gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
<samueldr> toggle "view result"
<gchristensen> kbd is the shorthand for <keycap> -- https://tdg.docbook.org/tdg/5.2/keycode.html
<samueldr> though a user can make any macro
<gchristensen> that is nice :)
<samueldr> so we could have option:[boot.kernelPackages] that automatically does the right thing
<samueldr> with work, that could also include embedding the HTML for a "hover" card like effect for more details in-place
<infinisil> Well everything should be possible with enough work
<samueldr> sure, but what I mean is that the asciidoctor-side is already well-defined
<infinisil> Yeah, that seems nice
<gchristensen> what is the run-time closure of asciidoctor like?
<qyliss> It'll have GCC in it
<qyliss> Because Ruby does
<gchristensen> oh and it depends on ruby
<gchristensen> oof
<qyliss> asciidoctor is written in Ruby
<infinisil> Eh I think this is negligible to the benefit we'll have from it
<infinisil> And I mean, which person who writes docs doesn't have a gcc already anyways
<infinisil> Okay I guess if more people write docs there's more, but still
<gchristensen> docs get generated on user systems semi-regularly, unfortunately
<infinisil> Huh why that?
<gchristensen> user modules :)
<infinisil> Oh, but that's disabled by default
<qyliss> We can kill the Ruby GCC runtime dep
<qyliss> I wouldn't worry about that
<gchristensen> or bugs in our module, causing docs to change when options are set
<qyliss> It's only there so programs can ask what compiler Ruby was built with
<qyliss> And I'd be very surprised if any do
<infinisil> Yeah I think let's not worry about that, it's fixable
<samueldr> can that information be somehow built-in without bringing gcc in?
<gchristensen> I think it would be really really good to try and find a solution which doesn't involve ruby
<qyliss> samueldr: it wants to expose the path to gcc
<gchristensen> I am certain there will be a lot of pushback if there is ruby involved in docs generation
<samueldr> that doesn't invalidate my question :)
<samueldr> the path could be ROT13'd
<qyliss> gchristensen: really? why?
<samueldr> or even split at the middle of the store hash
<qyliss> samueldr: that would still break every program that used it, so not sure what the point would be
<infinisil> I don't think we should discriminate tools because of the language they're written in :)
<qyliss> gchristensen: pandoc supports (some?) asciidoctor
<infinisil> s/because of/based on
<qyliss> But if runtime dependencies are an issue Ruby is way better than Haskell
<samueldr> pandoc is worse imo, with haskell deps because of issues with non-x86_64 builds
<samueldr> qyliss: ah, I didn't think about *using* the path, only getting the path :)
<qyliss> If you want to get the path use nix-store -q :P
<gchristensen> asciidoc's closure (minus gcc) is like 840mb
<infinisil> That's pretty big..
* samueldr wonders if mruby is enough for asciidoctor
<samueldr> probably not, due to relying on gems
<infinisil> gchristensen: You mean asciidoctor right?
<qyliss> Bet it doesn't have to be
<gchristensen> yeah: nix-store --query --requisites $(which asciidoctor) | xargs du --bytes -s | grep -v gcc | awk '{sum += $1; } END { print sum }'
<infinisil> gchristensen: What about `nix path-info -S -h $(which asciidoctor)`?
<gchristensen> 946M
<qyliss> gchristensen: that still includes gcc's runtime deps, doesn't it?
<qyliss> gmp and mpfr and whatever
<gchristensen> (yeah, my gcc filtering isn't so smart -- but I'm not trying to lie and make it seem harder than it is -- I'm on board with switching doc techs for sure, but we must be aware of the run-time closure size of the tool we pick, and we'll receive a lot of pusback if we don't)
<infinisil> From who?
<qyliss> Oh no we definitely will
<gchristensen> eelco at a minimum
<samueldr> from closure size alone I'm sure
<gchristensen> (who is also on board with switching doc techs)
<infinisil> Okay eelco, but I haven't seen many other people being concerned about closure size
<gchristensen> I didn't try to get fancy with filtering out gmp and mpfr because I didn't know if maybe other deps depended upon it too
<samueldr> I am, but don't put any work into improving :(
<qyliss> I don't want to name names, but I know other people who would be
<samueldr> the aarch64 images are now too big to build on hydra (without compression)
<infinisil> Hmm okay I think I'm also a little concerned about it actually
<gchristensen> I am glad to hear thaht :)
<samueldr> and the x86_64 iso grew from IIRC ~6xx to 8xx MB since I started using NixOS
<infinisil> Is anybody trying to get rid of that gcc or should I give it a go?
<gchristensen> I think a friend of mine might have already done it in a patch they've prepared, but I don't know if they're going to submit it
<qyliss> Profspatch has
<gchristensen> nice!
<qyliss> let me find the pr
<infinisil> :o
<qyliss> Although he was doing it only for a new rubyMinimal
<qyliss> I think we should do it for all Rubies
<qyliss> But I understand why he doesn't want to block his PR with tat
<{^_^}> #75566 (by Profpatsch, 1 week ago, open): Add rubyMinimal
<qyliss> Can I get a flame graph of the runtime closure somehow?
<infinisil> Nice
<gchristensen> qyliss: what do you want it to show?
<infinisil> qyliss: I'm not sure how a flame graph makes sense for that
<gchristensen> I guess bytes?
<qyliss> yeah
<qyliss> maybe a flame graph isn't the term
<qyliss> but i think it is
<gchristensen> yeah I think it could be useful
<infinisil> Oh I get it
<infinisil> Hm, how do you visualize a DAG of sizes
<MichaelRaskin> Maybe with circles of different sizes and thickness of arrow proportional to the subgraph reachable from the arrow's target?
<gchristensen> oh right infinisil
<infinisil> MichaelRaskin: That might work yeah
<infinisil> MichaelRaskin: How about the graph, where every node has the size of how small the graph gets if removed
<gchristensen> infinisil: zfs has the same problem: if one big node is dependend upon by 2 paths, neither of them "Claim" that space
<infinisil> s/how/how much
<MichaelRaskin> Hoe much smaller, you want to say
<infinisil> Haha yes
<infinisil> Is today typo day?
<gchristensen> every day is typo day when you have Graham around
<infinisil> :D
<MichaelRaskin> Well, it is kind of single-step
<infinisil> I'm just building asciidoctor with the rubyMinimal to see how much better it gets
<MichaelRaskin> In such a graph you never see a two-rev-dep package that is the real problem
<infinisil> Well darn
<infinisil> GCC is still included
<infinisil> Specifically the mathematical and nokogiri ruby packages pull it in
<samueldr> infinisil: by adding `bundlerApp = bundlerApp.override({ ruby = rubyMinimal; });` to asciidoctor's callPackage I presume
<infinisil> Yeah pretty much
<infinisil> Here's the why-depends for gcc: `curl https://paste.infinisil.com/fuy_avNzww`
drakonis has joined #nixos-chat
endformationage has quit [Ping timeout: 240 seconds]
<colemickens> So if I disable upower, my laptop will stay on for 60+ minutes beyond what it thinks it can manage.
<colemickens> How do I "fix" this? Is this an intrinsic property of LI-ion batteries? Can I "reset" or calibrate the controller?
<infinisil> I think older batteries are just really bad at estimating the capacity
<drakonis> older batteries get worn out as well
<drakonis> and that predictably leads to reporting incorrect values over time
<gchristensen> colemickens: isn't that what running the laptop until it dies does?
tilpner_ has joined #nixos-chat
tilpner has quit [Ping timeout: 240 seconds]
drakonis has quit [Ping timeout: 245 seconds]
drakonis has joined #nixos-chat
<pie_> quoting doesnt really work in github issue search but i think last time i asked someone had a solution?;
<pie_> how can i search for issues on "nixos-option"
<pie_> why cant they at least properly index issues -_-
<pie_> do you guys know any like...toggle switch app things for linux? i want to have some things i can turn on and off on my desktop or somesuch
<sphalerite> pie_: you mean like start and stop a program?
<pie_> sure
<pie_> turn on / off containers. whatever
<pie_> obviously should be scriptable
<pie_> s/obviously//
<sphalerite> pie_: good question tbh, the only thing I can think of off the top of my head is putting a segment in swaybar for it. That assumes you're using swaybar though
<pie_> sounds like a tool that /would/ need desktop integration - would be cool if there was a semistandardized interface though but thats asking for too much
<sphalerite> pie_: not necesssarily, it could just be a little window
<pie_> true
<pie_> not as nice though
<pie_> sphalerite: hm, might not be that hard to code something up for a terminal either since the support clicking things
<pie_> but im probbly not going to try that right now
<pie_> i just googled temrinal dashboard
<pie_> this might lead to something cool..
<pie_> doesnt look clicky though
<pie_> hmm some random quote suggests > Some panes are interactive,
<pie_> oh no its written in go
<pie_> i dont know that
<pie_> loks like theres a command runner module at least
* pie_ looks for other dashboard implementations
sphalerite has quit [Quit: WeeChat 2.6]
sphalerite has joined #nixos-chat
tilpner_ is now known as tilpner
<ar> a/39
__monty__ has joined #nixos-chat
pie_ has quit [Ping timeout: 258 seconds]
pie_ has joined #nixos-chat
drakonis has quit [Quit: WeeChat 2.6]
pie__ has joined #nixos-chat
pie_ has quit [Ping timeout: 260 seconds]
<joepie91> adisbladis: I've been otherwise occupied for a while, so didn't get around to asking earlier... did you get anywhere with the keyboard thing?
endformationage has joined #nixos-chat
pie__ has quit [Ping timeout: 260 seconds]
drakonis has joined #nixos-chat
<MichaelRaskin> Fun fact: Linux kernel 9p over TCP support assumes IPv4 with no way to override
red[evilred] has joined #nixos-chat
<red[evilred]> Not awesome
<red[evilred]> Oof
<MichaelRaskin> I guess I should just find a place where enough IPv6 enthusiasts gather that someone gets a patch included
<lassulus> would it be a cool idea to have language specific categories on discourse.nixos.org? people asked about it in #nixos-de
<gchristensen> sounds cool
<MichaelRaskin> I think there _is_ a French language category, and like four people promising to look at it, and zero substantial posts there
<lassulus> that could also be the fate of a german language category, but is there a downside to having dead categories?
<MichaelRaskin> But #nixos-de is active, maybe a German language category would go better
<MichaelRaskin> I have no idea why the decision was not to add everything where at least a couple of people would be able to answer
<MichaelRaskin> I think non-speakers would need to mute them one by one
<MichaelRaskin> (_if_ traffic actually happens)
<MichaelRaskin> I guess one could add one language category per month, until we either run out of languages to add or get some non-English traffic
<lassulus> I will talk with some people at 36c3 and see if there is interest and if there is someone for moderation
<gchristensen> just had a postgres INSERT query take 12min :')
<MichaelRaskin> How much data you were inserting?
<MichaelRaskin> All of the Hydra dump?
<MichaelRaskin> (well, the largest table)
<samueldr> 12 minutes feel without actual deep knowledge, not that long for hydra-big-data
<samueldr> or might
<joepie91> TIL: online key duplication services from pictures are a thing
<samueldr> or might I say acceptably long?
<joepie91> this is a thing you can just... do
<samueldr> only from your own keys that you own
<joepie91> I mean as in, I didn't realize that there were actually shops doing this over the internet
<joepie91> I know that it's /possible/ to duplicate keys from a photo
<__monty__> Yes, but I "lost" *my* key, samueldr, I just happened to take picz of it last week.
<samueldr> I was trying to be funny, not sure how to classify this...
<samueldr> because I'm sure no one would ever send photos of keys they are not the owner of
<gchristensen> bummer, this database import took >1h and then failed
<worldofpeace> 😿
<joepie91> samueldr: to be fair, pretty sure you could walk up to a random sketchy key duplicator and give them a keying code, too :P
<samueldr> yep
<worldofpeace> gchristensen: (offers gift) 🎁 -> 💥 -> ✨. Hope it catches a break.
<gchristensen> thanks! :)
<gchristensen> hrm, vim is taking a long time to open this file. I thought it handled long files better than ... it is ...
<samueldr> I think it does when it doesn't try to do things like detect the filetype
<samueldr> or having plugins do stuff
<samueldr> (but I think you're plugin free)
<gchristensen> nix-shell -p vim, on a fresh machine
<gchristensen> ahh it is trying to do highlighting out of the box
<samueldr> that would have been my assumption
<gchristensen> hm it is doing something else, too. `syntax off` did the thing but still it is quite slow to open
<samueldr> line numbering
<gchristensen> ah
<samueldr> not trivial to number lines, you can't really look at the end to know :)
<gchristensen> something else, too :P
<gchristensen> eh its fine, I think I can patch this file with just grep and sed
<samueldr> sounds ominou
<samueldr> s
<gchristensen> 109G /tank/hydradbstorage/pg_dumpall.sql
<gchristensen> any of y'all fine folks good with postgres admin? :)
<MichaelRaskin> Do you have access to the old server? Maybe re-dumping to binary format (with text dump as last-measure backup) could be a better idea?
<gchristensen> I sure do
<samueldr> all I know about postresql database dumps is that there is only one format that works for me, and I never remember which one
<MichaelRaskin> Which suggests it is not text SQL
<samueldr> I prefer not trying to remember more details from those times for now :)
<MichaelRaskin> Intuitively, there should be a reason for custom format, so…
<gchristensen> "custom" is a strange name
<MichaelRaskin> With alternatives being text, directory, tar — not _that_ strange
<gchristensen> "custom" to who?
<MichaelRaskin> Obviously, to PostgreSQL
<gchristensen> as a person just keeping lights on I don't want things to be custom
<gchristensen> I want things to be standard
<MichaelRaskin> Well, you can also wait until the standard SQL text dump loads, I guess
<gchristensen> when I was reading the pg_dump options, I didn't read `custom` and think oh, must be custom to postgres -- I thought oh, must give me fine control over output
<__monty__> I am saddened by how the various variations of who have gone the way of the dodo.
<gchristensen> anyway, good to know that custom is standard :)
<samueldr> the owl whoots
<MichaelRaskin> I think actually the standard SQL output forat has the _most_ tunable options.
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
tokudan has quit [Quit: Dunno.]
tokudan has joined #nixos-chat
<infinisil> I'm so sick of seeing VPN ads everywhere..
<gchristensen> yup.
<gchristensen> love that snake oil
<samueldr> don't you want to be secure, and completely anonymous online?
<samueldr> hackers will steal your passwords in cybercafés!
<srhb> I'm more scared of my government stealing all my browsing habits, that's what they should be advertising obviously :P
<gchristensen> better give it to a company than let the govt steal it?
<samueldr> I print my browser history and send it to 1 parliament hill every year, just in case they lost mine
<drakonis> vpns are good for one thing, getting around region locks for things
<qyliss> VPNs can be good for pseudonymity too
<qyliss> as long as you can pay in cash
<qyliss> (Mullvad)
<samueldr> imagine if you had to send your browser history to the internal browser history service, like you send your revenues to the internal revenue service (replace appropriate locale equivalent)
<qyliss> especially for stuff like Freenode that won't let you register over Tor
<qyliss> You can do Tor -> VPN to register
<qyliss> that's an advanced use case, I'll admit
<drakonis> an interesting proposition there
<infinisil> There certainly are usecases. If only advertisements would tell them and deceive people
<infinisil> s/deceive/not deceive )
<infinisil> !
<drakonis> i sent in patches to fix up gzdoom to be on par with every other distribution
<drakonis> yay for source ports
<drakonis> doom game nights when, people?
<lassulus> Yay doom
<drakonis> please check
<lassulus> Patches where? Did I got mentioned?
<drakonis> yes
<drakonis> nixpkgs#76154
<{^_^}> https://github.com/NixOS/nixpkgs/pull/76154 (by Drakonis, 8 minutes ago, open): gzdoom: include missing soundfont files
<drakonis> this is why it sounds so wrong
<gchristensen> uh oh I wish I didn't know about gzdoom
<drakonis> zandronum for online play, i'll send in patches for that one too
<drakonis> have it merged, mr.
<drakonis> while i play this one noita game
<drakonis> kicking a minecart around for that sweet sweet gold bonus
<drakonis> zandronum seems to not have missing files, v. nice.
<drakonis> lassulus: does it work?
<drakonis> up next: quake injector
<lassulus> Uhm im currently busy. Will look later. But i never experienced sound issues in gzdoom
__monty__ has quit [Quit: leaving]
<drakonis> to be fair, i have run gzdoom on windows and other distributions and it sounds quite different from the one shipped here
<drakonis> due to missing soundfonts
<drakonis> doom's midi soundtrack should sound like what's on youtube rather than how it it does in nixos
veske has joined #nixos-chat
<gchristensen> this psql import has been running for 7h
<drakonis> hoo boy
<leons> Probably busy planting a lot of trees on your drive
<gchristensen> B-shaped I assume
<samueldr> as long as it doesn't turn pear-shaped
<drakonis> or apple-shaped
<gchristensen> LOG: duration: 2108506.903 ms
veske has quit [Quit: This computer has gone to sleep]