yup. Hope to rock this thing for at least 5-6 years :P
<oldfashionedcow>Though I am saving up for a laptop, bit hard to do when I can't get a "real job" :P
<nckx>AFAICS most fake ones pay better anyway.
<oldfashionedcow>:P not having tax helps
<Guest28>Do you have a GPU in it? I wonder how could games run on it.
<oldfashionedcow>Guest28: i do, a radeon pro wx7100
<oldfashionedcow>picked it up for just under 50 quid
<oldfashionedcow>gets the job done
<Guest28>8GB VRAM isn't even that bad
<oldfashionedcow>yea, really happy with it for the price
<oldfashionedcow>i'm personally not too much of a gamer, so can't comment on that myself
<Guest28>Well, having that much cores isn't useful for gaming since most games use max. 4 anyways.  So higher clock rate would be better and always wondered how that would look as of experience.
<oldfashionedcow>yea, if there are any benchmarks you want me to try just lemme know
<oldfashionedcow>it mostly helps for compiling, which is what i do most of the time :P
<Guest28>Well, you need 10 minutes for llvm.  This is more than enough for me.  Your probably take 10 minutes only for webkit, too.
<oldfashionedcow>Guest28: when i used webit it was about 25 iirc#
<Guest28>Those take at least 5 hours or so on mine.  That is one of the reason I switched.
<Guest28>Basically updating the system meant my PC was unusable for anything for the next 10 hours
<oldfashionedcow>:p yea, thank god for distcc for my laptop
<nckx>Guix doesn't support distcc granularity, but it can offload whole builds to a different host running Guix.
<oldfashionedcow>ahh that is intersting
<nckx>(Also, no ccache, sorroz.)
<Guest28>Do you crush big Rust projects as well with that machine?
<Guest28>(I mean generally, real big Rust projects doesn't exist yet AFAIK)
<oldfashionedcow>Guest28: yea pretty much - let me get the times for the rust source
<oldfashionedcow>(i compiled it from source for fun)
<oldfashionedcow>Guest28: 2023-06-17T23:31:16 >>> dev-lang/rust-1.69.0-r1: 20′05″
<oldfashionedcow>takes 20 minutes to compile rust from source
<Guest28>Who cares about optimization if you can just compensate it with power
<oldfashionedcow>(not me using -O3 and flto ;) )
<oldfashionedcow>thats my make.conf
<pjals>oldfashionedcow: mail me your computer :)
<Guest28>That is the thing I like about hardware.  They don't really degrade.  So you can buy a 10k$ machine 5 years later for around 1k$ and you have enough power for some years (only the inefficiency compared to newer hardware bothers me)
<Irvise_>GNUtoo: hi! I will send an email to the devel ML tomorrow regarding the addition of gcc-ada as a binary blob. Hopefully there will be no issues, fingers crossed :)
<nckx>Out of curiosity, what's the size of the blob?
<Irvise_>nckx: if the ones provided by Alire are accepted, then about 250MB
<Irvise_>But those come with plenty of extras. The Debian binaries are noticeably smaller. Lemme get their size
<nckx>The GNAT situation is so sad.
<meologism>installed a new guix vm, nothing in /run/user, shepherd wont start and on first login script is complaining about XDG_RUNTIME_DIR
<meologism>what am I doing wrong
<nckx>meologism: Could you share your system configuration?
<Irvise_>Just the GNAT part in Debian is 18 MB, 95 decompressed. Source:
<meologism>nckx: sure. I want to understand though, is /run/user/xxx supposed to be owned by system shepherd or the user one?
<Irvise_>nckx: we are aware of the issue. I provide a lot more info in my email with context and explanations. A small group of people are working towards a bootstrapping path (I am learning Scheme just for this).
<nckx>meologism: I thought it was created by elogind.
<nckx>Irvise_: Awesome.
<meologism>nckx: oh thats probably because i removed desktop services then
<nckx>Yep, that's where it usually comes from.
<nckx>If you want to trim it down you'll have to provide some commonly expected services ‘by hand’.
<nckx>(You can refer to gnu/services/desktop.scm for a list.)
<nckx>Irvise_: Do you speak Ada?
<GNUtoo>Irvise_: thanks
<Irvise_>nckx: a bit :) I do not work much with it. Only for hobby projects centered arround embedded development. But I would say anybody that reads Ada a couple of times can say they "speak it" x)
<nckx>I added ada-ed a long time ago in a naive hope that $omeone skilled might be able to use it to bootstrap GNAT. I still don't know whether that's a foolish thought.
<nckx>(It's Ada 83, not 95.)
<Irvise_>nckx: I talk about it in my email ;) Long story short, no, it cannot be used.
<nckx>I might have borked my e-mail server at the moment ( ͡° ͜ʖ ͡°)
<nckx>Which list?
<Irvise_>I will send it tomorrow, it is too late to proof read it now :P
<nckx>I look forward to reading it and despairing.
<Irvise_>I do ensure despair. That is what I focus on, as there is no other alternative than adding a binary for the time being.
<GNUtoo>Haskell also has a binary if I recall well
<nckx>Which is a gigabyte unpacked, by the way, which is really nice.
<Irvise_>It comes with a C/C++ compiler to be fair. And a lot of src :)
<nckx>Yeah, the fact that it was 68M compressed stopped me from swearing. Whatever's in there, it's low entropy.
<Irvise_>Indeed :)
<patched[m]>Seems like appservice kicks you from the new matrix room if you leave the old one
<nckx>Define new and old?
<nckx>(Since no new rooms have been created.)
<nckx>Old =, new =
<nckx>It's probably because, at a certain level of the bridge's abstraction, they are the same room. I assume this will stop when EMS removes the portalling.
<nckx>One reason I'd rather let EMS nuke their portals, rather than risk making my own mistakes :)
<nckx>What if I tombstone, and this kicks everyone out of for whatever reason? Etc.
<podiki[m]>hello from matrix (again), will this infinite loop if i'm still bridged....
<nckx>I don't think so.
<nckx>But can you try leaving just to test if it kicks you?
<podiki[m]>did not but that would have
<nckx>You just missed: But can you try leaving just to test if it kicks you?
<podiki[m]>thanks for whatever you did nckx !
<nckx>Or maybe the Matrix history caught it, I have no idea how any of this.
<nckx>No idea how any of this at all.
<podiki[m]>i did just leave the libera room, after seeing a message i sent from here go through
<nckx>podiki[m]: Oh dear. So I'm guessing you did get kicked?
<podiki[m]>err...was kicked by appservice but rejoined now
<podiki[m]>let's see, since this time i'm not in libera chat when i joined here
<nckx[m]>ACTION sighs.
<podiki[m]>ACTION flinches at every incoming packet
<nckx[m]>ACTION has adjusted the room name again :)
<nckx>podiki[m]: You're *not*? OK…
<podiki[m]>i was in libera room (via matrix) when I first joined the matrix room, even though I left soon after; maybe that's why i was kicked?
<podiki[m]>so far so good now...
<nckx>I'm still not sure what exactly patched[m] meant, but now has a warning *not* to leave, because whatever.
<mfg[m]>i also switched the room to
<mfg[m]>so far nothing strange happened to me
<nckx>ACTION already has a concussion because they did the TikTok speedboat challenge, doesn't really need this too.
<podiki[m]>well if one gets kicked should only rejoin the matrix room if on matrix, seems okay now
<nckx>I'll take your word for it.
<oldfashionedcow> * | nckx already has a concussion because they did the TikTok speedboat challenge, doesn't really need this too.
<nckx>Don't drink and water-ski.
<oldfashionedcow>can't drink yet :P
<oldfashionedcow>nckx: how was water-sking
<oldfashionedcow>that sounds fun
<podiki[m]>well a few messages haven't come through from irc to matrix....either slow or issues just like the other bridge :(
<oldfashionedcow>podiki[m]: most likely slow, based off my experience with matrix
<podiki[m]>slow is one thing, but dropped/out of order messages (if they show) is another
<podiki[m]>really disappointing with all this tech and can't reliably send/receive basic text at the very least
<nckx>oldfashionedcow: Er, well, great fun as always right until I face-slapped the water something silly…? The problem is I got straight back on the horse, which in retrospect I shouldn't have. Anyway, enough off-topic chats from me.
<nckx>podiki[m]: I also noticed a >1 second lag. I was hoping it was just the Web UI being slow. ☹
<podiki[m]>it has dropped 3 messages in last few minutes based on our log
<podiki[m]>on my end at least
<oldfashionedcow>folks what do you say i try guix on a pi
<oldfashionedcow>already done gentoo
<oldfashionedcow>so why not guix
<nckx>podiki[m]: What's an example of a message that didn't arrive?
<oldfashionedcow>nckx: *imagine if it was that one right there ;)*
<oldfashionedcow>but tbh i think it might be specific users
<nckx>Byzantine chats.
<oldfashionedcow>as i remember when i used matrix, i just couldnt see some user's messages sometimes
<oldfashionedcow>never got to the bottom of it
<nckx>podiki[m]: I don't see a difference between the two Matrix rooms, so it doesn't appear to be the new bridge…?
<nckx>oldfashionedcow: Oof. I'm starting to understand why Matrix love is low.
<Guest28>I only see issues with the IRC and Matrix bridge.  Why is that even a thing.  Seems like it does more trouble than be useful.
<nckx>Not splitting the community is nominally a very good thing.
<oldfashionedcow>^ yea i agree with nckx
<nckx>Both sides are sizeable.
<nckx>We can't just dump one.
<oldfashionedcow>nckx: I gave it a pretty fair chance - I did genuinly try it, and there some things I really liked. For once discord with matrix was perfection almost - not having to use discord's crappy js client was *amazing*. But, the bridge's can be very hacky/dodgy, and matrix "rooms" is a problem as tbh matrix is mostly designed FOR bridges
<oldfashionedcow>I loved the nheko client, it was pretty sweet
<oldfashionedcow> sway on guix too?!
<oldfashionedcow>okay this has *really* got me excited
<nckx>ACTION uses Guix/Sway.
<podiki[m]>nckx: when you said "i'll take your word for it" for instance never appeared for me
<oldfashionedcow>nckx: oh how is it?
<oldfashionedcow>i love sway
<podiki[m]>i use matrix since i only have a few direct matrix rooms/people I use, but bridge all chats to use one client (whatsapp, signal, discord, irc)
<Guest28>Hmm, understandable.  What is the core issue with the bridge problems?
<podiki[m]>in my time here i've never seen a rhyme/reason to dropped messages, just random and happened more recently with the bridge getting worse. we'll have to see on this end now
<nckx>Hmm, I was opped when I sent that… surely not…
<oldfashionedcow>okay so practically 99% of the software i use is there
<nckx>ACTION does an opped thing.
<oldfashionedcow>the only one missing is librewolf, but i like to keep gui apps as flatpaks anyway
<nckx>Oppity oppity lalala.
<podiki[m]>Guest28: the bridge doesn't work well, went down a lot, and caused issues on matrix and libera people managing
<nckx>Nope, that's not it.
<podiki[m]>another missing message for me was "LAMOOOOOO" from oldfashionedcow
<oldfashionedcow>podiki[m]: i think the sheer power behind that emotion was too great for matrix ;)
<Guest28>podiki[m]: But both parties (IRC and Matrix) work together to make that bridge work? I don't understand if both support that bridge that it has issues
<podiki[m]>both were soon after i rejoined after being kicked, after then seems okay
<nckx>podiki[m]: But I did see a ‘reply’ or whatever Matrix calls it for that, so wasn't sure if it was actually missing or just translated.
<oldfashionedcow>podiki[m] | both were soon after i rejoined after being kicked, after then seems okay
<oldfashionedcow>yep that would be it
<oldfashionedcow>I had that issue myself
<nckx>Guest28: As Libera tells it (and has been lamenting for many months), there was no ‘and’.
<nckx>Just EMS pointing the hose at Libera and mostly looking the other way.
<nckx>(EMS represents ‘Matrix’ here.)
<oldfashionedcow>Folks I just noticed I'm a lot more "free" than i thought
<Guest28>Ah I see.  Wha/Wh is EMS?
<nckx>Element Matrix Services.
<oldfashionedcow>looking at my licences, I only use 3 pieces of unfree software
<podiki[m]>"I had that issue myself" from oldfashionedcow didn't appear over here too
<oldfashionedcow>packettracer, which i don't use anymore; intel-microcode for cpu and linux-firmware
<podiki[m]>i do run my own matrix server, maybe that is another variable though shouldn't matter
<nckx>Now, I'm biased towards Libera, but I also trust the people who were lamenting.
<Guest28>But doesn't Matrix live through bridges? No one wants to replace their working solution but with a bridge you can create backwards compatibility.
<mfg[m]>i also run my own matrix server it's always a bit slower and takes long to laod some things
<nckx>podiki[m]: If is just too unreliable, running our own bridge is another option. Just not one I look forward to.
<mfg[m]>but synapse is just slow software
<civodul>ACTION had an annoying static networking failure:
<civodul>anyone else using it?
<podiki[m]>agree bridges is a great feature of matrix, having interop with other protocols is really important for these things
<podiki[m]>nckx: the question is: is it the hosting or the underlying (bridge) software?
<podiki[m]>i suppose i can produce logs of this room on my end as well as server logs and try to actually help debug this....annoying though, it shouldn't be difficult to at least not lose messages
<Guest28>civodul: Got a question since you are a maintainer of GNU Shepherd as well.  I am using a Raspberry Pi 4b with GNU Guix system.  I can't use the latest version since Shepherd is always using 100% CPU usage on one core and is therefore unresponsive (can't run herd status or reboot).  Jetomit pointed out this issue:
<Guest28> and in fact my Raspberry Pi has no RTC.  Therefore in the first minute it always thinks it is 1970.  Basically my question is: should I make an issue or not for shepherd?
<Guest28>(now clue if shepherd even uses this, therefore unsure how to proceed)
<nckx>podiki[m]: I'm not going to report this at this time (sorry), I still have 2 unanswered tickets open… I *hope* this stabilises out after the big disablement. If not, we'll see.
<civodul>Guest28: thanks for the heads-up! we should check with wingo how to fix it in Fibers
<civodul>we could also work around it in Guix if you have an idea
<podiki[m]>nckx: no problem, I'm not in a rush with all the changes, let's see how we fare for a bit
<nckx>podiki[m]: Of course, if you want to report it yourself, I'll happily help provide any extra info (or change any settings) if needed.
<podiki[m]>nckx: I'll keep an eye on it for now and see after a bit. i'd rather do guix hacking than matrix/irc debugging :-)
<nckx>ACTION nods cautiously.
<podiki[m]>ACTION makes no sudden moves and doesn't look it in the eye
<Rovanion>I just encountered a bug where chmodded /dev/null 664. I am unsure how that came to be, but my system became a bit unruly after the fact.
<Rovanion>Was probably masked from an earlier install.
<nckx>Of Guix?
<nckx>ACTION looks like this: 🤨
<oldfashionedcow>ACTION also looks like that
<Rovanion>Guix, on Ubuntu.
<nckx>I can see this happening if /dev/null didn't exists, but…
<nckx>I also can't reproduce this.
<Guest28>civodul: Okay, now I can track it.  I don't have an idea for a workaround (new to Guile Scheme and GNU Guix, Shepherd and so on).
<civodul>Guest28: actually the Fibers issue above suggests a fix, trying it now
<patched[m]><nckx> "I'm still not sure what exactly..." <- Joined :matrix, then left :libera, upon which appservice kicked me from :matrix as well
<nckx>But only once, right? (Not that that's great UX, but it's what podiki[m] reported, who is still here now.)
<Rovanion>nckx: What if you try `systemctl mask guix-daemon && sudo ./` on a system where the service guix-daemon does not exist?
<Rovanion>That is how I can reproduce it.
<Rovanion>eh, `sudo systemctl mask guix-daemon`.
<nckx>The only foreign system I have handy is currently ‘guix system init’ing its way out of that sorry state, but when it's done I'll try that before I reboot.
<nckx>It's SystemRescueCD, which is Arch-based, but this being Ubuntu-specific would be just too weird.
<Rovanion>nckx: No, it should be a systemd-thing.
<patched[m]><nckx> "But only once, right? (Not that..." <- Only once, yes.
<nckx>Rovanion: Oh fun, it's running sed on a symlink to /dev/null and sed must be changing the mode for funsies. We could wrap this around a sanity check, but why the devil is guix-daemon masked in the first place? Some edge cases are worth caring about. Some aren't.
<nckx>patched[m]: Fun, more funsies.
<nckx>I guess this one's worth caring about.
<nckx>Or we provide an uninstaller that doesn't leave leftovers like lazy humans do.
<nckx>Oh, boring, looking at the code it's just an explicit ‘chmod 664’. No weird bug. Boo.
<Rovanion>nckx: Yeah, chmod on a symlink to /dev/null. The weird thing is that the symlink doesn't get written over by the cp preceding it.
<nckx>That's documented cp behaviour.
<nckx>I think --force is the thing to use here.
<Rovanion>I guess you have to go one way or the other. When implementing cp I mean.
<nckx>That just ignores the error 😒
<nckx>There must be a cp option we can use to save a precious line.
<Guest28>Why can't I only select boot or esp flag in gparted? I doubt with both the Pi would boot correctly
<Guest28>Ah nevermind, figured it out.  You need msdos partitional label
<nckx>It says esp is an alias fo—yes.
<nckx>AFAIK on GPT, the ‘flag’ is just a UUID.
<nckx>Type UUID, that is.