IRC channel logs


back to list of logs

<lechner>Hi Guix, is the Emacs-native way to handle packages preferred in Guix (as opposed to distro packaging)? Thanks!
<ryanprior[m]>Not sure how you mean that lechner, are you asking if you can use M-x package-install alongside Guix? If so: yes, Emacs native pacakge management works fine with Emacs from Guix.
<lechner>ryanprior[m]: thanks! it was more like, should i?
<ryanprior[m]>It's just pros and cons. I use Spacemacs which has its own package handling with upgrades and roll-back, so I don't feel the need to use the Emacs packages in Guix for most things. If you want to have a fully reproducible Emacs setup then using Guix home & the Emacs packages in Guix would give you that.
<lechner>ryanprior[m]: okay, thanks so much! as a spacemacs user you may perhaps find these XKB symbols interesting
<ryanprior[m]>I created which turns my Caps-Lock and Enter into extra Ctrl keys (when held down like modifiers) while maintaining their normal functions.
<ryanprior[m]>Maintaining normal function of keys is very important to me - I still use both my left & right control keys, so I don't want to map them to something else. I actually use my Caps Lock for E_CONSTANTS and for YELLING AT PEOPLE and so I don't want to remap that either.
<nckx>I guess you don't spam newlines much? Neet idea.
<ryanprior[m]>What do you mean spam newlines?
<nckx>Hold down your return key for whatever reason.
<ryanprior[m]>I can still tap Enter repeatedly & get a bunch of newlines. Only acts as Ctrl when you hold it down while you press another key.
<nckx>Yeah, it's not common.
<ryanprior[m]>Oh yeah I basically never hold down a key for the repeating behavior. I wouldn't say never, but not enough for it to bug me.
<ryanprior[m]>Mostly if I want to emphasize something is realllllllllllllllllllly boring or that a cat is loooooooooooooong
<nckx>Of course in emacs (—yes, and vim, thank you person in the back) you don't need to. I'm struggling to think of a good example, but I very occasionally do, I think. Not a deal breaker though.
<nckx>Hi lfam.
<lechner>maybe for section headings in markdown?
<nckx>I was thinking specifically of newlines and don't write Markdown, but sure, something like that.
<nckx>Anyway, I used to be one of those anti-remapping-I-want-to-be-comfortable-at-any-PC people, but that changed several years back, and now I'm glad I did.
<nckx>‘the end’.
<nckx>What I actually wanted to ask: no Wayland support [planned?], ryanprior[m]?
<ryanprior[m]>I want to totally redo the implementation because it sucks right now
<ryanprior[m]>I have a local branch where I rewrote it to use kmonad instead of xcape, which is better already
<ryanprior[m]>And kmonad works with wayland, so that already is a big improvement
<ryanprior[m]>The issue is, kmonad is a complex beast, way more software than is needed for home-row-control. So I would like to learn how it performs its tricks and write a simple minimal program that does just what I want.
<jgart>is 502ing for anyone else?
<ryanprior[m]>I just seized on the excuse and published my kmonad rewrite:
<ryanprior[m]>If anybody tries it I'd be interested in feedback
<ryanprior[m]>I'm getting 502 as well, it was working 20 minutes ago
<jgart>I've been thinking for a while now how I should use kmonad and I still haven't been able to decide
<jgart>ryanprior[m]: thanks for sharing!
<jgart>I'll check it out
<nckx>kex_exchange_identification: read: Connection reset by peer
<nckx>Oh noes.
<nckx>Did a reconfiguration go terribly wrong?
<nckx>RIP berlin 💀
<nckx>But it's there, and happily serving 502s, that's what's odd.
<nckx>According to the ☝ log there was no reconfiguration due to other unrelated bugs…
<nckx>apteryx: Here perchance?
<nckx>Oh and of course it's Friday night so no remote hands for 2 days.
<jgart>mala: I'm looking forward to more guix vids! ;() Those were great!
<lechner>Hi, spamhaus says my server is infected. Is there a package to detect rootkits?
<nckx>Not in Guix AFAIK (or can search). But since when does Spamhaus diagnose those…?
<cehteh>since when do you have that ip?
<nckx>And option #2, misconfiguration, is still so much more likely than a rootkit.
<nckx>Now, there is lynis.
<cehteh>usually when you setup a new server or get a new 'static' ip for some reason you have to do some work to get it unlisted
<nckx>I guess I've been lucky enough that I'd not use ‘usually’, but it's unfortunately common.
<nckx>Especially if you use the DOs of this world.
<cehteh>my former isp sometimes changed my 'static' business IP for some reasons .. i was always pretty upset becasue of having a lot work with that, reconfigure stuff and get things unlisted at a lot blacklist providers
<cehteh>hence former
<cehteh>no i am at dynamic ip at another provider but have some 2Eur cloud frontend for terminiating things that need static ip's
<cehteh>i put my firewall in the cloud haha
<lechner>it was listed today, and have had it for two months. it's a linode though, and they are pretty tough. the server passes open relay tests
<cehteh>sometimes happens when some neighbour does sh*t, spamhouse and others first try to blacklist entire blocks not single IP's
<nckx>Yes. It's worth asking for a delisting if this is the first time. Their checker tool doesn't seem to indicate how large the block is.
<nckx>Silly tip: if you use separate mx/submission ports, make sure you test them both for open relaying. Some simple on-line tools don't.
*nckx AFK.
<cehteh>its pita but after double checking that your server is really not misconfigured and/or trojaned its then usually easy to do, still sux when your IP gets blacklisted at half a dozen services and delisting may take some time
<cehteh>at the blacklist provider themself and then a lot other companies cache the blacklists .. over multiple levels
<cehteh>can take a week or longer in worst case
<lechner>they blocked a 64 bit subnet on ipv6, which i think is what linode assigned so i'm not sure it's a neighbor. i'm on the XBL exploits list supposedly for advertising a fake HELO, although i do not see it myself. luckily, i have independent providers but i still have to find out what happened. the thing is, i switched to Guix a week ago
<lechner>the HELO is
<bjc>can someone point me to these guix videos of mala's?
*the_tubular is interrested in new guix videos
<cehteh>lechner: of course you need the real hostname there
<lechner>cehteh: which one do you see?
<cehteh>220 ESMTP OpenSMTPD
<cehteh>250 Hello [], pleased to meet you
<apteryx>nckx: now yes
<nckx>Hi! Did you reconfigure berlin?
<nckx>I can't SSH in.
<apteryx>lechner: I only use guix to manage my emacs pacakges; it's cleaner.
<apteryx>and if you mix both, you risk "tainting" your Elisp libraries with different versions, which could introduce issues
<apteryx>nckx: I only tried; but it was failing early
<lechner>cehteh: it's presumably when sending i think
<apteryx>nckx: can't ssh either
<cehteh>lechner: i just telneted it
<lechner>cehteh: into the host?
<lechner>you mean port @%
<lechner>cehteh: thanks for checking. i had a debian trojan once, so they do exist
<wdkrnls>Dear Guix, I still cannot update because of emacs-deferred and I while ddg tells me there was an issue filed, the issue tracker tells me "Bad Gateway". Does anyone know the latest news?
<nckx>wdkrnls: I don't know about the package, but you can use <> while the bug tracker's down.
<nckx>It happens to me too. Just runs $forever.
<wdkrnls>It's strange that they mention they had to uninstall emacs-deferred to make it update. I don't even have it in my profile.
<nckx>Indeed, I assumed they meant they had to uninstall emacs-md4rd.
<nckx>To remove the dependency. But 🤷
*nckx → 😴💤
<nckx>'night Guix.
<wdkrnls>good night
<apteryx>nckx: are you messing up your nights again?
<apteryx>ooh, apologies, my client hadn't scrolled down. good night, nckx
<horizoninnovatio>greetings. Been trying to get a python app running, this is the error: ERROR: Failed to load dependency PyQt5
<horizoninnovatio>* dependency PyQt5. PyQT5 is installed.
<Christoph[m]>Maybe an environment variable is missing? See
<ulfvonbelow>just to let everyone know before I go to sleep, the emacs-deferred test issues are caused by emacs 28.1 no longer supporting the old calling conventions for 'start-process-shell-command'. AFAIK emacs-deferred still works, but the tests use the old calling conventions, so this patches them:
<civodul>Houston, we have a problem: berlin.guix closes connections on port 22
<civodul>rekado, apteryx: do you happen to have an alternate access?
<mroh>ulfvonbelow: Good work, ty! This needs a push to master. Did you send a patch?
<PurpleSym>Hm, `guix reconfigure` fails because the test-suite of guix-1.3.0-27.598f728 errors out 😕
<jpoiret>PurpleSym: seems like
<PurpleSym>Hm, jpoiret, that looks different. Running `guix time-machine -- build guix` now.
<jpoiret>i've been able to build guix today on a recent-ish commit
<PurpleSym>Must’ve been a transient issue. Now it’s buiding again.
<sughosha>Hi all, does anyone know how to enable MAC address randomization for NetworkManager in Guix system?
<faust>Hi guix!
<faust>just have a question, how i can upgrade golang package to latest?
<f1refly>thunar-volman is broken for me. I have gvfs installed globally, but thunar and thunar-volman are installed for my user profile. when I open thunars settings and go to the 'advanced' tab, it tells me that gvfs is not available and that thunar-volman won't work
<f1refly>Any idea why thunar-volman doesn't find the installed gvfs?
<faust>i just run guix build go --with-source=go@1.18.2=./goroot
<faust>and it finished with success
<faust>but i dont find go in .guix-profile/bin/
<faust>how i can fix this?
<faust>i have /gnu/store/zx2k3vxal1a10s533ws9rxpq6ivvpabj-go-1.18.2/
<cbaines>faust, guix build builds things, guix install will build things and add them to a profile
<faust>which cmd do i need?
<cbaines>well, do you want to add something to your users profile?
<faust>yes i want sim link  binary
<cbaines>then use the second command I mentioned, guix install
<faust>cbaines guix install but which parameters?
<faust>cbaines guix install  /gnu/store/zx2k3vxal1a10s533ws9rxpq6ivvpabj-go-1.18.2  ?
<nckx>civodul: You're not on guix-sysadmin (should you be)?
<cbaines>faust, that depends entirely on what you want to install. If you want to install the go package you previously built, use the same parameters as you passed to guix build
<nckx>civodul: Even the alternative access does not provide access.
<nckx>apteryx: <Are you messing> I mean… the answer was still yes :)
<nckx>But much, much less.
<faust>cbaines thanks!
<ekaitz>hi all I packaged glfw, anyone wants to give it a try before we push it?
<ekaitz>any gamedev around?
<ekaitz>also, should I make two versions one with wayland and other with x11 or both at the same time are ok?
<unmatched-paren>don't we already have glfw?
<unmatched-paren>`guix show glfw` works
<unmatched-paren>it's quite a popular library, i'd be surprised if it wasn't on guix
<ekaitz>oh I tried yesterday and I didn't find it
<ekaitz>probably I misspelled? hahaha
<ekaitz>spent 15 minutes of my life doing nothing!
<unmatched-paren>happens to the best of us :)
<ekaitz>anyway I'm still wondering, should we distribute a wayland version too?
<unmatched-paren>hm, you have a good point there
<unmatched-paren>problem: that would require us to distribute a wayland version of all GLFW-using apps
<unmatched-paren>strange that the guix package doesn't depend on wayland
<ekaitz>require? not necessarily
<ekaitz>yeah, glfw by default builds for x11
<unmatched-paren>i'd think depending on both would be fine; gtk does that
<ekaitz>yeah but the lib has a flag to say if it should use one or the other
<unmatched-paren>oh, really? never both?
<ekaitz>take a look there
<ekaitz>it mentions a flag
<ekaitz>(have to go, brb)
<unmatched-paren>i think USE_WAYLAND might just enable wayland support without disabling x support
<unmatched-paren>might be wrong tho
<unmatched-paren>since wayland support may be experimental/unstable
<nckx>One that supports both, unless there are real disadvantages.
<nckx>(There's already a glfw in gl.scm, but it doesn't mention Wayland in its inputs.)
<nckx>civodul: Is the ‘unimplemented’ here a programming error making this otherwise mundane network error worth reporting? <>
<nckx>Nice URL.
<nckx>I was using mosh, so there's no more backtrace.
<nckx>ekaitz: Old messages; server burped.
<nckx>But ‘we already have glfw’ remains.
<ekaitz>aiight unmatched-paren nckx I'll try adding wayland support and see what happens, maybe we should add it
<civodul>nckx: no idea; that's with the current 'guix' package?
<nckx>civodul: Yes.
<civodul>uh, first time i see it
<civodul>is that reproducible or transient?
*jess yawns
*nckx puts out snax.
<nckx>civodul: ‘Transient’…
*jess bites snax, bites nckx
<civodul>PurpleSym: what test suite failure did you see?
*nckx no is snak :(
<civodul>nckx: re berlin, access over Tor isn't helping if that's what you mean; i suspect it could be shepherd rejecting incoming connections because the max threshold has been reached
<nckx>No, I meant the remote serial connection.
<civodul>oh you have that?
<nckx>Even logging it at the console throws a PAM error.
<civodul>ah, what kind?
<nckx>I do, but I haven't used that super power yet; rekado has.
<nckx>berlin login: root
<nckx> login: PAM Failure, aborting: Critical error - immediate abort
<civodul>anything on tty12?
<nckx>OK, if I'm doing this I need to get my decryption keys, that'll take a mo.
<nckx>No GPG here.
<civodul>ah alright
<civodul>i grepped Linux-PAM source for PAM_ABORT but it's hard to guess what could be going on
<faust>i want guix pull but without upgrade my linux core
<unmatched-paren>so you don't want to update linux-libre?
<unmatched-paren>you need to "pin" the kernel package
<faust>not libre just linux )
<unmatched-paren>i see. you should ask in the naughty channel's IRC then :)
<faust>but how i can pin it ?
<unmatched-paren> <- have a look at this.
<unmatched-paren>once you have an inferior, you can (lookup-package-inferior "linux") and use that as your (kernel)
<unmatched-paren>in your system config
<unmatched-paren>sorry, (lookup-inferior-packages "linux")
<unmatched-paren>*(lookup-inferior-packages inferior "linux")
<unmatched-paren>actually (first (lookup-inferior-packages "linux"))
<PurpleSym>civodul: worker(s?).scm failed, but I don’t have logs and the second try succeeded.
<unmatched-paren>because (lookup-inferior-packages) returns a list of all matching packages
<faust>unmatched-paren i need to specify version for linux
<civodul>PurpleSym: hmm ok
<unmatched-paren>i think (lookup-inferior-packages inferior "linux") will return the most recent version in the inferior first anyway
<unmatched-paren>so you don't need to specify a version of "linux"
<faust>unmatched-paren but i whant to keep my current version of linux
<unmatched-paren>well, what's the most recent commit of the channel that provides linux with the version you wish to keep?
<unmatched-paren>you should specify that as the inferior's commit
<unmatched-paren>s/inferior's/channel's/ s/commit/commit in the inferior/
<unmatched-paren>and then the most recent version _will_ be the version you want to keep
<unmatched-paren>as far as (lookup-inferior-packages) is concerned
<faust>unmatched-paren oh) but is it possible to specify just a linux version? which i whant to keep
<unmatched-paren>it depends on whether that version still exists in the channel
*unmatched-paren afk
<janneke>sneek: seen dannym?
<sneek>Dynom?, pretty sure was seen in #guix 2 months ago, saying: But thank you lilyp for the clarification.
<janneke>thanks sneek, i'll take that as a "no"
<civodul>sneek: seen dannym
<sneek>I last saw Dynom in #guix 2 months ago, saying: But thank you lilyp for the clarification.
<civodul>hmm, not any better
<civodul>howdy janneke :-)
<faust>unmatched-paren looks like i can point package version
<faust>(first (lookup-inferior-packages inferior "linux" "5.4.21"))
<Dynom>janneke: don't think you are looking for me are you? q;)
<nckx>civodul: …aaand I'm back. You're going to love this as much as I do. Console access currently requires a working SSH forward… from berlin. :-/
<nckx>Somebody (R?) suggested using another node, but I can't find the rest of that discussion. Maybe there was none.
<janneke>hey civodul
<janneke>Dynom: hi, no that was sneek's guess :-/
<nckx>I think we just have to wait until Monday.
<faust>unmatched-paren but in which file in need to set lookup-inferior-packages?
<civodul>nckx: oh oh, interesting :-)
<nckx>“ ” ☺
*janneke is working to resurract the arm bootstrap
<nckx>with a tesseract.
*janneke has been found out, damn
<nckx>Which is about as comprehensible as most of the bootstrap is to me at this point, I just ignorantly admire its geometry from afar.
<nckx>* beautiful geometry, I meant to write.
<janneke>i have kind of the same about the arm bootstrap, esp wrt to eabi/v4/v7 support etc.
<nckx>civodul: Without MDC help, I can't think of a way out of this puzzle box.
<nckx>Barring an nginx exploit (wasn't there a recent RCE in Linux…?) — no, no, hush.
<rekado>sorry, do I need to do something?
<rekado>I just got back home (but might leave again)
<rekado>nckx: yes, node 130 is connected to the public switch, but I don’t think anyone actually assigned the second public IP, so it’s not going to be easy to connect to it.
<rekado>serial console on 130 is difficult, because *all* the build farm nodes are misconfigured wrt serial.
<rekado>there seem to be *two* consoles connected to the same device, and input is sent to either of them at arbitrary points, so I can’t even log in reliably or type a command
<rekado>looks like this if you’re curious to know:
<nckx>rekado: Ohai. I *just* sent you an e-mail asking the same.
<nckx>Thanks for the deets.
<nckx>No, I don't think there's much you can do in a short time slot.
<nckx>rekado: …yikes.
<faust>   in Pinning package versions section i dont follow in which file i need to put this code
<faust>can any one help me?
<nckx>Considering ‘only’ issues. is affected at this point, not the Web site or CI (previous status quo at least), I don't think it's worth the risk going for a quick fix.
<cbaines> looks to be down
<nckx>That's what I meant by affected.
<nckx>faust: Your system.scm, the one you pass to ‘guix system reconfigure’. But that channel has its own separate support channel on this network.
<cbaines>ah, sorry, I didn't parse your message properly
<nckx>Nah, I don't know why I went for corporate euphemism either. ‘Down’ is what it is.
<rekado>what happened?
<rekado>is this a reconfigure gone wrong?
<nckx>We don't know.
<nckx>Well, unless someone did reconfigure without telling us, but reconfigure was actually broken last apteryx & civodul tried yesterday.
<nckx>Somebody reported issues. 502ing last night, I tried to SSH in to poke it, that failed, then I sent my mail & went to bed. That's all I know.
<faust>nckx thanks!
*nckx didn't end up in bed for long but the intention was there.
<civodul>nckx: i didn't try to reconfigure yesterday, but apteryx mentioned that the sysadmin modules needed an update
<nckx>OK. I thought I saw you coordinating a reconfigure to fix the ‘In progress’ issue. Maybe I was mistaken.
<rekado>fixed the serial on 130 by booting with just "console=ttyS1,57600n8" instead of "console=tty0" "console=ttyS0,57600n8".
<rekado>I cannot see the boot output this way, but at least there’s only *one* console instead of two that share the same input
<rekado>let’s see if this other public IP will work here…
<oriansj>and for whoever finally packaged kiwix, *THANK YOU*
<faust>on guix reconfigure got error
<faust>ice-9/boot-9.scm:1685:16: In procedure raise-exception:
<faust>error: xfce-desktop-service-type: unbound variable
<nckx>faust: Did you import (gnu services desktop) or (use-service-modules desktop)?
<nckx>Both are equivalent, you need only one.
<nckx>* More clearly: did you (use-modules (gnu services desktop) …) or (use-service-modules desktop …)?
<faust>nckx yes (gnu services desktop) helped!
<faust>guix thaks for your support!
<nckx>Great. That's one of the lesser aspects of the ‘it's all just a programming language’. You have to more or less know which modules provide what, or search for them. Here you could have used ‘guix system search xfce-desktop’ and translated the ‘location’ field to its (foo bar) form, but I admit that's all a bit obscure.
<nckx>faust: Thanks :)
<rekado>to use the second public IP we need support from the network team; the node’s mac address needs to be explicitly permitted.
<nckx>I'm guessing they don't work week-ends.
<avp>Hello, Guixers.
<faust>nckx how i can navigate guix code in emacs?
*avp trying to package spaCy (
<nckx>Hi avp. faust: I don't use any advanced emacs features, sorry, I just grep for stuff and ‘know’ most things already. Maybe someone else has a better answer.
*nckx AFK.
<rekado>I’ll be off next week and not around computers for the most part; but I can submit a request next week anyway.
<avp>I've sent Guix patches with "cymem" and "preshed" Python modules that are required for spaCy.
<rekado>(all that does is give you an alternative path to access the serial interfaces of all nodes, but I guess that’s all you need to reboot and fix berlin.)
<faust>oh my reconfiguretion is fineshed, but now i got this
<faust>sudo: you do not exist in the passwd database
<faust>how i can fix this?
<faust>is it possible to rollback my configuration without root?
<ryanprior[m]>Not a system-wide configuration, but a profile or user configuration sure
<faust>but how i can user to /etc/passwd without root?
<ryanprior[m]>Well, probably there's a way and I just don't know what it is
<faust>i need reinstall os?
<ryanprior[m]>That would be one way, you might also be able to boot from a live USB and fix /etc/passwd and then reboot
<rekado>you could reboot into an older system generation
<nckx>rekado: <all that does> Yes, I think that's enough here, but who knows. That's what we thought about the jump through berlin itself :)
<nckx>Is ruby-ruby-prof broken for everyone? No 'tutes either.
<civodul>rekado: thanks for looking into it!
<civodul>so you rebooted it, right?
<civodul>cbaines: my tests for the make-inetd-constructor changes in the Shepherd look conclusive (
<civodul>so i think i'll go ahead and release 0.9.1 tomorrow or so
<cbaines>great :) I haven't got around to testing anything yet unfortunately
***yqrl is now known as yarl
<yarl>502 on
***Alex[m]1 is now known as Alex[m]11
<nckx>It is known. Enjoy our other fine subdomaines which, for now, do work.
<nckx>Also, for bug reports, this:
<nckx>Patches are tracked separately on this one; add ʿ-patchesʾ for those.
*civodul added a .guix-authorizations file in the shepherd repo
*nckx noticed. Yay!
<civodul>every time there's a patch-removing patch, "git am" fails when i pipe the email right into it
<civodul>i think someone explained me what to do a while back
<civodul>but i forgot
<nckx>pashencija[m]: Thanks for giving cross-compilation a valiant go.
***Xenguy_ is now known as Xenguy
<bricewge>Hi Guix!
<Tcache[m]>bricewge: hi
<Tcache[m]>nckx are you the one running relay bridge irc to matrix
<nckx>No, that's Matrix themselves for all of Libera.
<bricewge>I have the bad habit of keeping a personal branch with a too much custom commits and use it to reconfigure my system « sudo -E ~/project/guix/bricewge/pre-inst-env guix system reconfigure »
<Tcache[m]>nckx: Oh i forgot lmao ;D
*nckx .oO …‘bad’? Oh dear :)
<nckx>Tcache[m]: Complaints?
<bricewge>Unfortunatly, it clutter my GUILE_LOAD_COMPILED_PATH, I end up with errors like `In procedure abi-check: #<record-type <service-type>>: record ABI mismatch; recompilation needed`
<Tcache[m]>nckx: Nope
<Tcache[m]>a question
<Tcache[m]>A stupid question
<bricewge>Do you have a clean way of using such custom branch, and not having to delete all thoses guile compiled module from time to time?
<nckx>I won't say those don't exist, but that wasn't one. I doubt many people know how Matrix <-> Libera works.
<Tcache[m]>So if am from matrix no need to register a nick using nickserv?
<bjc>bricewge: clean is in the eye of the beholder. or, in my case, a make target: ‘make clean’ will get rid of all the old .go files
<bricewge>Tcache: You better register your nick, or you will get the [m] on the IRC log
<nckx>I'm no expert on Matrix, but I don't think the bridge changes that (***[m] nicks aren't magically protected AFAIK, but IK not much). I registered mine. I recommend you do, unless told otherwise by a competent.
<Tcache[m]>bjc: Agreed
<nckx>Well, sure, if you want to avoid the [m] of shame that's true as well.
<unmatched-paren>stupider question: why not just use an IRC client?
<bricewge>Or someone could impersonate you on Libera if other people don't check that [m] tag
<nckx>bricewge: Avoid what, exactly? Are those ABI warnings bogus? If they aren't, they are there for a reason, and you need to ‘make’.
<Tcache[m]>nckx: Well how can i register my nick on matrix
<Tcache[m]>bricewge: lol
<bjc>nckx: my experience has been that ‘make’ alone is not enough. i've always needed to clean first
<nckx>Uh something something appservices something. Of course it's some arcane chat with a service bot, because IRC, but it's twice as worse because also Matrix. Let me load up my Matrix account and see if it's saved.
<Tcache[m]>unmatched-paren: Well ill come back to IRC soon
<nckx>bjc: Yeah, clean-go.
<nckx>That was totally implied :-x
<bricewge>bjc, nckx: These ABI ar from my config repo, not the Guix one
<bricewge>Here is an exemple
<nckx>Tcache[m]: Better idea: here's a how-to.
<bricewge>I already `make clean` `make clean-go` when pulling guix repo
<nckx>Oh, you're using --load-path.
<nckx>That's news :) No, then I don't know.
<bricewge>But my question is not that specific, how do you use a custom branch for guix and personal config repo in a sane way.
<bricewge>My setup always feel like a hode-podge and because of that I rarely look forward to rebase on master :/
<nckx>--load-path XOR sane, but that's just me. I don't know your situation. I thought you were doing what I do, which is just ~/guix/pre-inst-env guix system blah /etc/guix(a git repo)/system.scm
<nckx>but I don't use multiple channels like that.
<bricewge>Yeah, it's easier that way.
<Tcache>Ok am on irc now
<nckx>Now you should be able to register with NickServ.
<Tcache>I missed IRC
<unmatched-paren>try `//msg NickServ help`
<bricewge>But I want to be able to test, develop and use multiple channel without having to `git commit` each changes I'm making
<unmatched-paren>i mean `/msg NickServ help`
<nckx>(I thought you could do that from Matrix, too, but 🤷)
<bricewge>I thought too
<Tcache>nah i dont know
<akonai>What does the /etc/static directory contain?
<Tcache>My friend is offline so
<nckx>But I already had an NS account so didn't test that.
<nckx>akonai: Only links into /gnu/store (it's such a link itself). So it can never be modified.
<nckx>But Guix isn't very diligent about how or when it's used, the other half of /gnu/store links just live elsewhere in /etc.
<akonai>I see
<nckx>I don't think there's much discipline, or I fail to see the rules.
<nckx>It's inherited straight from Nix.
<bricewge>Some software don't like having their /etc configuration with the right imposed by /gun/store
<Tcache>Hei guy's how i can remove my old nick which i even forgot password of
<unmatched-paren>i think you need the password to drop the old nick
<nckx>It's fine to ask here, but you might be better served in #libera.
<Tcache>Ahh i dont even have that, my old nick is Nutte
<pashencija[m]><nckx> "pashencija: Thanks for giving..." <- Have you seen my patches? :)
<nckx>s/might/will almost certainly/
<nckx>I only randomly happened to see your btrfs-progs report because I pointed someone at and it was then the top bug, and it looked easy 😉
<nckx>So no.
<nckx>(Yes we know that issues.guix is down, before anyone asks.)
<nckx>Oh is that the flag API stuff?
*nckx AFK.
<pashencija[m]><nckx> "Oh is that the flag API stuff?" <- Yes
<pashencija[m]>I sent three patches that seem to work here
<rekado>nckx: I’ll be out tomorrow for most of the day; but I guess I could reboot berlin in the afternoon.
<rekado>the problem with that server is that it doesn’t boot reliably since the move to the SSDs with btrfs.
<rekado>last time I rebooted half a dozen times before the disks were ready.
<apteryx>rekado: not sure if it'll help but I re-added the 20s rootdelay since that problematic boot
*nckx peeks in.
<nckx>No pressure.
<pashencija[m]>So why btrfs?
<pashencija[m]>Btw issue web interface seems to be a small problem. Everyone can access email
<nckx>Can you elaborate pashencija[m]?
<apteryx>pashencija[m]: zstd compressed file system; and no running out of inodes
<pashencija[m]>apteryx: Sounds good. Sad it's not working properly
<pashencija[m]>Actually, there's no btrfs support in Image API. I guess I should add one later
<nckx>There's no support for most Guix-supported file systems, is there? It's all rather ad hoc at the moment.
<pashencija[m]>nckx: I am afraid I do not really understand your question
<apteryx>yeah, it's been annoying. the plan is to move to another large array we got access to since, and continue experimenting with that 6x drives SSD Btrfs RAID10 array on another, very similar server
<nckx>> Btw issue web interface seems to be a small problem. Everyone can access email
<nckx>Which interface (the one that's down?); what does ‘access email’ mean.
<pashencija[m]>nckx: I mean there seems to be no need to hurry with web interface as long as email lists work properly
<unmatched-paren>they probably mean reading the site, and checking their own email list subscription
<pashencija[m]>unmatched-paren: Exactly
<nckx>pashencija[m]: Aaah. I thought you meant there was a ‘small problem’ *with* the interface :) All good. Yes, I agree, plus there's always still, emacs-debbugs, …
<nckx>It's not worth rebooting in a hurry. The important services are still up.
<unmatched-paren> seems to redirect to
<pashencija[m]>nckx: There's only support for ext* and partial support for fat16
<pashencija[m]>There will be full support for fat16 as soon as my patches are applied
<nckx>Yeah. I was implying that it might make sense to ‘unify’ it with (gnu build file-systems) and the like, so anybody adding the next fs can implement everything there, not chase around random APIs.
<nckx>I don't know if that makes sense! Just an idea.
<pashencija[m]>I am really surprised guix doesn't have many mirrors like debian or ubuntu do
<pashencija[m]>It would be nice to have these, especially they can just copy things without re-signing them
<nckx>I just remember that adding, say, jfs/bcachefs/…, I had to ‘grep olderfilesystemname’ around the tree to find all the places I needed to extend.
<nckx>pashencija[m]: They're slowly coming. cbaines just mailed something to that effect today, mentioning the nar herder, although I haven't read it yet.
<pashencija[m]>nckx: Sounds compilcated
<nckx>Historically, mirroring Guix was ‘hard’ because you couldn't just point rsync at it like most mirror operators expect.
<pashencija[m]>1. file-systems are not generated in the API you mention
<pashencija[m]>2. Image api uses genimage, so each FS needs special support
<pashencija[m]>3. Most file systems are never really used in images
<nckx>I'll take 1 & 2, let's leave 3 aside where it belongs.
<nckx>Each file system needs ‘special support’ in Guix.
<nckx>I should be working, anyway 😉 o/
<oriansj>pashencija[m]: umm anyone can setup a guix binary mirror really easily
<pashencija[m]>nckx: It could be easy done with proper documentation.
<pashencija[m]>Also, combining APIs would
<pashencija[m]>1. Take a lot of time and effort. I spend all the time I can spend on porting GUIX to arm devices and fixing things around.
<pashencija[m]>2. Possibly break backward-compatibility. Image API is documented and used not only inside GUIX with supported boards, but also outside it with 3rd party devices.
<nckx>oriansj: Different kind of ‘mirror’.
<nckx>Night all.
<pashencija[m]>Ofc I wouldn't mind if someone made Image partition creation for all the supported file systems and unified API, but it's not something I find possible or worth-the-effort
<pashencija[m]>Also, that might need extending genimage as it does not support random file systems at the moment (the list is limited for GPT and unlimited for MBR)
<pashencija[m]>Good night, nckx
<GNUtoo>hi, is the blog up to date?
<GNUtoo>It seem to not have the latest blog posts anymore
<oriansj>well to be fair there are only a few things one could mirror: the code itself (git clone covers that), the package source tarballs (not easy out of the box but possible), the built binaries (which I linked how to generate), the issue tracking (not sure if mirroring is setup for that yet) and community interaction [irc/mailing lists] (multiple ways to mirror)
<GNUtoo> also has "502 bad gateway"
<rekado>GNUtoo: we know
<unmatched-paren>hm, i didn't know there was new blog posts that hadn't appeared; where should i find them other than the guix-artwork.git repo, which makes it hard to see which are the new ones?
<unmatched-paren>also: what specifically does `--tune` do? does it enable processor-specific optimizations in e.g. gcc that would make it hard to reproduce binaries on different machines?
<GNUtoo>ok thanks
<GNUtoo>unmatched-paren: for tune I think there was some explanation on a website dedicated to using guix on high performance computers
<GNUtoo>It goes in a lot of details that are really interesting
<unmatched-paren>ah yes
<unmatched-paren>it does \o/
*unmatched-paren wonders whether it would affect the performance of supertuxkart and minetest
<unmatched-paren>which is currently very poor on this laptop
<unmatched-paren>how would i enable tuning in a package declared in guix home?
<GNUtoo>unmatched-paren: Good idea: it would be neat to be able to use Guix to do benchmarks
<GNUtoo>There is an old version of phoronix-test-suite that is packaged in parabola but I'm not good enough to be able to hook guix in its tests somehow
<GNUtoo>*good enough in php
<GNUtoo>Though for specific games like supertuxkart massive performances can sometimes be gained by enabling or disabling specific options in the game itself
<GNUtoo>for instance in xonotic some settings (like shadows that are calculated in real time and so ) are really demanding while some other are not that demanding
<unmatched-paren>sometimes supertuxkart makes the graphics output freeze up, even on this pretty new laptop
<GNUtoo>For me it works fine on an old desktop computer with an NV50 GPU
<unmatched-paren>i think the problem is mainly that i'm using an integrated GPU
<unmatched-paren>since my nvidia GPU seems to be too old for nouveau to work effectively
<unmatched-paren>i plan to get a framework laptop at some point, which seems to have far better integrated graphics...
<unmatched-paren>(not sure if they're even integrated. they're apparently "Intel Xe" graphics)
<unmatched-paren>oh, no, it's not
<unmatched-paren>wait no it can be either
<GNUtoo>Nvidia gpu too old for nouveau are really old stuff
<GNUtoo>It means that they need to be older than the RIVA TNT / TNT2
<GNUtoo>ah s/old/new
<GNUtoo>There is still llvmpipe but it can also be very slow