IRC channel logs


back to list of logs

<formbi>how to use lbzip2 in a package definition?
<formbi>I mean how to use it to uncompress the source tarball
<lechner>nckhexen: sorry, the point was to paraphrase; i was not sure that mentioning the other channel was allowed here
<two[m]><formbi> "I mean how to use it to uncompre..." <- try this... (full message at <>)
<two[m]>replace "lbzip2" "--unpack" with the correct command
<nckhexen>florhizome[m]: Both to debbugs? Hmm. I'm also very tired & must be missing something. Night!
*nckhexen → 😴💤
<mroh>gn8 nckhexen
<two[m]>* replace (string-append (assoc-ref inputs "lbzip2", * "lbzip2") ", * "/bin/lbzip2") "--unpack" with
<two[m]>(corrected since mroh's message)
<two[m]>> <> I mean how to use it to uncompress the source tarball... (full message at <>)
<two[m]>and add libzip2 to inputs
<mroh>and add a (list ...) after arguments, around #:phases (or `).
<lechner>Hi, what's a rational way to manage the system configuration of multiple machines, hopefully using (use-modules) to share code, without accidentally applying a wrong configuration somewhere?
<taijiavila>Hi, I need some help with guix translation. I am in weblate
<taijiavila>I have a question regarding @result{} #<unspecified>
<taijiavila>Should I translate it? Like @resultado{} #<sin especificar>
<daviid>taijiavila: @result{} is a texinfo command, you can't translate it
<taijiavila>And the next part?
<daviid>taijiavila: i'll leave guix folks answer the second part, but i would keep it as is, because that's a guile 'message', that is what guile will display no matter what locale the user is using, i think (?)
<taijiavila>Ok, thx!
<taijiavila>Let's see if someone from translation team is present
<florhizome[m]>*nckhexen yes to debbugs
<jab>This seems pretty cool:
<jab>no-js forum software.
<apteryx>what would be the most minimal terminal that guix carries, next to xterm?
<apteryx>in terms of size closure
<vagrantc>wild guess would be rxvt* ?
<vagrantc>might even be smaller than xterm
<apteryx>rxvt-unicode is pretty small indeed
<apteryx>smaller than xterm
<iyzsong>guix size reports that -> st: 94.9, xterm: 118.6, rxvt-unicode: 173.0, foot: 177.6 (MiB)
<apteryx>how can I set my system timezone to UTC?
<apteryx>is 'Etc/UTC' the one to put in /etc/timezone?
<apteryx>(via the 'timezone' field of the OS)
<jackhill>anyone else having problems with some gnome stuff segfaulting after the recent updates? On one comperter Files (nautalus) has the problem and gnome-control-center on another. `guix gc --verify=repair,contents` turns up clean on one computer, but repeadetly "repairs" spice on the second.
<jackhill>also, why would `guix gc --verify=repair` not be able to repair an item?
<apteryx>lechner: perhaps guix deploy? and making the common stuff shared as common modules
<jackhill>wooo, I got rid of the problematic spice (it was from an old generation), but gnome-control-center still segfaults
<lilyp>jackhill: repair chokes on grafts iirc
<lechner>apteryx: thanks! that looks cool. i hope that command is limited to 'root' on the coordinating machine
<jackhill>lilyp: ah. well, ideally I wouldn't have to repair, but…
<nckhexen>apteryx: That should work, but presumably you tried it already.
<Lumine>Good morning
<sektor[m]>How's it going?
<Lumine>Just made coffee
<Lumine>And figured out why circe was acting out
<Lumine>How are you?
<civodul>Hello Guix!
<sektor[m]>Not too shabby. How do you tell a service it depends on the audio system bding up?
<Lumine>Nice, nice. Not too shabby that is.
<hnhx[m]>Do any of you guys use Guix on a laptop? How could i improve battery life?
<hnhx[m]>My T60's battery life isnt the best, i want to get the most out of it
<civodul>hnhx[m]: hi! you could try running 'powertop'
<civodul>that gives hints as to what's consuming energy
<hnhx[m]>i use a very minimalistic dwm setup so its not any of the software that is causing issues im sure
<hnhx[m]>it just has a bad battery
***robin_ is now known as robin
<efraim>perhaps the tlp service? the T60 is a lenovo laptop.
<abrenon>hey guix
<unmatched-paren>hnhx[m]: I use both tlp and thermald, they both seem to help a bit
<florhizome[m]>and thinkfan
<unmatched-paren>florhizome[m]: presumably that only works on thinkpads?
<florhizome[m]>I discovered nbfc, which I am trying to send a patch for, which is more general though.
<unmatched-paren> <- ah, looks like there isn't a framework configuration yet
<florhizome[m]>yeah there isn’t one for my laptop either, I use a different one atm
<sektor[m]>My laptop really does need a new battery.
<sektor[m]>In a bad way.
<hnhx[m]><unmatched-paren> "hnhx: I use both tlp and..." <- thanks
<hnhx[m]>i will try it when i get home
<unmatched-paren>hnhx[m]: keep in mind that thermald is Intel-only
<hnhx[m]>ye my t60 has an intel cpu
<hnhx[m]>T7200 to be specific
<unmatched-paren>hnhx[m]: you should be able to just add (service tlp-service-type) and (service thermald-service-type) to your system services
<unmatched-paren>(they're exported by (gnu services pm), btw)
<hnhx[m]>and i dont need to config anything?
*unmatched-paren opens their system config to check
<unmatched-paren>no furthur configuration required..
<ss2>I did:
<ss2>I managed to reduce the SSD's temperature by 10°C this summer.
<ss2>and the mainboard cooled down a bit too, and effectively managed to reduce the fan speeding up too.
<unmatched-paren>ss2: how did you know which options would be optimal? :)
<ss2>I wouldn't recommend thinkfan though.
<ss2>reading and testing until the system actually became quite unstable.
<hnhx[m]>unmatched-paren: cool
<ss2>unmatched-paren: <- the docs are an easy read too.
<unmatched-paren>hnhx[m]: i should have said "no configuration required, unless you want to fine-tune it like ss2 did" :)
<hnhx[m]>ye i think i will go with default settings
<hnhx[m]>also i think i should get a new battery, even though the one i have rn is new too, but it only lasts for like an hour so i kinda feel like i got scammed
<ss2>med_power_with_dipm isn't a default, and may be unstable. But it is the most effective in keeping SSD's in a passive state while keeping the temps down. 10°C alone was a lot I thought once I observed the system cooling down by itself.
<ss2>unmatched-paren: I also got the tp-smapi-module in to the repos too. That's the module where you can manually take control over thinkpads bettery charge. Pretty cool module.
<ss2>with that you can sudo tlp fullcharge, or discharge your battery on a whim.
<ss2>But I'm still wondering why Debian has got a better config still, because the fan there spins down even further.
<unmatched-paren>I'm gonna attempt a home-dbus-service-type
<unmatched-paren>because a mako service would need to depend on it...
<Kabouik>"deleting `/gnu/store/trash'" is busy since about 5-10 min with very high CPU usage after a `guix package --delete-generations-1m && guix gc`, should I worry?
<Kabouik>--delete-generations=1m sorry
<civodul>Kabouik: that phase takes a while (though it's faster if you run a recent guix-daemon)
<civodul>ah no, sorry
<unmatched-paren>morning civodul :)
<civodul>i was thinking about the "deleting unused links" phase
<civodul>hey unmatched-paren :-)
<Kabouik>I'm not too worried because it's only /gnu/store/trash, but it's still busy with high cpu usage
<civodul>Kabouik: anyway, probably nothing to worry about
<unmatched-paren>what does /gnu/store/trash contain, anyway?
<Kabouik>3600 folders that guix package --delete-generations=1m moved there (I suppose?)
<Kabouik>"I supppose" meant that I assume that it's guix package --delete-generations that put them there, I didn't mean that as a pedantic reply at all, sorry if that's how it came out
<Kabouik>It's just that I'm not sure, it's my first --delete-generations ever.
<unmatched-paren>Our dbus manpages appear to be a bit broken.
<f3n1x>Given my current ongoing task (Python developement), i'm prompted to make use of the '-devel' Python package. Fine. In guix, would it be 'python-devtools' ?
<unmatched-paren>f3n1x: no, it looks like python-devel is headers and stuff for Python's C extension interface
<unmatched-paren>Ah, right, of course, I forgot the -devel convention was for headers on Red Hat-based distros...
<unmatched-paren>the ``python'' package should just come with the -devel stuff on guix
<sektor[m]>And the kernel kept compiling.
<sektor[m]>Almost an hour.
<f3n1x>Thxs,thxs, thxs unmatched-paren
<linj>I am new to guix. When reading, I says "Rust is currently unavailable on non-x86_64 platforms". Is that still true?
<unmatched-paren>linj: yes
<linj>s/html_node/html\_node/, s/I/it/, s/x86_64/x86\_64/
<unmatched-paren>The build takes so much memory, it OOMs on x86
<linj>I think you have aarch64 machine, no?
<unmatched-paren>and mrustc doesn't yet support anything other than x86{,_64}
<unmatched-paren>(we build rustc from mrustc)
<florhizome[m]><ss2> "reading and testing until the..." <- For my x230 it helped a lot to keep temps down
<nckhexen>Weird sed reported to the sed police:
<attila_lendvai>i'd appreciate if someone knowledgable could look at why `gpaste-client --version` dies:
<unmatched-paren>nckhexen: i don't believe the _ -> \_ seds are bogus; maybe the _ showed up as italics on matrix, so they edited it to escape them?
<nckhexen>Hmm. (_ is italic now? In my golden antebellum years it meant underline, but fine.) Which brings us to: Elements ‘history’ view… isn't:
<nckhexen>Note the ‘it’, twice.
<nckhexen>So it's not italic/underline/blink there, but then that view is probably bogus to begin with.
<unmatched-paren>nckhexen: I believe matrix uses markdown syntax. But yeah, I stand corrected.
<nckhexen>Oh, wow, instareply on GitHub.
<ss2>Underline can also mean italics on terminals that can not render italic fonts.
<nckhexen>We digress however.
<nckhexen>Seems like both bugs are known. So that's something.
<nckhexen>unmatched-paren: Heh, I hit the GitHub italic button and yep, underscores. Markdown is silly.
<unmatched-paren>nckhexen: You can also do *...* for italics. For some reason.
<unmatched-paren>**...** or __...__ is bold.
<unmatched-paren>Silly indeed.
<nckhexen>That's the one I knew, and used when I was typing the original ☺ And to think this was meant as a (ill-conceived) statement against ‘complex’ mark-up languages…
<nckhexen>That == **.
<nckhexen>Er, *.
<nckhexen>I could have used sed for that but that would have been cruel.
<nckhexen>ss2: Which ThinkPad do you have? I didn't know that TLP could set charging thresholds. I've been manually invoking tpacpi-bat at startup all this time.
<unmatched-paren>sneek: later tell muradm: yay, greetd-wlgreet-sway-session has been merged into master! :D
<sneek>Got it.
<unmatched-paren>sneek: botsnack
<unmatched-paren>Kabouik: ...and it looks like your cobib package has been merged too :)
<Kabouik>Oh nice, I didn't check my emails and missed that. Great. Looking forward to having nmail merged now. :> (I already use my local version)
<sektor[m]>Now to figure out how to debu/play with services in the pre-inst repl.
<unmatched-paren>I just have my Sway workspace 1 set up with qutebrowser on the right, senpai at the top left, and aerc at the bottom left, so I usually see the merge emails quite quickly.
<Kabouik>My typical workspaces are nmail1/nmail2 on WS1, Icecat or Nyxt on WS2, and some other stuff on WS3 so I also usually see emails immediately
<Kabouik>But doay nmail2 was not running because I'm late at work and needed to minimize distractions
<Kabouik>Turns out I still have IRC somewhere… :>
<unmatched-paren>Odd, my laptop just crashed even though it's at nearly 100% battery charge.
<nckhexen>I didn't know that made them crash-proof, I'll have to stop using tpacpi-bat now 😉 How did it crash?
<unmatched-paren>No idea.
<unmatched-paren>Just suddenly shut down for no reason.
<unmatched-paren>Ooh, looks like someone had the same thing and it seemed to have happened "at the exact time TLP kicked in to poll the power settings".
<xd1le>hi guix
<unmatched-paren>xd1le: Hello.
<nckhexen>I'd recommend pstore, but actually I don't because I still suspect it was to blame for eventually deadlocking my UEFI NVRAM.
<nckhexen>Hi hi.
<nckhexen>unmatched-paren: Oof.
<nckhexen>(How would one know that?)
<unmatched-paren>nckhexen: They checked the logs.
*unmatched-paren doesn't know which logs they checked though.
<nckhexen>A very polite crash that flushes the logs. I want those.
*nckhexen is describing a working pstore, sigh. Has anyone used it recently?
<nckhexen>I tried pointing it at a partition but that failed.
*unmatched-paren does ``guix search pstore'' -> nothing...
<ss2>nckhexen: x61s, not all modern models are supported:
<nckhexen>Right, smapi doesn't support mine (which is why I never bothered adding it alongside acpi-call-linux-module, sorry you had to 😛). Part of me would be surprised if TLP didn't support $any_backend though, but an equal part of me wouldn't.
<ss2>It's perfect on my old machines.
<nckhexen>You still find decent batteries for them? That's nice.
<ss2>Nope, unfortunately not. I just asked for a refund of a third brand.
<ss2>And a new one from China.
<ss2>*just ordered
<ss2>I'm seriously considering to get into the business myself on replacing the batteries. But for that I need a (fireproof) workshop.
<nckhexen>I have (extremely briefly and hypothetically) considered trying that…
<nckhexen>I'd rather pay someone but that doesn't seem to be an option.
<nckhexen>I've had extreme luck (that's it really) with the eBay stock I bought so far, but my model's not as old as yours is yet. That won't last.
<ss2>There's not a high enough demand to do it. I'd be qualified in doing such things, but I'm not trained well enough for it.
***wielaard is now known as mjw
<unmatched-paren>Alright, I've disabled TLP. Let's hope that doesn't happen again :)
<unmatched-paren>I did have an issue once with TLP on my last laptop, when I tried using it on Ubuntu; the kernel panicked when I shut it down.
<nckhexen>ss2: Oh, I meant a friend or acquaintance. I agree it's not a viable business. Have fun even shipping them.
<nckhexen>unmatched-paren: OK, you've made an excellent case not to bother trying TLP after all, thank you.
<unmatched-paren>I think I'll stick to thermald, which has the rather significant advantage of being under the intel/ organisation on GitHub
<unmatched-paren>nckhexen: Eh, it'll probably work much better on a ThinkPad. And I'm not even sure it was TLP that caused that.
<unmatched-paren>Besides, I used it on Guix System on my last laptop for ages with zero issues.
<ss2>I'm pleased with tlp. But yeah, I did mention that it is trial and error to get it right. You're system will crash. It gets even worse with docking stations. I was determined enough that I could at least keep my laptop switched on in this summer's heat wave.
<nckhexen>Okido. I'll let you know if I ever try it, and especially if it goes boom.
<nckhexen>Silly question, but why do you both use these services to begin with?
<ss2>To run them at a lower temperature. Nothing else.
<nckhexen>Does your machine overheat without them? Significant batter life? Something else?
<nckhexen>* Significant batter bettery life.
<ss2>Try idling a laptop that was designed for room temperature at 30--35
<nckhexen>ss2: So it's not that the machine shuts down if you install a distro and just start using it.
<ss2>°C room temperature.
<nckhexen>Well, I'd (perhaps naively) expect that to feed right back into the temperature & hence feedback loop, and it would just downclock sooner, possibly always. I don't really understand why the core temperature sensors would ‘care’ or make assumptions about how much heat they *could* evacuate… But then I'm no TDP science man.
<ss2>no, they'd get so hot you couldn't touch them anymore. Like 45--50°C around the case. With TLP it all went down to 39,40. Surprisingly the biggest difference was with the SSD.
<nckhexen>Interesting (your SSD comment was what prompted my original message. I'm pretty sure I have the same SATA link state by default, but with those results it now bears double-checking.
<ss2>the problem with, for instance my model, is that the sensors where designed around the fact of the machines themeselfs heating up, but the fan is not designed to remove the heat from the whole case. Only the CPU. Which is just a design error, unless you sit in an air conditioned room on a cool table.
<nckhexen>Right. I guess that was my real question, and I was indeed a bit naive.
<ss2>yeah, I never realised that SSDs can be such radiators. But that has to do with the controller not downclocking.
<ss2>tbh, the more modern laptops from the last five years have gotten significantly better, since the main goal has been to keep the battery running for as long as possible.
<ss2>that wasn't the case 10 or 15 years ago. The goal there was more throughput and a bit powersaving.
<sektor[m]>Ther we go, making progress. My problem was that I misheard #~ as #`; not so sure how that happened.
*nckhexen nods.
<ss2>It makes a difference though. With my settings, my battery would last 6 hours (okay, the quality lacked) from 3,5 hours.
<nckhexen>‘The quality lacked’? Performance?
<unmatched-paren>ss2: Ah, makes sense that you'd want to get the temperature down as low as possible if you live in a place with summer heatwaves. Fortunately, we don't get serious summer heatwaves (yet; thanks, global warming...) here (in the west of Scotland, not known for warm temperatures :P)
<ss2>nckhexen: "dodgy battery replacements", that's why.
<nckhexen>When I was (extremely reluctantly) choosing a 'phone, it was very interesting to read vastly different reviews on performance & throttling and after a while realising there was a geographic pattern. Mostly India. The subcontinent murders bad cooling, apparently.
<nckhexen>ss2: Ah.
<ss2>unmatched-paren: yes, that's the reason. I've been caught in crazy heat waves the last years now and since I've turned my focus on to keeping machines just as cold as possible.
<ss2>Even having to remove plastic modems from being screwed against walls. Everything just melts now. I've even gone that far to move cpu intense processes into the night. Machines will just pop right up to 70°C and don't really want to burn through them either.
<hnhx[m]>what country are you from?
<ss2>I'm not jocking. :P
<hnhx[m]>i mean i live in hungary so the weather here should be similar but i didnt notice overheating in any of my machines
<unmatched-paren>I suspect the heatwaves will eventually catch up to us, given the horrific heat in London and the wildfires across England last summer.
<jonsger[m]>apteryx thanks for migrating icedove to gexp! Could you assist me in making the "thunderbird-sources" in native-inputs a gexp as well? I tried in the past but failed...
<ss2>It can be very dry and humid here where the heat will sit for weeks. I'm adjusting to it now.
<Kabouik>I'm facing a "fatal error: linux/limits.h: No such file or directory" when installing some package in R. I think limits.h is something that comes with kernel-headers and/or kernel-devel in other distros, what would be the way to install/compile those R packages in Guix?
<unmatched-paren>Kabouik: are you installing those R packages with Guix?
<unmatched-paren>the linux headers are in linux-libre-headers, btw
<Kabouik>No, with R's install.packages()
<unmatched-paren>You should probably package them in Guix instead.
<unmatched-paren>It's not good practice to use language package managers on Guix.
<Kabouik>Eh, that will probably be a significant deal breaker for me, there are far too many R packages to install or even just try before understanding they're not the solution
<Kabouik>R is a statistical software with its builtin packages, not really a package manager
<unmatched-paren>You can use the package managers, it's just better to make Guix packages.
<Kabouik>Not builtin, I meant third-party
<Kabouik>So for emacs plugins, Guix users also favor Guix packages over emacs builtin plugin manager?
<unmatched-paren>As you may expect, we have a *lot* of emacs packages packaged.
<Kabouik>I thought those plugins (either for R and emacs) work only within their "parent" program, so installing them with Guix seems a bit overkill to me (especially if one needs to write the packages definitions in the first place)
<unmatched-paren>Actually, we also have quite a few R packages, since many Guix users are scientists.
<Kabouik>That was my understanding too, but for instance we don't have r-ggplot, which is massively used in R, and I can't imagine no one in the Guix community uses it
<Kabouik>We even have Guix packages that work in combination with ggplot
<Kabouik>(Oh actually we have it)
<unmatched-paren>Ah, it's called r-ggplot2. Which would be confusing, I suppose.
<Kabouik>I knew the actual package name is ggplot2 (ggplot is the function), but was confused because `guix search ggplot` returned a lot of other packages before that one, I missed it
<unmatched-paren>Kabouik: Also, you don't have to write the package entirely yourself.
<Kabouik>I suppose I can import it yes
<unmatched-paren>Mhm, guix import cran PKG.
<Kabouik>But then I have to test it, submit a patch, and wait for its inclusion, while my statistical problem is still pending
<sleepydog>we would appreciate it! :P
<unmatched-paren>Kabouik: Well, you already have a collection of local packages that haven't been upstreamed yet, haven't you?
<unmatched-paren>Kabouik: You could also consider starting your own channel for these packages.
<Kabouik>I can't imagine telling my intern "Err, yeah apparently we need that package to solve your issue. I'll submit a patch to Guix and then I can help you. Hold on." I'm joking of course, I'd probably try to package it and build it locally first, but in some cases this may be a huge time sink even if not waiting for a merge
<unmatched-paren>You do have a point, of course. You could maybe use package.install() in the short term, note the package's name down, and then, when you have time, make a Guix package for it.
<Kabouik>(Don't get me wrong, I'm not complaining or ranting, I'm trying to clearly understand the workflow of other R uses in Guix before making choices and going in a way that could be suboptimal)
<Kabouik>Typically, should I install packages in my user profile (guix install), in my system (guix system reconfigure), or in some specific profile for R (I don't intentionally use multiple guix profiles yet so I wouldn't know how to do it)? I'm asking because I think every `guix command` gets a tad longer every time I install a package in my profile. Currently with 185 packages it's already much longer than it used to be.
<Kabouik>Could grow substantially if my R packages are installed through guix
***MysteriousSilver is now known as vetrivln
<florhizome[m]><florhizome[m]> "nckxhexen turns out I am getting..." <- I would Like to get back to this: i am getting this in a Mail from my Provider. I use gut sendmail with msmtp as sendmail Program. When i just send normal Mail it works, but Patches Like this, it doesn't.
<sleepydog>Kabouiki: I'm not an R user, but I use guix for ocaml development. I usually just use the `opam` importer for guix to add packages to a guix.scm file that I use `guix shell` with. Sometimes it needs some fiddling, and that's annoying.
<Kabouik>I see sleepydog, thanks. Yes, that "Sometimes" exception is the issue, it usually gets in the way exactly when you can't take the time to deal with it.
<sleepydog>that is true. I have no time constraints on this work, as it's not part of my day job
<Kabouik>And I doubt colleagues would understand the situation either when they're waiting for my part
<unmatched-paren>Kabouik: You can use ``guix shell'' instead, to avoid installing tons of R packages
<Kabouik>I have an ultimate fantasy goal, learning emacs and using it as my R editor/repl. But until then, there are many small roadblocks that I may hit as I learn Guix (which is expected, and I'm still happy with the process), R packages being one
<Kabouik>Learning emacs will be another. :>
<sleepydog>it's OK to take your time :)
<rekado>Kabouik: ESS is nice
<rekado>Kabouik: you can use ‘guix.install()’ from within R
<rekado>it’s provided by the ‘guix-install’ package on CRAN, available as r-guix-install in Guix.
<rekado>this lets you install R packages that are in Guix, but also packages that are not in Guix and that would need to be imported
<rekado>it imports packages in the background, saves them to ~/.Rguix and then installs them from there.
<rekado>so you don’t have to submit a patch first
<Kabouik>Yes I found about ESS when I started that fantasy; still haven't tried though because I still have too much to learn with emacs keybindings first. Interesting about guix.install(), I didn't know about that one. Cool. Not great for sharing scripts with others or for submitting them to material repositories though.
<rekado>you’ll just accumulate definitions for missing packages in ~/.Rguix
<rekado>for sharing scripts with others you can send along the ~/.Rguix file
<rekado>or rather ~/.Rguix/packages.scm
<Kabouik>Others *not using* Guix I meant (or, should I say if honest, not even using Linux)
<rekado>to reproduce the environment you’d just set GUIX_PACKAGE_PATH=$HOME/.Rguix
<rekado>use ‘guix pack’ then
<Kabouik>Right, that sounds like a relevant use case for it
<rekado>you can share your environment symbolicly for those who use Guix or as a binary blob for those who don’t
<Kabouik>I believe online script repositories and reviewers/editors wouldn't be too happy about it, they'd most likely expect .R or .Rmd files. But for what it's worth, as long as I properly document how to reproduce the work, I might be allowed to send my scripts as a bunch of pdf or even png files. That'd be mean, though.
<rekado>an R file can never be enough
<rekado>it is evaluated inside a context that determines what the behavior will be
<rekado>you can share this context with Guix
***Dynom_ is now known as Guest8312
<Kabouik>I think guix.install() will be very handy rekado, thank you
<apteryx>jonsger[m]: what thing which would work for thunderbird-sources without a label would be to define it as a variable, and refer to it from the copying phase as #$varname
<apteryx>(and not declaring it as an input)
<civodul>apteryx: hi! what's the rationale for providing a default timezone?
<civodul>i thought it might lead users to install a system with a wrong timezone
<apteryx>civodul: if you don't care to specify onee, UTC seems a good default
<apteryx>it's also one less extraneous thing to specify when building docker images, for example
<apteryx>does that make sense? :-)
<civodul>well, i'm not sure why one would be fine with having the system on the wrong timezone
<civodul>i guess that can be the case if that's a throwaway server or Docker image etc.
<apteryx>right. the idea came years ago when I was generating docker images with Guix, and in this context a few things appeared as clutter to the system definition, such as the bootloader and timezone fields.
<antipode>linj,unmatched-paren: I've heard that mrustc supports aarch64-linux nowadays
<gnucode>morning guix!
<unmatched-paren>antipode: Ah, yes, looks like it does.
<gnucode>antipode: yup.
<unmatched-paren>gnucode: Hello.
<gnucode>unmatched-paren howdy!
<ric342[m]>How is armv7 support doing?
<ric342[m]>Is it better supported on guix than nix?
<unmatched-paren>ric342[m]: unlikely
<unmatched-paren>it's supported, but probably not too well, since the vast majority use x86
<unmatched-paren>and there are only a few people developing arm support
<unmatched-paren>nix is far more popular, so i'd guess its support is probably better
<ric342[m]>It is not officially supported on nix
<unmatched-paren>oh, is it not?
<gnucode>which ARM is 64bit? do we call it ARM 64?
<ric342[m]>And there is no binary cache
<unmatched-paren>gnucode: aarch64 and armhf
<ric342[m]>Other than community one
<unmatched-paren>ric342[m]: Oh, really? Nix has no official substitute server?
<gnucode>I personnally want to play with the MNT reform on guix...when it gets 8+ GB of RAM.
<ric342[m]>No, and they don't list the armv7 at all on their package search
<unmatched-paren>ric342[m]: guix supports x86, x86_64, armv7, aarch64, powerpc64le, and riscv64
<ric342[m]>But they still have things like u boot for armv7 devices and apparently some people use it
<unmatched-paren>though with anything other than x86, x86_64, or maybe aarch64, you should expect a tough time
<ric342[m]>Probably riscv more worth the effort than armv7 now
<nckhexen>unmatched-paren: indent_size = 4
<nckhexen>That is evil.
<unmatched-paren>nckhexen: huh?
<unmatched-paren>our cpp files have four-space indents.
<nckhexen>No, it's true. It's just evil.
<nckhexen>I think I'd erased that from memory TBH.
<unmatched-paren>I meant to send that EditorConfig file ages ago, but totally forgot about it :)
<unmatched-paren>nckhexen: By the way, do you think 85 is a decent "max line length" value?
<nckhexen>I understand what you're going for, and I think it's OK. There's no ‘soft limit’ or ‘show a ruler here’ setting? Although… I guess editors will treat this as a soft limit (no?), so maybe it should be 80?
<unmatched-paren>nckhexen: I think it's up to the editor.
<nckhexen>What does yours do¿
<unmatched-paren>Vim-editorconfig shows a ruler and tries to realign the text if you go past it.
<unmatched-paren>But it won't complain if you move the text back.
<nckhexen>Oh, that could be annoying for an 82-character URL.
<unmatched-paren>I'm... not entirely sure what emacs-editorconfig is supposed to do.
<unmatched-paren>It doesn't seem to do anything with the max line length value.
<nckhexen>I don't know. I think 85 is OK, but only because I assume everyone™ will see that and realise ‘oh, they really mean aim for 80’. Maybe that assumption is dated.
<nckhexen>If people just keep typing until their bell dings, it's not ideal, but it shouldn't be common IMO…
<unmatched-paren>It seemed to me that going a *little* past 80 was quite common, so i just went a little above.
<apteryx>there's usually easy ways to keep things under 80
<unmatched-paren>nckhexen: Aha, if I do :display-fill-column-indicator-mode it displays a ruler.
<nckhexen>Right. I use that in emacs. But that's not controll{ed,able} by .editorconfig, or is it?
<unmatched-paren>It seems to be.
<unmatched-paren>According to that issue, at least.
<nckhexen>I was terribly unclear, I meant outside of emacs.
<nckhexen>(Although that issue was new to me, so thanks.)
<unmatched-paren>nckhexen: "Outside of emacs"? How do you control an Emacs procedure outside of Emacs?
<nckhexen>I see :, I think vi(m). :display-fill-column-indicator-mode does not look like an emacs command.
<unmatched-paren>Ah, I'm using evil-mode.
<unmatched-paren>Should've wrote M-x ...
<nckhexen>I thought it was some fancy vim thing (I know raw vi, but I know vim is different).
<unmatched-paren>I don't believe vi{,m} permits dashes in command names.
<unmatched-paren>It uses camel case, e.g. :PackerInstall
<nckhexen>The size of this <list> does not bode well for standardisation of anything not in it. Anyway. Bit of a tangent by now. As replied: LGTM.
<unmatched-paren>nckhexen: Hmm, should I add a copyright header, actually?
<nckhexen>I'd say no, it's just a list of facts which is not copyrightable, knowing full well that doing so opens up a can of worms for many existing copyright lines in Guix :)
<nckhexen>I'm not hungry though so I'm not going to open it. If you really want to, you can try & see if anyone protests. I don't think I will.
*nckhexen away, bye o/
<unmatched-paren>And thanks :)
<attila_lendvai>gnome-shell-extension-clipboard-indicator saves the clipboard history by default to the disk in cleartext, including passwords. and the setting is called something obstruse. does this constitute a security issue? it sure surprised me.
<sektor[m]>Did something happen to the UNISAV module?
<sektor[m]>Found a fatal error about it when I tried to boot the installer.
<sektor[m]>Maybe it was the container that made it angry.
<antipode>sektor[m]: I'm not finding any 'UNISAV' modules, is this a public thing?
<antipode>attila_lendvai: It's a known thing not considered a security issue by GNOME, IIRC.
<sektor[m]>Oh, uvesafb.
<gnucode>well I think I have a really hacky first test for my opensmtpd service records...
<attila_lendvai>antipode, i know. i reported it to the authro, who said it's all fine. what i'm asking is what the Guix crowd thinks.
<antipode>I think that writing encryption stuff in applications is possible, and that leaving it to generic disk encryption is less prone to problems
<antipode>The kernel implements some nice disk encryption, if everyone used it, then no need to reinvent it per-application.
<antipode>OTOH not everyone uses it.
<antipode>authro = author?
<antipode>I first thought it was some other distro thing ...
<xd1le>ric342[m], efraim has an unofficial riscv64 substitute server for guix. If you're interested you could contact him about it.
<xd1le>He does a lot of working building and testing Guix packages for riscv64-linux
<mitchell>Can someone tell me how to build the guix-binary tarball that's used to install guix on a foreign distro?
<mitchell>of course as soon as i ask the question i stumble upon the documentation
<antipode>mitchell: From (guix)Binary Installation:
<mitchell>that's the one. Right at the very bottom lol
<antipode> The binary installation tarball can be (re)produced and verified simply by running the following command in the Guix source tree: make guix-binary.SYSTEM.tar.xz
<lechner>apteryx: thanks for the pointer@
<attila_lendvai>disk encryption is not enough for a simple security hole in a browser that blindly scans and collects well-known files... saving the clipboard to disk greatly increases the attack vector.
<zpiro>anticomputer: the first thing you need to know for crypto in guix, is that is protected from tampering. It is not proected from runlevel tampering, and for your needs in user space your bosses will trust the level of crypto if government trust is required for employment and living.
<zpiro>anticomputer: "above your paygrade and TMSC pay grade level of tampering".
<zpiro>TMSC having Nvidia as the first major customer for their industry level open policy and intellectual property service level and survey of production line.
<zpiro>That is the biggest fab in the world, and biggest and most powerful industry in Taiwain.
<nashdidan[m]>Hi guys. Any idea why importing a crate seems to fail with 'Throw to key
<nashdidan[m]>`gnutls-error' with args `(#<gnutls-error-enum The TLS connection was
<nashdidan[m]>non-properly terminated.> handshake)' see:
<unmatched-paren>nashdidan[m]: sounds like a transient network error
<unmatched-paren>try it again
<zpiro>That runtime is secure, is one of the reasons why Twaitan is a in conflict and another Ukraine situation, "we will never give open up".
<nashdidan[m]>unmatched-paren: seems to work now, thanks!
<nashdidan[m]>woooow. It happens quite a lot espeiclaly with recurssion...
***mark_ is now known as mjw
<rekado>I’d like to document all things containers in the Guix Cookbook. What do you think: would it be acceptable to include an example that uses the Guix Science channel?
<rekado>the example is about running a Guix System container with an RStudio Server system service, which currently resides in the Guix Science channel.
<apteryx>rekado: is the Guix Science channel committed to the FSDG?
<apteryx>then I do not see a problem
<unmatched-paren>Hi two[m] :)
<rekado>only free software goes there, but there will be more bundling or gnarly javascript packages
<apteryx>it could be more convenient to not require the use of a channel and demonstrates with just what's included in Guix proper, but if there's a good reason to demo Guix Science, I think it'd be OK.
<two[m]>is it possible to get search paths not as bash commads, but as a simple `PATH=/gnu/store/...` like format?
<two[m]>(from guix shell --search-paths)
<rekado>I guess I could limit the example to Guix services and reserve the RStudio / Guix Science stuff for a blog post … :)
<apteryx>sounds good
<rekado>two[m]: should be possible through the Guile API
<unmatched-paren>two[m]: maybe we could add a --search-paths-format option
<unmatched-paren>which accepts either ``shell'' or ``table''
<unmatched-paren>so ``export PATH="/gnu/store/.../bin:/gnu/store/.../bin${PATH:+:}$PATH"'' would become ``PATH /gnu/store/.../bin /gnu/store/.../bin''
<nckhexen>two[m]: It's not Bash, by the way. :+ is POSIX shell.
<rekado>two[m]: here’s something you can use:
<rekado>this will print all environment variables to be set as FOO=bar pairs.
<nashdidan[m]>When creating package definition for a rust app which requite pkg-config, is it... (full message at <>)
<unmatched-paren>nashdidan[m]: use-moduling does absolutely nothing but bring a module's variables into scoke
<unmatched-paren>scoke? scope :)
<unmatched-paren>nashdidan[m]: you do need to specify (native-inputs (list pkg-config)), yes
<unmatched-paren>you might also need to (inputs (list openssl)), not sure
<nashdidan[m]>Does is matter that pkg-config is actually already installed...? I'm not sure
<nashdidan[m]>why it's failing.
<unmatched-paren>with ``guix install''?
<unmatched-paren>guix builds are isolated from the environment...
<nashdidan[m]>Yes, just while I'm trying things out for myself...
<nashdidan[m]>Oh, ok. got ot. 10x
<unmatched-paren>a package's build environment won't be able to pick up anything outside it, by design
<jcmdln>Anecdotally, how long does `guix system init` take for others? I think I've been waiting an hour which seems like I might have an issue in my `config.scm`
<unmatched-paren>jcmdln: Hmm. What's the output on stderr look like?
<unmatched-paren>s/stderr/the terminal/
<jcmdln>repeat `substitute` messages
<unmatched-paren>substitute: with nothing after it?
<jcmdln>```substitute: updating substitutes from ''... 100.0% 49.5 MB will be downloaded```
<jcmdln>The messages have differing sizes so it appears to be doing something
<unmatched-paren>Hmm, sounds like it's working.
<unmatched-paren>Just taking its time about it...
<unmatched-paren>You do have to download quite a lot of stuff when you're initialising your system.
<unmatched-paren>WOO \o/ home-dbus-service-type works!
<rekado>jcmdln: depends on your connection to and whether you’re able to download binaries for the version of Guix you’re using.
<jcmdln>Sure, though my `config.scm` declares a handful of packages. I'm doing a minimal install as part of building a VM in Packer.
<jcmdln>ignore the silly indentation :)
<podiki[m]>unmatched-paren: home-dbus-service is...for running a dbus daemon as a user?
<unmatched-paren>podiki[m]: yes
<unmatched-paren>It's for things like Mako (which I'm making a home-service for) which require a session-specific D-Bus to work
<podiki[m]>speaking of services, I need to prepare peroxide packages and service for submission; just been putting off cleaning up ~30 go packages
<unmatched-paren>so now i can write (requirement '(dbus-session)) in home-mako-shepherd-services
<two[m]><Cairn> "Nice!" <- unfortunately it has the same error again
<two[m]>i was lucky
<jcmdln>Ah, looking at `traceroute` it's my ISP that has a bad/slow route. Sorry for the noise and thanks for the quick feedback unmatched-paren rekado !
<Cairn><two[m]> "unfortunately it has the same..." <- Darn! Not sure what to say then.
<gunnargrop[m]>Hey everyone! I'm very new to Guix so excuse me if this is a stupid question, but how come GDM can't launch a Gnome Wayland session, but SDDM can?
<cbaines>gunnargrop[m], the GDM can, it just needs configuring to do so
<gunnargrop[m]>cbaines: Oh okay, thank you. I read the documentation on desktop-services where it states you'll need to use the "sddm-service instead of GDM as the graphical login manager" if one wishes to use Wayland, so I assumed it really was the case
<unmatched-paren> :D
<cbaines>gunnargrop[m], you're potentially reading the documentation for the current release. Things have got better since then :)
<gunnargrop[m]>cbaines: oh I think I was! Thats great news, thank you for the link :)
<two[m]>what is ld-wrapper-0?
<unmatched-paren>two[m]: that's our ``ld-wrapper'' package
<unmatched-paren>it's on version 0
<two[m]>what does it do?
<two[m]> * what is its purpose?
<two[m]>i only see it makes build fail because it replaces unicode characters with question marks
<unmatched-paren>description: The linker wrapper (or `ld-wrapper') wraps the linker to add any missing `-rpath' flags, and to
<unmatched-paren>+ detect any misuse of libraries outside of the store.
<two[m]>where can i report its bugs?
<unmatched-paren>it comes with guix
***emacsen_ is now known as emacsen
<lechner>Hi, is anyone using podman? I see Error: open /etc/containers/policy.json: no such file or directory
<gabber>i get a "guix home: error: no target of type 'activate' for service 'wireguard'" when trying to add a wireguard service to my home configuration. does wireguard-service-type only work for system configurations?
<unmatched-paren>gabber: You can't add system services to a home configuration...
<unmatched-paren>Home services are prefixed with home-.
<gabber>huh :) my bad
<unmatched-paren>And they come from (gnu home services ...), not (gnu services ...) :)
<unmatched-paren>There aren't that many right now, but I've sent a few patches adding more, and I'm working on one right now.
<mroh>lechner: podman is more or less unusable in guix (even if you provide/generate these /etc files), see
<mroh>It's a feature: There is always something to hack on :)
<gabber>unmatched-paren: sounds cool! what services are you currently working on -- if i may ask?
<unmatched-paren>gabber: I've sent home services for senpai and dbus, and I'm working on one for mako
<unmatched-paren>senpai is an irc client, btw
<unmatched-paren>i'm also going to write a service for configuring the aerc mail client
<jab>unmatched-paren: not to bad-mouth guix home...but is it me or are their services a little different than system services?
<unmatched-paren>jab: of course?
<jab>syntax wise
<unmatched-paren>(service home-bash-service-type (home-bash-configuration ...))?
<unmatched-paren>looks mostly the same to me..
<jab>as in, system services seem more abstract and further away from the actual configruation file
<unmatched-paren>hmm, which ones are you thinking of?
<unmatched-paren>I assure you, that's not the intention
<jab>unmatched-paren: maybe I'm being a little weird...
<jab>I thought I vaguely recall someone's sway home service and it felt a little low level to me.
<jab>again not to point fingers. I am glad that we have it...
<unmatched-paren>jab: Guix doesn't have a sway home service
<unmatched-paren>RDE does though
<unmatched-paren>You might be thinking of that
<jab>unmatched-paren: it must have been in RDE then...
<unmatched-paren>Guix Home *was* originally in RDE until it was upstreamed, though
<jab>does guix home support creating arbitrary directories?
<unmatched-paren>you could do that by extending home-activation-service-type...
<jab>my use case would be create Documents, Downloads, and then softlink those files to docuements and downloads
<jab>maybe I will have to give that a try.
<unmatched-paren>which allows you to execute arbitrary code on ``guix home reconfigure''
<unmatched-paren>sure, you could probably do something using (simple-service) and home-activation-service-type.
<jab> will add that to me list of things to try
<unmatched-paren>jab: this'll probably work: ``(simple-service 'create-desktop-directories home-activation-service-type #~(for-each (lambda (dir) (mkdir (string-append (getenv "HOME") "/" dir))) (list "Documents" "Downloads")))
<unmatched-paren>this basically will run that gexp on ``home reconfigure''
<unmatched-paren>and of course you can modify that to do the symlinking too :)
<jab>unmatched-paren: thanks!
<gabber>unmatched-paren: awesome! i have my self-rolled emacs-background-daemon home-service running. do you think such a thing would be of interest upstream?
<unmatched-paren>gabber: yes! definitely! :)
<unmatched-paren>home-emacs-service-type would be a welcome addition to many peoples' configurations
<unmatched-paren>Including mine! :D
<unmatched-paren>I'll review it for you if you ping me when you send it.
<unmatched-paren>I won't be able to merge it though.
<jab>unmatched-paren: have you asked for commit rights?
<unmatched-paren>jab: I'm not able to do that until I reach 50 commits
<unmatched-paren>I currently have 37
<jab>that's the bar!?
<unmatched-paren>though I have many more in waiting
<jab>I think I have 2 or 3 commits max. :)
<singpolyma>Shoulda made smaller commits ;)
<jab>singpolyma: hahaha
<unmatched-paren>jab: Eh, I think it's a reasonable bar.
<unmatched-paren>You need to have been involved for >6 months, too
<jab>unmatched-paren: it makes sense.
<unmatched-paren>sure, you can't just give commit access to a random stranger
<jab>unmatched-paren: should we try to merge your create home directories service?
<unmatched-paren>jab: no
<unmatched-paren>it's too tiny
<unmatched-paren>jab: You also need three committers to vouch for you, too.
<jab>unmatched-paren: I would submit to you that newbie users may have a hard time creating directories via home service w/o that bit of code. (guess that makes me a newibe)
<jab>also, if you disable that simple service 'create-desktop-directories...would those directoies still exist?
<unmatched-paren>jab: Yes, I think so.
<jab>just as a thought that what we would want in that service?
<jab>not trying to poke fun at your code. You did just get me that much closer to using guix home. So thanks for that!
<unmatched-paren>jab: also,
<unmatched-paren>simple-service is just a shorthand
<unmatched-paren>that allows you to do this kind of trivial service extension without writing a whole service-type
<jab>unmatched-paren: what if I wrote a proper service that created those directories. Would an actual service written out be helpful for guix home?
<unmatched-paren>jab: Eh, maybe you could have a go.
<jab>I'll add it to me list.
<unmatched-paren>There's probably another service-type for gexps to be run when the service is remove
<unmatched-paren>for deleting the directories
<unmatched-paren>But you should probably check that they're empty first.
<rekado>the dummy package in gnu/system/images/wsl2.scm will need to have its license field changed
<jab>unmatched-paren: that's a good idea!
<rekado>#f is no longer valid