IRC channel logs


back to list of logs

<anon2029>I was going to ask what the plan for packaging Emacs 27 was, but it seems to have been done in the last day or so. But why was there a delay? I thought Guix (System) was rolling release by default
<lfam>anon2029: Guix is a rolling release distro, but that doesn't mean it offers same-day updates, or even any kind of schedule
<lfam>Usually, if a package is not getting updated quickly, it means one of two things. Either nobody cares enough to do the work, or the work is hard
<lfam>For Emacs, the work is hard, as in this case there was a miscommunication about the update process and a lot of dependent packages were broken
<lfam>Here is a link to the discussion about it:
*nckx RETs.
<nckx>pkill9: If there's a good reason to package a newer ytdl for the time being (like, apparently, the developer taking a holiday, which I'm sure they deserve) we could do it. If significant things are broken.
<lfam>anon29: If there is something that seems egregiously out of date, please feel free to let us know, or even try updating it yourself!
<lfam>anon2029 ^
<nckx>So it's exactly what I expected, liblouis = _loader[""] needs to be monkey-patched to use an absolute path, assuming ‘_loader’ supports that. But that's the general solution.
<jonsger>what is the herd equivalent to `systemctl suspend`
<nckx>‘Not herd's job’ 😛
<nckx>Use loginctl, or echo mem > /sys/power/state.
<lfam>We could make an alias!
<lfam>That's effectively what `systemctl suspend` is
<nckx>Why stop at herd? emacs suspend!
<vagrantc>is there a way to disable suspend/hibernate entirely?
<nckx>(If services ever support sleep/wake-up service hooks it would make sense to do it via herd, although that sounds fragile.)
<vagrantc>sometimes elogind likes to suspend the system as soon as it starts on a system that is unable to resume ...
<lfam>I don't think we really support hibernation (unfortunately), so no need to worry about that one ;)
<vagrantc>what's weird is switching to old generations still have the problem, except for generations without elogind
*nckx coughs feebly, it's coming, just one more bug™...
<vagrantc>so it doesn't appear to be upgrade related ...
<nckx>vagrantc: Did you set elogind-configuration to disable handle-lid-switch etc?
<vagrantc>nckx: it's not clear *why* it's going into suspend ... if it's because of handle-lid-switch or not...
<lfam>That sounds annoying
<vagrantc>it used to not do this, and on generations i *know* it didn't used to do it, it's doing it now ...
<vagrantc>hard to troubleshoot, too ...
<nckx>vagrantc: Do you use a desktop environment?
<vagrantc>nckx: just sway and some elogind stuff
<anon2029>lfam: would there be a way to define my own Emacs package and build it from source? I don't use Guix for Emacs packages and this experience has not encouraged me to rely on the main Emacs package if it's going to be held back by all of its Guix-exclusive dependencies (which wouldn't be the case with straight.el)
<nckx>I wonder what a raw system without elogind does on laptops when you close the lid.
<nckx>I'd assume (hope) nothing, but I've not actually tried.
<vagrantc>wow, and now this other machine is kernel oopsing and panicing
<vagrantc>not my day for guix :/
<nckx>The default for logind systems is to handle-lid-switch → suspend, so no mystery there, unless it's been disabled.
<nckx>I went there last week ☹
<lfam>anon2029: Sure, custom packages are a first-class use case for Guix. You would first define your package, and then use it via a "Channel"
<vagrantc>nckx: as in the hardware lid switch is broken?
<lfam>Feel free to ask for help!
<vagrantc>nckx: i would assume handle-lid-switch would only matter when ... the lid is closed?
<nckx>As in my system kept oopsing when I closed it. I'm sure Guix was not to blame, my kernel is... a colourful quilt.
<jackhill>In the emacs case, there was also emacs-next, which had 27 for a while.
<nckx>And when I plugged it in with the lid closed it would suspend when I opened it.
<nckx>Not sure if that's what you mean?
<jackhill>Now we just need package-with-emacs-next for the elisp packages
<vagrantc>nckx: lid open from before it had power
<vagrantc>guess i could try booting with the lid closed :)
<nckx>Some people on the Internet report disabling handle-lid-switch + handle-suspend-key fixing suspend on boot even with no switches involved. It's worth a try.
<vagrantc>true enough
<lfam>Wow! at these write-derivation performance improvements. Two 4% improvements in a row
<nckx>(I agree it doesn't make sense!)
<lfam>That's awesome civodul
<anon2029>lfam: Thank you, I'll give those a read
<nckx>raghavgururajan: Wrapping orca with an LD_LIBRARY_PATH containing liblouis should also work (grep gnu/packages for LD_LIBRARY_PATH + wrap-program for many examples). I prefer substitute*ing _loader, but both approaches are OK.
<civodul>lfam: tx, though we need many more of those!
*civodul -> zZz
<nckx>Good night.
<nckx>I just saw those commits ~8% isn't bad for free.
<lfam>It's actually more than 8%
<lfam>Well... I guess it depends on how it was measured. If each commit was measured without applying the other, it would add up to >8%
<nckx>I mused about that for a while, then decided to just write 8 😉
<lfam>You had the right idea ;)
<nckx>It could be less, too.
<lfam>Yeah 🤦
<nckx>I suspect 4% is (shocking!) a rounded number anyway.
*nckx calls Phoronix.
<raghavgururajan>nckx: Thank you!
<sneek>raghavgururajan, you have 2 messages!
<sneek>raghavgururajan, nckx says: if you don't have a backlog.
<sneek>raghavgururajan, nckx says: Patching to ‘liblouis = _loader["/gnu/store/58ihlq6rzqz8iqyp7v5fk7f34r004f0v-liblouis-3.14.0/lib/"]’ works. Now it complains that I'm not running at-spi-bus-launcher, which is 100% true.
<raghavgururajan>nckx: That is perfect! Thank you so much.
<raghavgururajan>sneek: Have some botsnack, will ya?
<drgibbon>hello there, is KDE 5 available on guix these days? I seem to remember not, but looking at the packages it seems that there are KDE5 things available :)
<mroh>aha, seq seems to kill emacs27...
<pkill9>there are kde packages available, but not the full desktop environment, because nobody has desired to add it
<pkill9>or, nobody who desires to add it has been able/bothered to
<anon2029>I'm doing a reinstall of Guix on another laptop, just a barebones TTY at the moment. Can anyone recommend some config.scm examples for users of window managers? Or good config.scm examples in general. It would be pretty nice to install all my usual programs in one fell swoop (in particular, both Icecat and fonts for it to use), but I'm not really sure which modules the various programs fit into, or the distinction between use
<anon2029>and use-package-modules. Are there inventories of modules and their contents available?
<anon2029>Apologies for the wall of text.
<pkill9>anon2029: there's an example in guix's source:
***catonano_ is now known as catonano
<drgibbon>pkill9: ok thanks!
<anon2029>To make the story short, is there a list of modules and their contents available, or do I have to guess what module a package would be in and then guess the name of the module?
<str1ngs>anon2029: guix show package will let you know what module it's in.
<str1ngs>ie for guix show bash. location: gnu/packages/bash.scm:124:4 is (gnu packages bash)
<str1ngs>anon2029: guix search is also helpfule
<anon2029>str1ngs: thank you, this has been a bit of a head-scratcher for me
<jackhill>heh, there really is an xkcd for everything. I'll know we've made it when there is one about Guix :)
*raghavgururajan couldn't sleep
***ezzzc6 is now known as ezzzc
<xelxebar>str1ngs: Would you be able to build linux-libre@4.9, but with CONFIG_DEBUG_INFO enabled?
<xelxebar>I'm looking at gnu/packages/linux.scm to see if there's an easy way to enable that option.
<xelxebar>Okay. I think I could give you a patch to gnu/packages/linux.scm capable of building the kernel with debug syms
<mitchk>I'm getting consistent Xorg lockups - and I'm tailing logfiles in /var/log/ while it locks and I don't see any fresh errors come across. Sometimes Xorg will come back if I let it sit, othertimes I have to restart xorg-server
<mitchk>tailing through remote ssh, is there any other logs I should be looking at?
<lfam>I would look at /var/log/messages
<lfam>Or dmesg
***sneek_ is now known as sneek
<str1ngs>xelxebar: if you have expression I can build it for you.
<xelxebar>str1ngs: Thanks. Actually, I figured full debug symbols aren't necessary.
<xelxebar>I'm trying to get perf to see the symbols for my external kernel module.
<str1ngs>xelxebar: no worries let me know if you need anything built.
<peanutbutterandc>Hey there,
<peanutbutterandc>I am trying to add a wrap-program phase... where should it go in the %standard-phases? After 'install? o.O
<rekado>peanutbutterandc: yes, that’s a good choice.
<str1ngs>xelxebar: here's a hack for CONFIG_DEBUG_INFO #t. this adds the '("CONFIG_DEBUG_INFO" . #t) pair to %default-extra-linux-options. my set! expression is not advisable but it works for one offs like this.
<str1ngs>xelxebar: I use GUIX_PACKAGE_PATH so you might need to change this from a module to use -l
<peanutbutterandc>rekado, Oh hey there. Thank you very much. Um, another question, is it possible to use (package-inputs this) to get the inputs of a package from the definition itself? O.O
<str1ngs>peanutbutterandc: not this but you can use a let I have an example here one sec.
<peanutbutterandc>str1ngs, I think I read somewhere about self-reference to the package being defined. In the manual....
<str1ngs>peanutbutterandc: that would be helpful. but this is not exactly true inheritance like goops so not sure how you would do that.
<peanutbutterandc>str1ngs, I see... that definition seems to be a child package that inherits from a package. Mine is an independent one... but I'll see if I can do something... thank you
<str1ngs>peanutbutterandc: I misunderstood, my bad
<peanutbutterandc>str1ngs, It's all right (:
<str1ngs>peanutbutterandc: in phases you can reference inputs
<str1ngs>so maybe that is what you need?
<peanutbutterandc>str1ngs, teach me sensei! Yes. I need to reference the inputs of the package currently being defined
<str1ngs>in the context of a phase those?
<peanutbutterandc>I am an idiot. I think %build-inputs might be what I need.
<peanutbutterandc>I'll try and see
<str1ngs>okay does that work outside phases?
<str1ngs>normally I use #:key input in the phase lambda
<peanutbutterandc>str1ngs, Probably not. I haven't tried it yet. I have only used it in phases... what does the lambda get called with, normally? The build inputs? o.O I have been using '_'
<leoprikler>(lambda* (#:key inputs outputs ... #:allow-other-keys))
<str1ngs>I believe they become derivatives so they are called within the guix-daemon
<str1ngs>I pray I got that right
<leoprikler>I think it's derivates, but close enough
<str1ngs>did I screw that up too? :P
<str1ngs>leoprikler: btw I do now have a working webkitgtk with webrtc. but now I need to translate it to something that works with code.
<str1ngs>it does work with MiniBrowser to some degree though. does not help I don't use anything that uses webRTC lol
<str1ngs>also I think in the process I found some potential performance tweaks for webkitgtk. hope to roll those back to webkitgtk proper.
<str1ngs>baring any freedom issues.
<xelxebar>str1ngs: Thanks. Here's an alternative approach:
<str1ngs>xelxebar: nice, ah much better
*str1ngs throws his set! out
<xelxebar>Wait, I just realized that make-linux-libre* isn't public...
<xelxebar>Anyway, I feel like the "right" way to do this is create a debug output on all the kernels :p. That's going on my backlog of things to look at.
<str1ngs>xelxebar: yah though technically this could be used with ./pre-inst-env I guss
<str1ngs>xelxebar: if you find something that works an you need this built let me know.
<xelxebar>Cheers. I'm just trying to debug this kernel module at the moment.
<str1ngs>I think you don't actually need ("CONFIG_DEBUG_INFO" . #t) right now?
<str1ngs>probably the linux kernels need a debug output. if they don't already?
<peanutbutterandc>Okay, I find myself repeating myself while (wrap-program ...)-ing. I'm adding all the inputs/bin to PATH, basically. And I would really like to not repeat myself. Any tips please?
<peanutbutterandc>Okay, the thing I was looking for was this-package. (at the bottom of the page)
<peanutbutterandc>it is a reference to the current package that is being defined.
<peanutbutterandc>.... but I can't seem to be able to access it inside phases.
<xelxebar>str1ngs: In fact, I believe I do need 4.9 with debug symbols.
<xelxebar>May I lean on your offer again?
<str1ngs>xelxebar: no problem, which expression do you want to use?
<xelxebar>This is really odd. But 4.9.234 just came out. I was able to grab it from and apparently it has symbols... O.o
<str1ngs>oh nice
<xelxebar>Just noticed.
<str1ngs>is 4.9 a LTS kernel?
<xelxebar>Sorry for flip-flopping a million times on you.
<xelxebar>str1ngs: Yeah, lts.
<str1ngs>xelxebar: now worries I have this 4.9 with debug expression I created so just ping me. or if you have your own expression
<str1ngs>xelxebar: I changed my substitute server so builds have a faster turn around time.
<xelxebar>Oh nice!
<str1ngs>I bought a Ryzen 9 3900X just so I create the qtwebengine package lol
<str1ngs>now it's useful for kernels and webkitgtk
<str1ngs>it would be interesting to create a secure a P2P network for offloading guix builds.
<xelxebar>That would, indeed, be interesting!
<str1ngs>I'm intrigued by P2P infrastructure for example this looks very cool
<str1ngs>P2P issues an pull request all rolled into one
<xelxebar>I find it encouraging that they are aware of pijul :)
<xelxebar>Looks pretty cool.
<xelxebar>I wonder if they are handling the project discovery aspect in a decentralized way as well...
<str1ngs>though the are rewriting radicle in rust instead of haskell and the UI uses electron. so they might have lost there way lol
<str1ngs>oh I never heard of pijul
<xelxebar>Huh. Haskell or rust in combo with electron gives me some cognitive dissonance.
<xelxebar>str1ngs: The main idea behind pijul is pretty nifty. It looks like a better patch model than darcs.
<str1ngs>I was just reading that. how distributed is it I wonder.
<str1ngs>the thing I like about radicle is users can submit issues/patches from the command line all via P2P
<str1ngs>so it lowers the barrier for contribution.
<xelxebar>The coolest aspect is the idea of *antiquing*.
<xelxebar>Yeah, pijul as a version control tool is decentralized just like git, so if radicle could support modular backends, radicle + pijul would be like VCS heaven. heh
<str1ngs>also hard to move from git it's so ubiquitous now
<xelxebar>In a similar vein, I'm a bit annoyed that zoom has become so ubiquitous overnight.
<str1ngs>ironically now that you mention zoom. I'm currently working on getting experimental support for webRTC so people can use things like jitsi with nomad.
<str1ngs>in webkitgtk that is.
<zzappie>xelxebar: omg zoom... people sending me links to zoom assuming that everyone use it...
<zzappie>most amazing part is when it says like: donwnload this executable and run it to join
<guix-vits>sneek: sneek?
<sneek>I could be wrong, but Sneek is just picky…
<guix-vits>Hello Guix. I want to use a make-linux-libre* in my config.scm. I included the (gnu packages linux) in my config. But when i try to use this variable, i gett error and suggestion: (use-modules (gnu packages linux)). I tried define with @@, as in Guile Manual, and that works. But manual tells not to use @@ (only for last resort and debugging). What to do?
<str1ngs>guix-vits: is the kernel only going to be useful to you?
<guix-vits> str1ngs: i just want to use a #:extra-options for my customizations. I'll report the things (as was suggested) that seems universally useful.
<guix-vits>#:extra-options is a lot better (?) that have a "alist something replace" with edited copy of kernel config.
<str1ngs>well yes but keep in mind if you are creating a kernel that other can use to might be better to add it to (gnu packages linux) or create your own channel
<guix-vits>str1ngs: Yes is to @@?
<guix-vits>Or there is a better way to use procedures that aren't public?
<str1ngs>for something adhoc @@ is fine to a degree. I think you can use module ref something . but the result is the same.
<guix-vits>OK, thanks.
<guix-vits>BTW: why make-linux-libre* isn't public?
<str1ngs>there is away module excursion and add your kernel dynamically to (gnu package linux) if you are into meta programming :P
<str1ngs>it's probably private to encourage people to contribute kernels
<guix-vits>Meta? Using i can use Alt key.
<str1ngs>meta programming. is programs the write programs
<str1ngs>guix-vits: here
<str1ngs>guix-vits: here's an example of save module excursion.
<guix-vits>Thanks, str1ngs. Will read.
<str1ngs>guix-vits: the main point is you can do set-current-module (resolve-module '(gnu package linux))) define your custom kernel. and it will leave after
<str1ngs>err (set-current-module (resolve-module '(gnu packages linux))) btw I have not tried this with guile 3
<guix-vits>str1ngs: it's like #include? Temporary copies things from module "above cursor".. err..?
<guix-vits>*but temporary
<str1ngs>'save-module-excursion with (set-current-module (resolve-module '(gnu packages linux))) is like entering the gnu packages linux module and adding things dynamically within that namespace.
<str1ngs>#include is not compatible
<guix-vits>str1ngs: *competetive?
<str1ngs>guix-vits: a better analogy is dynamically editting the file that #import includes.
<str1ngs>guix-vits: with scheme you are an artist modeling with clay. :)
<guix-vits>(my-avatar "dog-on-meth.jpg") ;; :)
*str1ngs can only draw a stick man
<str1ngs>so your dog is a chihuahua?
<guix-vits>str1ngs: sometimes. As this is a imaginary dog, it can be anything.
<str1ngs>ah? the makings of a Picasso
*guix-vits M-x query RET Picasso
<str1ngs>(define-key global-map (kbd "C-c s") 'query)
<str1ngs>C-c s Picasso
*str1ngs pets 🐕
<str1ngs>guix-vits: are you using nomad daily?
<str1ngs>nice, I think ibuffer needs more improvements.
<guix-vits>To C-f et all working in it?
<str1ngs>in ibuffer? should not need it really
<guix-vits>shortcut for rename()?
<str1ngs>that a good idea. also markings would be better then d
<str1ngs>I don't want to invest to much in ibuffer I'd like that to be inferiour-ibuffer then write an ibuffer with gtk controls. not use a text-buffer
<str1ngs>text ibuffer is more portable.
<str1ngs>back to guix though, does your definition work now?
<guix-vits>str1ngs: some syntax error: body should end with an expression. Looking..
<roptat>hi guix!
<sneek>roptat, you have 1 message!
<sneek>roptat, raghavgururajan says: Long ago, we were discussing OpenVPN service type on ML. I have a doubt on how to use ".pem" file. Would you be able to provide me an example code-snippet to use
<roptat>raghavgururajan, sorry it's been too long, the paste doesn't exist anymore it seems
<str1ngs>guix-vits: this might not work you sill need @@ :(
<guix-vits>str1ngs: Yes it seem to work. I think to place the (set-current-module ... and (define linux/simon at the top of the config.scm. Should i somehow unset it after (what was set with set-current-module)?
<guix-vits>(define my-kernel
<str1ngs>guix-vits: is not make-linux-libre better to use?
<guix-vits>str1ngs: IDK, the definitions in linux.scm use one with * (make-linux-libre*)
<str1ngs>guix-vits: this works for me
<str1ngs>guix-vits: forget save-module-excursion it won't help you if you still need to use @@
<guix-vits>str1ngs: currently i just use set-current-module (without save-module-excursion), define, and then place the rest of my config. Wrong?
<str1ngs>you don't need set-current-module either see my example
<guix-vits>str1ngs: OK. So, @@ is better to use there? Manual tells against.. call it "impolite method"..?
<str1ngs>@@ is not advisable though it might be okay in this ad-hoc way. better to ask in #guile
<str1ngs>ideally guix should provide a public interface. but I don't know enough about (gnu packages linux) to comment on this.
<str1ngs>the right way to do this is create a channel or add it to gnu/packages/linux.scm and use pre-inst-env.
<str1ngs>guix-vits: @@ is like having your dog locked inside your fence. but someone jumps over the fence and pets him anyways.
<guix-vits>str1ngs: #ilovelennart: (define vits-config (set-current-module (resolve-module '(gnu packages linux)))) (define linux-standing-in-hammok ...) (set-current-module vits-config) ...
<guix-vits>@@ and two short defines!? No way.
<str1ngs>guix-vits: no just use my example
<str1ngs>forget set-current-module
<str1ngs>assuming this is a config.scm
<str1ngs>I though save-module-excursion would get around @@ but it doesn't so you don't need set-current-module with @@
<str1ngs>I guess @@ is evil
<guix-vits>str1ngs: Thanks for explanations. BTW, don't miss the linux-libre-arm64-generic in linux.scm: it has #:extra-options:
<guix-vits>... #:extra-options (append `(("CONFIG_XYZ" . #t)) %default-extra-linux-options)))
<str1ngs>%default-extra-linux-options is not public despite the % prefix
<str1ngs>so you'll have to @@ that too
<str1ngs>this is why it's not good to use @@
<nckx>G@@d morning Guix!
<guix-vits>I'll get no warning about %default-extra-linux-opitons is unresolved?
<guix-vits>Hello nckx.
<guix-vits>sneek: nckx?
<sneek>Last time I checked nckx is a nice 👤️
<guix-vits>sneek: story-teller?
<guix-vits>sneek: ghost-of-king-lisandus?
<guix-vits>sneek: lil-timmy?
<nckx>sneek: botsnack?
<nckx>Oh dear, I'm a botsnack.
<guix-vits>nckx: vitssnack?
*nckx .oO we shall all be but botsnacks when the event comes.
*guix-vits not a bot. honestly.
<nckx>First-world problem but why has the emacs update changed my font size again 🤦
<str1ngs>guix-vits: WebRTC 1.0
<guix-vits><fat>A second question: Are You sure?</fat>
<str1ngs>guix-vits: what do you use that has webRTC jitsi?
<guix-vits>Yes, Jitsy
<str1ngs>okay I'll have to test with jitsy somehow
<str1ngs>epiphany doesnt even have webRTC :P
<guix-vits><fat>.. In Guix. How can we know for.. Debian for example?</fat>
<str1ngs>sneek: later tell guix-vits because I use ubuntu and it requirs a new webkit and branch
<sneek>Will do.
<pkill9>civodul: i don't suppose you'd know where to put a hook in guix code to execute a command before adding to the store with local-file? so it would execute the command when guix builds the source
<pkill9>I want to add bleeding edge packages that have source set to a local git repository, and have it execute `git pull` on the source repository before doing what 'local-file' does
<bdju>guix package: error: emacs-next: unknown package
<bdju>what's going on?!
<bdju>I know emacs 27 released now but then I would've thought emacs-next would be 28 or something. am I expected to just switch to regular emacs now?
<civodul>bdju: yes, for now at least; there are discussions to re-introduce emacs-next
<bdju>okay, good to know.
<civodul>pkill9: you can try using 'git-checkout', which performs a git fetch at the time you access it
<civodul>so instead of using (local-file ...), just write (git-checkout ...)
<str1ngs>sneek: later tell guix-vits jitsi seems to be working in Nomad. so looks like webRTC should work. need to test more though.
<sneek>Will do.
<pkill9>civodul: will that redownload it each time?
<civodul>it maintains a cached checkout in ~/.cache/guix/checkouts
<civodul>the first time it clones, and then it just fetches
<pkill9>ooo nice, that sounds like what i want
<pkill9>hmm, it says invalid field specifier
<pkill9>i might be using git-checkout incorrectly, probably
<civodul>(git-checkout (url "..."))
<str1ngs>xelxebar: looks is indeed linux-libre@4.9 on I have a copy cached on my substitute server just encase.
<pkill9>yes that was it, thanks
<joshuaBPMan>morning guixoids!
<civodul>sneek: later tell jonsger guile-ssh segfault reproduced!
<sneek>Will do.
<pkill9>that worked, thanks civodul
<pkill9>joshuaBPMan: there is already a way to make bleeding-edge packages, use as the source (git-checkout (url ...)) from (guix git) module
<joshuaBPMan>pkill9: ahh. That's cool. How well does it work?
<joshuaBPMan>Also I had a guix server working for a while, and I broke it already.
<pkill9>joshuaBPMan: it works exactly as i wanted it to work
<pkill9>it keeps a local git checkout and fetches updates each time guix builds the package
<joshuaBPMan>pkill9: that's pretty awesome. Which packages are you running on the bleeding edge?
<pkill9>joshuaBPMan: youtube-dl
<pkill9>that one is great for bleeding edge
<pkill9>as it's regularly updated for compatibility but doesn't change anything that would require new package arguments or inputs
<joshuaBPMan>pkill9: I suppose they have to constantly update it to be able function.
<oliverp>any advice on backing up a Linux system before attempting a fresh install of Guix?
<pkill9>and the stable release is a month old and doesn't work with youtube
<oliverp>there seems to be a few but conflicting options available..
<nckx>oliverp: Depends entirely on which parts you want to back up and how you want to restore them later. If you want to roll back as if the installation never happened, programmes like Clonezilla can efficiently (compared to dd) back up your entire disc including partition tables & file system layout. I don't know how easy it is to extract specific files out of such an image.
<nckx>If you just want to save ‘your files’ now to copy them to the new installation later: I just use rsync, or even tar | zstd | nc (faster but extracting individual files is much slower), to my home server.
<nckx>I wouldn't use a ‘real back-up’ solution like Borg for this, since they are intended for ‘real’ (scheduled, incremental) back-ups.
<oliverp>nkcx: thanks for your answer: I think mostly intrested in backing up the /home and the most important systemfiles
<oliverp>I see
<nckx>If you have the space & the bandwidth I'd just rsync/tar/whatever your entire root filesystem (as root so it can read everything, and maybe excluding /gnu and /var/guix to save time). You'll back up to much but no stress of finding out you forgot something later. Worth it.
*guix-vits e.. khm.. merges again.
<guix-vits>sneek: Was Geht's?
<sneek>guix-vits, you have 2 messages!
<sneek>guix-vits, str1ngs says: because I use ubuntu and it requirs a new webkit and branch
<sneek>guix-vits, str1ngs says: jitsi seems to be working in Nomad. so looks like webRTC should work. need to test more though.
<guix-vits>Meow. Cool.
<PurpleSym>Is there any way I can `guix pack` the most recent guix from git? (i.e. the one you get when `guix pull`-ing) --with-source does not work :/
<raghavgururajan>Hello Guix!
<guix-vits>Hello there.
<raghavgururajan>sneek, later tell roptat: Oh shoot! Here is the new link,
<sneek>Got it.
<guix-vits>raghavgururajan: use e-mail ;)
<raghavgururajan>guix-vits: I set not to expire this time. :-)
<raghavgururajan>oliverp: Oh hey o/
<civodul>PurpleSym: you want to run "guix pack whatever" with the latest guix, right?
<civodul>you can always do: guix pull && guix pack whatever
<civodul>or: guix time-machine -- pack whatever
<PurpleSym>civodul: No, my actual plan was to modify the install script, so it takes an arbitrary tarball. And I wanted that tarball to contain the git HEAD of guix, so I don’t need to run `guix pull` after installing.
<guix-vits>civodul: Hello. Can be Guix package manager packed in `guix pack`?
<PurpleSym>(because `guix pull` takes like 10 minutes)
<raghavgururajan>nckx: Nice! For the first time, I came to know about clonezilla. Btw, check this out,
<civodul>PurpleSym: ah! then you could probably do: guix pack guix --with-git-url=guix=master
<nckx>> packaged by most popular Linux distributions > doesn't mention Guix 😮
<raghavgururajan>nckx: Don't worry. I already took care of it. 😉
<nckx>You sent them a message?
<nckx>I'm happy to see such a healthy variety of Free back-up programmes.
<nckx>It can be confusing, but it's a complex topic with many use cases.
<PurpleSym>civodul: Nope, same error as with --with-source: guix pack: error: getting attributes of path `/gnu/store/mzfkrxd4w8vqrmyrx169wj8wyw7r8i37-bash': No such file or directory
<nckx>raghavgururajan: Oh, I was aware, thank you!
<civodul>PurpleSym: ah that's another problem, something must be wrong with your installation
*raghavgururajan sends a message to fsarchiver to include Guix
<raghavgururajan>nckx: Cool! :-)
<nckx>I haven't used it myself, is all.
<PurpleSym>civodul: You sure? That bash belongs to bootstrap/bash as far as I can see.
<civodul>PurpleSym: right, but in general, 'guix' commands cannot refer to store items that are not actually present in the store
<civodul>(or that's a bug :-))
<civodul>however, that can happen if your store or store database is somehow "corrupt"
<PurpleSym>Hm, okay, I recently GC’d a bunch of stuff. Let me see if I can --repair the items referencing it…
<civodul>which itself can happen if you passed the wrong --localstatedir in the manual build of Guile (the manual has something about it)
<peanutbutterandc>I am trying to package something from git that, on it's README, has an invocation to `git submodule update --init` before starting the actual build
<nckx>‘The standards say the gun must be aimed at the feet by default.’
<roptat>raghavgururajan, I think you can use the "ca" option in the openvpn-client-configuration
<sneek>roptat, you have 1 message!
<sneek>roptat, raghavgururajan says: Oh shoot! Here is the new link,
<peanutbutterandc>Now, I am trying to add a phase to do that. But I have a feeling it should go BEFORE set-SOURCE-DATE-EPOCH
<peanutbutterandc>which means that I have to set PATH and everything
<civodul>nckx: true, the standards are very clear in that respect :-)
<roptat>raghavgururajan, I don't think the pem extension is going to cause any issue to openvpn
<nckx>I know 🙂
<peanutbutterandc>is there any word of caution/etc that anybody would like to give me regarding that?
<peanutbutterandc>Do I have to unset PATH or something, once that phase is done?
<civodul>nckx: but seriously, i think ./configure detects such misconfigurations rather well
<civodul>so dunno if that's still a problem
<roptat>peanutbutterandc, I don't think that'll work, git submodule update will try to download stuff, but the build environment doesn't have network access
<nckx>Someone else has been here ‘recently’ (last few months?) with a similarly borked system.
<peanutbutterandc>Oh wait, perhaps there is something else! Maybe I can do the submodule update thing in (origin)
*nckx tests what ./configure says.
<peanutbutterandc>roptat, I see. Perhaps the (origin ...) part of the package has something. I will go and look at the documentation. Please do let me know if there is something else
<roptat>peanutbutterandc, sounds more reasonnable
<roptat>although I'm not sure if git submodule will give you something that's reproducible
<peanutbutterandc>roptat, It's a certain tag that I am building from. And I was reading this morning about git submodule and it seems it only downloads the submodule at a particular commit (as defined by the devs of the superset repo)... so maybe?
<nckx>Can't we work around the standard somehow? Similar to
<raghavgururajan>nckx: Done!
<nckx>Hehe. Ballsy.
<nckx>(☝ the guy can be annoying as hell, but he tells a good story.)
<raghavgururajan>roptat: I see. Would you be able to provide me an example snippet that goes into the services section of config.scm?
<raghavgururajan>nckx: Me?
<nckx>No, raghavgururajan, not you. Bryan Cantrill.
<raghavgururajan>nckx: Ah I see.
<nckx>He is not a fan of the GPL, and prefers rhetoric to argument.
<peanutbutterandc>Okay wow this is cool. So inside of (origin, under (snippet, I can have (use-modules (guix build utils)) and then do (invoke "stuff") there. Cool.
<raghavgururajan>Do we have dstrace,
<nckx>Yep! You can transform the source arbitrarily. In Guix, it's mainly used to remove problematic (non-source or non-FSDG) content.
<nckx>Lol, someone's been researching Brian.
<raghavgururajan>Yeah !!
<nckx>Linux + dtrace is a whole saga. It never happened. bpftrace is sort of equivalent now.
<nckx>I think the only dtrace-proper port is Oracle's CDDL one and I don't know if it's maintained.
<nckx>Nor if it's as ‘good’ as the BSDs'.
<raghavgururajan>I see.
<nckx>Didn't you package sysprof? It uses bpftrace to provide dtrace-like features.
<raghavgururajan>Yes I did
<nckx>bpftrace is a bit of a messy mess (typical for Linux) but it does work.
<raghavgururajan>Ah I see.
<raghavgururajan>I didn't know that. I assumed it used strace
*nckx exchanges dirty looks with CONFIG_IKHEADERS.
*raghavgururajan poutine with extra cheese-curds
<nckx>Religious debates (‘is poutine “food”?’) are off-topic in #guix.
<PurpleSym>civodul: You were right, my store was somehow corrupt and it built the tarball now. Thank you :)
<nckx>raghavgururajan: You were right and I was mistaken, Sysprof doesn't use bpftrace. I wonder what I'm confusing it with.
<raghavgururajan>nckx: Ah, no worries
<raghavgururajan>You were joking about poutine right? 😅
*raghavgururajan double-checks with
<peanutbutterandc>Okay... I am slightly confused. It seems I can (invoke "git") inside of (origin (snippet ...))... but then I checked the (envion) inside of it, and it seems that it has the same environment as my machine. So essentially, it is invoking git from 'outside' - my normal PATH
<peanutbutterandc>that isn't very cash money (reproducible)
<peanutbutterandc>so, that idea of invoking `git submodule update --init` from (origin ...) is a failure.
<peanutbutterandc>any other ideas, please?
<leoprikler>collect a lot of git references and stitch them together?
<peanutbutterandc>leoprikler, Hmm... how would I go about doing that please? This sounds like some major wizardry
<peanutbutterandc>can't believe this is the first time any guix packager has stumbled upon submodules. But a quick `grep submodule *` in gnu/packages does return very promising results
<peanutbutterandc>I think I'll go for adding a new phase before all-other-phases trick then... for now
<nckx>raghavgururajan: Of course. Enjoy!
<leoprikler>git submodules are only seldom used in a way that meshes with guix
<raghavgururajan>nckx: Yum Yum Yum!
<leoprikler>they are usually either first/third party software, that often deserve their own packages
<leoprikler>only rarely it's one project scattered over many submodules, that don't pull in anything else
<peanutbutterandc>leoprikler, I am currently trying to do something like: 1. have `git` as a native-input, 2. add a new phase before the very first phase, in which invoke git submodule in (assoc-ref %build-inputs "source")... Hmmm... I should probably consider using the submodule as an input then.... hmm
<leoprikler>no internet access during build for you
<leoprikler>you lazy little
<peanutbutterandc>Ah... no internet access during build.
<peanutbutterandc>okay. I need to figure out something else then
*civodul learned that current Nix doesn't have search paths, generated etc/profile, and "profile hooks":
<peanutbutterandc>okay. I don't have any ideas.
<peanutbutterandc>Maybe I should request a feature? for (guix git-download) to make provisions regarding git submodules too? Perhaps a git-fetch* or git-fetch-recursive method, that also initializes all submodules?
<leoprikler>there already is recursive checkout
<leoprikler>but again: think twice before using it
<roptat>peanutbutterandc, use recursive? in git-reference?
<peanutbutterandc>roptat, is there an example? or, which section of the manual describes it? o.O
<peanutbutterandc>also what does it do? (looking it up myself too)
<peanutbutterandc>Ooooh! (url ...) (commit ...) (recursive? ...) Cool thingy!
<peanutbutterandc>Everytime I hit what I think is a wall with guix, it turns out it isn't a guix-wall, rather just a documentation wall. Thank you very much!
<roptat>yes, it recursively checks out sub-modules (according to the comment in (guix git-download))
<peanutbutterandc>roptat, Yes, I just read that too! (:
<peanutbutterandc>Now I can rule the galaxy
<roptat>I wonder if we even properly document all these build-side functions
<roptat>actually, do we document any?
<peanutbutterandc>roptat, some of these build side function's documentation are avilable via guix repl if I ,use (ice-9 session) and then (help stuff) but not this one, it seems
<joshuaBPMan>hey guix people!
<joshuaBPMan>hmmm, getting certbot to work is a little harder than I thought it would be.
<civodul>joshuaBPMan: hey! you're using the cerbot service?
<joshuaBPMan>civodul: trying to.
<joshuaBPMan>that's my current server configuration.
<joshuaBPMan> works, and my guix system is serving it. Now I'm trying to set up https.
<joshuaBPMan>certbot tells me that when the ACME server tries to verify my server, nginx is saying 404-not found.
<peanutbutterandc>okay, it seems that even when I am using (recursive? #t) in (uri (git-reference ...) ...) in (source ...) for a package, the checkout does NOT have the submodules initialized
<nojr>is anyone having issues with counsel-mode after upgrading emacs ? my error is that it won't find "counsel"
<civodul>joshuaBPMan: ah you probably miss an nginx "location" block for /.well-known, no?
<joshuaBPMan>civodul: I suppose... I thought the cerbot configuration did that for me? am I mistaken?
<civodul>yeah you have to DIY :-)
<civodul>as in
<nojr>Debugger entered--Lisp error: (void-function counsel-mode)
<nojr>how ??
<joshuaBPMan>civodul: thanks for the tip bro!
<nojr>seems it's an issue with ivy-rich
<joshuaBPMan>civodul: thanks again. I'm going to give that a try.
<peanutbutterandc>I am not sure what I am doing wrong, but the git-checkout does not seem to have it's submodules initialized. Can somebody please take a look at this package definition?
<peanutbutterandc>It fails during 'configure phase. But currently I am only worried about (git-reference not respecting (recursive? #t)
<joshuaBPMan>civodul: thanks! I now have a certificate!
<joshuaBPMan>peanutbutterandc: are you sure it is not recursive? just guessing
<joshuaBPMan>nojr: is it installed via guix ? as guix install emacs-counsel-mode ?
<peanutbutterandc>joshuaBPMan, Affirmative, captain. I have a separate checkout on my machine, and I manually did the submodule init and src/lib/libfoo becomes populated with a few things, however, /gnu/store/this-package.checkout does not have the same directory populated
<peanutbutterandc>joshuaBPMan, Also, I am intrigued by .... is that just a test?
<joshuaBPMan>civodul: thanks soo much man! You've really saved me! And I'll have to look at that berlin.scm file. There's soo much there that I could learn.
<joshuaBPMan>peanutbutterandc: well thanks super annoying major general. Have you tried shooting at the enemy? :)
<joshuaBPMan>I wonder why it is was not populated...
<nojr>joshuaBPMan: it's bundled with ivy, it seems there were some updates that I missed and now several functions are not working like swiper
<peanutbutterandc>joshuaBPMan, If you would like to give it a shot, please do:
<joshuaBPMan>peanutbutterandc: I am hoping to turn it into a business. I think the reason the linux kernel works so well, is that most of its developers are paid to develop it. I think the Hurd would probably make great strides if we could turn it into a business.
<joshuaBPMan>peanutbutterandc: what are you hoping to make with slow mode video?
<peanutbutterandc>joshuaBPMan, Whoa. Hurd is ready for production use already? o.O I thought guix-hurd still had a long way to go... but this seems like a super cool thing nevertheless
<peanutbutterandc>joshuaBPMan, I poked around with the program once. And now I just want to use it to learn guix packaging. But it's a really great program. I don't know. Probably some random slowmo video
<joshuaBPMan>peanutbutterandc: fair enough.
<joshuaBPMan>just a moment...I'll give it a shot.
<joshuaBPMan>please bear in mind that I only have two commits to guix, and both were documentation fixes....
<joshuaBPMan>though I do brag to friends that I am in the guix commit history. that if I had nerdy friends. Most of my friends are not comptuer nerdy.
<nojr>BTW sounds cool B)
<peanutbutterandc>joshuaBPMan, I have been poking around with that package definition for hours now, and I am desparate.
<joshuaBPMan>peanutbutterandc: so I get a wget to your url.
<joshuaBPMan>mv 9ped to 9ped.scm
<joshuaBPMan>guix package -f gped.scm
<joshuaBPMan>guix package: error: cannot install non-package object: #<unspecified>
<joshuaBPMan>are you getting the same error?
<joshuaBPMan>I think you are supposed to put the package object at the end of the file. as in end the file with "\nslowmovideo"
<peanutbutterandc>joshuaBPMan, You should `guix build -L /path/that/contains/9ped.scm/but/not/the/file/itself --verbosity=2 slowmovideo`
<joshuaBPMan>peanutbutterandc: ahhh. that makes more sense.
<peanutbutterandc>so if 9ped.scm is in ~/Documents, `guix build ~/Documents --verbosity=2 slowmovideo`
<nojr>I just read the docs at ivy's GH page and ther is no indication counsel-mode is removed, yet I cannot invoke it in my init file, this happened right now after I upgraded all my emacs related packages
<joshuaBPMan>nojr: I'm pretty sure counsil mode is still there.
<joshuaBPMan>is it located in .emacs.d/elpa/consel-mode/ ?
<joshuaBPMan>Or is it stored in /gnu/store somewhere?
<nojr>joshueBPMan: I can't call it in my init file says its definition is void
<nojr>it's installed via guix so in /gnu/store
<joshuaBPMan>nojr: take a look at your messages....
<joshuaBPMan>C-x b *messages*
<joshuaBPMan>C-x b *Messages* in emacs
<rndd>hi everyone! where i can find
<rndd>tput command
<joshuaBPMan>up toward the top you should see, Loading
<joshuaBPMan>autoloads...try to open that file.
<joshuaBPMan>maybe that file will tell you what function is autoloaded.
<joshuaBPMan>peanutbutterandc: it's building
<joshuaBPMan>lots of warnings.
<peanutbutterandc>joshuaBPMan, May I see some of the warnings please?
<joshuaBPMan>I think some of the errors are due to my packages...
<joshuaBPMan>it's talking about jmacs, which is a package I created...
<joshuaBPMan>that's the tail end of the errors that I see.
<nojr>joshuaBPMan: it's fixed by installing emacs-counsel, I think before they were bundled together
<joshuaBPMan>nojr: awesome!
<peanutbutterandc>joshuaBPMan, I don't think it's slowmovideo.... mostly something else. not slowmovideo
<peanutbutterandc>it's all right. Thank you nevertheless for trying to help me out
<joshuaBPMan>peanutbutterandc: I wish I could be more helpful, but I'm not quite a guix developer yet. :)
<peanutbutterandc>joshuaBPMan, It's all right. You were talking with Mr. Courtes just a while ago about hurd and stuff. I'm sure you'll get there soon enough
<joshuaBPMan>peanutbutterandc: thanks for the vote of confidence!
<rndd>bytheway found tput in ncurses
<rndd>did anybody tried to pack common lisp?
<leoprikler>joshuaBPMan: your error looks like you've set your GUILE_LOAD_PATH to something completely wrong
<NieDzejkob>rndd: tput is provided by ncurses
<joshuaBPMan>leoprikler: oh...good to know.
<rndd>NieDzejkob: sure, anyway thank you. do you know anybody who i can ask about packaging common lisp in guix?
<joshuaBPMan>rndd: I'm surprized that we don't have common lisp packaged.
<rndd>joshuaBPMan: sorry, i wrote it wrong. I mean packaging common lisp packages
<rndd>i hade an issue with one library
<rndd>i have*
<joshuaBPMan>rndd: ahhh. I don't know if we have a standard for creating common lisp packages...
<rndd>joshuaBPMan: asdf-build-system
<bdju> got this build failure building emacs-clojure-mode-5.12.0
<bandali>bdju, known issue:
<bdju>bandali: alright, thanks.
<bandali>no problem
<bdju> got this build failure building hedgewars-1.0.0 (warning, very large)
<leoprikler>sneek, later tell peanutbutterc you give me too much credit, origin inputs are used in all of guix
<leoprikler>sneek, later tell peanutbutterandc you give me too much credit, origin inputs are used in all of guix
<leoprikler>typing names is hard :(
*leoprikler → afk
<NieDzejkob>does your client not have tab completion for them? I can type le<TAB> and get leoprikler autocompleted
<NieDzejkob>hell, even l<tab> works because you spoke so recently
***Guest53005 is now known as KE0VVT
<civodul>joshuaBPMan: glad it worked! looks like, let's hope you can offer Guix GNU/Hurd hosting :-)
<joshuaBPMan>civodul: I hope so. I hope to turn it into something. If there are any aspiring hackers that want to help, then please let me know. I only own I'm open to any suggestions.
<rupicapra[m4>I feel a bit more confident in my scheme (and general programming) skills so I am back to guix!
<str1ngs>welcome back rupicapra[m4 !
<rupicapra[m4>But I will still have a lot of questions in due time! :D
<bdju>sneek: later tell joshuaBPMan I think I found a typo on it says "ganster" which should probably be "gangster", also "privledge" should be "privilege"
<sneek>Will do.
<bdju>what has emacs-clojure-mode as a dependency? I tried to upgrade all things except it and it didn't work. I also realized I never put it in my manifest
<lfam>You could use `guix refresh --list-dependent emacs-clojure-mode` to check
<bdju>thank you. I see emacs-evil-collection in the list, which is in my manifest
<lfam>There is also `guix graph --type=reverse-packages emacs-clojure-mode | xdot -` if you want a visual representation
<lfam>And `guix package --search=. | recsel -e 'dependencies ~ "emacs-clojure-mode"' -p name` if you want to use recutils
<lfam>`guix package --search=. | recsel -e 'dependencies ~ "emacs-clojure-mode"' -p name,synopsis` might be more useful
<bdju>wow, that graph thing is very cool
<bdju>thanks for the tips, I'll write these down
<lfam>Yeah! It's not as useful when there are a lot of nodes in the graph, which happens often
<lfam>But it's nice to have
<lfam>Xdot is really nice. I learned about it recently, after years of rendering the graphs as images
<joshuaBPMan>Hey #guixoids
<sneek>joshuaBPMan, you have 1 message!
<sneek>joshuaBPMan, bdju says: I think I found a typo on it says "ganster" which should probably be "gangster", also "privledge" should be "privilege"
<joshuaBPMan>bdju: thanks for the spell checking!
<bdju>you're welcome :)
<joshuaBPMan>I also got have my guix system serving two websites! That's awesome!
<leoprikler>NieDzejkob: but what if they're logged off tho?
<leoprikler>I'm not using Emacs for IRC, so I don't have dabbrev
<nckx>joshuaBPMan: Cool! Needs www. support though 😉
*nckx was confused by the reported typoes above being nowhere to be found on