gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
jasongrossman has quit [Ping timeout: 245 seconds]
ivan has quit [Quit: lp0 on fire]
ivan has joined #nixos-chat
lejonet has quit [Ping timeout: 245 seconds]
jasongrossman has joined #nixos-chat
gspia has joined #nixos-chat
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.2)]
jasongrossman has joined #nixos-chat
endformationage has quit [Ping timeout: 244 seconds]
Drakonis has quit [Quit: WeeChat 2.4]
<colemickens> Drakonis_: I'm jealous of all of the new Guix stuff. Has something happened that they have gained a larger following, more money for various new efforts, etc?
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-chat
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-chat
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-chat
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-chat
<ashkitten> so tired of java build systems
<ashkitten> janky garbage
<ashkitten> literally i couldn't get this app to work in android studio even though it was a pre-setup android studio project
<ashkitten> and dont even get me started on how many increasingly meaningless layers of abstraction i had to go through to hand-trace the handling of a single event when i couldn't get the thing to build
<ashkitten> im not even exaggerating when i say i had to go about 9 layers deep just to find any actual logic. it took me 2 hours.
cjpb has joined #nixos-chat
cjpbirkbeck has quit [Ping timeout: 245 seconds]
Myhlamaeus has joined #nixos-chat
cjpb has quit [Quit: Quitting now.]
ixxie has joined #nixos-chat
<pie_> colemickens, whats new in guix?
<pie_> cocreature, idk maybe there's some correlation with the archive thingy collaboration?
<pie_> or was that maybe a result of something else as well?
<pie_> i dont keep up with guix at all
<pie_> huh their website is looking kinda fancy
<pie_> i wonder if lisp vs Nix has anything to do with it
<pie_> i think theres some conceptual appeal in a fully lisp stack (and theres a blog post talking about mes bootstrapping from what i see) but i would prefer something that enfores purity and i like a bit of syntax...
<pie_> :p
* pie_ shrugs some more
<colemickens> Guix devops, guix workflow, the effort to get all of their package sources into an (ipfs?) archive, etc
<pie_> i wonder if it has anything to do with nix being in c++ :P
<pie_> im probably thinging about a bunch of superficial things, what do i know
<pie_> colemickens, guix workflow?
<pie_> colemickens, do you think there might be some reason guix is fundamentally easier to develop? or is that the wrong question?
<pie_> i guess im trying to ask something along the lines of "huh....ok now what"
<ar> https://finitestate.io/wp-content/uploads/2019/06/Finite-State-SCA1-Final.pdf >On dozens of occasions, Huawei engineers disguised known unsafe functions (such as memcpy) as the “safe” version (memcpy_s) by creating wrapper functions with the “safe” name but none of the safety checks.
<colemickens> I don't know. Maybe the fact that it's all Guile is appealing and makes tooling easier? But really I don't know and am curious myself, when from most reports, NixOS is more functional due to more pkgs being available, and yet, Guix seems to be gaining attention
<pie_> hm
<pie_> maybe someones workin on stuff in the background xD
<pie_> to be fair we've had stuff about software archival in the issue tracker for ages
<pie_> i started an issue myself but nothing really happened
Myhlamaeus has quit [Read error: Connection reset by peer]
<pie_> people work on their things, if you want something to go forward you have to take ownership of it
<pie_> i guess
<pie_> sidenote: i have felt some uncertainity on if i worked on something whether it would be merged or considered of use
<pie_> on that note i should get my POC into a usable state but id have to task switch again sigh :V
<pie_> colemickens, sidenote: if nix turns out not to be the best thing ever any my knowledge doesnt transfer i will be a little sad :p
ixxie has quit [Ping timeout: 245 seconds]
<colemickens> It's hard for me to say. I have my own opinions about Nix and the tooling, despite my love for NixOS. I really don't know anything about scheme/guile/guix, but my naive impression is that its a more holistic approach since "everything is in Guile".
<colemickens> I feel like I ought to try to send some PRs for Guix (or w/e they renamed GuixSD to)
<pie_> > despite my love for NixOS
<{^_^}> undefined variable 'despite' at (string):255:1
<pie_> i know that feeling
<pie_> colemickens, i feel like i ought to play with it at some point
<ar> /39
<sphalerite> wait, guixsd isn't called guixsd anymore?
<joepie91> pie_: I tend to design whatever I work on to work standalone / drop-in-able, so that I don't need to deal with convincing a maintainer that some new approach is worth doing
<joepie91> rather I make the thing available for people directly, and if enough people want to use it, it proves itself :P
<pie_> joepie91, hm. I mean I could provide it as a secondary library in nixpkgs heh
<pie_> probably good for a transition period anyway
<ashkitten> sphalerite: it looks like theyre just calling it guix system now
<joepie91> well, in the case of nixpkgs, and overlay (and in the future, a flake) is probably a good approach
<pie_> joepie91, did you look at my thing yet? xP
<joepie91> (this is why I was quite happy about overlays being introduced)
<pie_> true
<joepie91> pie_: which of them? :P
<ar> is guix still FLOSS-only?
<pie_> i dont want to maintain this forever myself though
<ashkitten> ar: yes, it's gnu
<pie_> "look i am your father" "gnooooooo"
* pie_ shows himself out
<ashkitten> wh
<jasongrossman> Not only is guix FLOSS-only, it looks like it won't ever have ZFS. That's my biggest problem with it.
<joepie91> pie_: I have looked at it a bit, but have not had time to look at it properly yet :P
<ashkitten> isn't the cddl not recognized as a free license by the fsf? id assume thats why
<pie_> joepie91, im trying to play with prolog right now but the thought of maybe finishing that thing this week is starting to distract me again -_-
<ashkitten> anyways im off to sleep, i suppose
<pie_> ashkitten, good sleepy tiems
<pie_> (no idea if its actually night)
<ashkitten> wish me luck, i'll need it tonight
<joepie91> pie_: work on what you enjoy :P
<pie_> or good night ashkitten local time
<pie_> joepie91, i hate everything
<joepie91> ashkitten: good luck
<pie_> jk
<pie_> joepie91, would be nice to get some feedback on how ridiculously overengineered it is, but i guess id need to clean it up more for that :(
<pie_> basically reviewing this thing is kind of involved...
<joepie91> pie_: it'd likely be a lot easier to get feedback if you add some 'wayfinding' documentation that describes what pieces are where, how the pieces interact, how it's different from the existding approaches, what new problems it solves, etc.
<joepie91> pie_: basically, frontload the high-level architectural view of things for people
<pie_> yeah i guess ive been putting that off because its not actually completely settled in yet
<joepie91> instead of having them infer it from the code
<pie_> well the implementation isnt
<joepie91> that's fine, documenting incomplete things is often a good way to find spots that still need work
<pie_> the interfaces seem to be largely clear to me - there is the one document but its a mess. i guess ill start with that.
<joepie91> just add a marker that the docs are a draft
<joepie91> and may become wrong over time
<pie_> in my limited experience im finding the holes by trying to write POCs
<pie_> i guess ill only really be happy once ive had a shot at two language infrastructures
<pie_> alternatively noone will use it and i can decide to drop it and not try to contribute anything major to nixpkgs again :P
<pie_> ok thanks for the feedbacc
<joepie91> well, writing POCs is important too, but it finds you different holes, mostly pragmatic onesa
<joepie91> ones*
<joepie91> documentation works well for architectural "hey, this doesn't actually make sense in general" holes
<joepie91> doing both gets you the best results :P
<pie_> yeah i try to do both. i dont know about helping me see "doesnt make sense in general" xD
<pie_> sidenote ive tried to reach out for architectural help but i feel like i havent really gotten it. maybe i did it wrong though.
<pie_> hm. actually. I guess I havent tried a discourse post yet.
<joepie91> very few people have both the interest and insight into a topic to help with complex architectural questions for any one given topic, in my experience
<joepie91> I've stopped asking people broad architectural questions for that reason
<joepie91> (and those few who do, are usually busy :P)
<pie_> also time
<pie_> yeah that
<pie_> i figured i guess i just havent given up on trying yet
<pie_> the uncertainity of whether anyone will care is a bit frustrating
<joepie91> so instead, I usually just ask people to describe their usecases, workflows, I give them an option A and option B and ask them to pick, stuff like that - basically scoping things down to the point that they don't need a broad understanding of the architecture to provide useful feedback
<pie_> i mean. i guess *i* care which is why im bothering at all, but its still a lot of work.
<joepie91> pie_: that's the unfortunate reality of this sort of stuff... most people typically really don't care until it Just Works for their problem
<pie_> i guess i kind of tried asking people if they like withPackages, but i havent directly asked "Do you like withPackages"
<joepie91> which can be quite demotivating
<pie_> i think thats fine, for most people
<pie_> but like....idk
<joepie91> I've more or less stopped trying to design for people's stated interests directly, in that sense, and just design according to an optimal design based on in-the-wild usecases and problems and issue threads I've seen, and the only hard objective is to make something that I would want to use and anybody else is a bonus
<pie_> i mean i dont expect most people to care about random system components
<pie_> yeah i guess that makes sense
<joepie91> and only having yourself as your greatest critic is a lot less stressful :P
<pie_> i guess nbp said i could send him an email or something but thats kind of high effort heh...
<joepie91> tl;dr don't let yourself burn out, don't worry too much about whether other people will find a thing useful
<joepie91> worst case you'll have gained design experience
<pie_> i guess im being a bit of a hypocrite because im always going on about documentation and here i am not wanting to write it
<joepie91> pie_: well, give it a shot anyway :)
<pie_> well thats the thing, i need to/he wanted a proper description of whats going on
<pie_> which isnt an unfair criterion
<pie_> ive been getting there but i guess i need to grow up some more and do everything myself
<joepie91> pie_: another reason to write those design docs :)
<pie_> * i have some they are just meaningless to others ;P
<joepie91> well, ones that aren't then :P
<joepie91> anyhow, I do need to go take a shower...
* joepie91 just woke up
<srhb> sphalerit: Um... I don't know why our train stations are confusing. I think it's mostly lack of signage compared to other EU places? :P
<srhb> sphalerit: I have noticed that I feel better informed in eg. Germany >_>
<pie_> sphalerite, i have a pet peeve about germany having hard to find or lacking street signs though...
<pie_> maybe its just me
<srhb> Ace public transportation though.
<jasongrossman> pie_: My local train station (in Australia) is on a small road that comes off the main road through my city. There are literally no signs mentioning the station on the main road. Can't get worse than that.
<pie_> jasongrossman, nice
<sphalerit> srhb: it was very unclear to me which platform we were on :p
<sphalerit> pie_: I think it's just you - or the places you've been in
<joepie91> I find German train stations confusing too
<joepie91> no trouble with Dutch and Belgian train stations though
<joepie91> and the UK ones I've been to were also pretty easy to navigate, other than the sometimes entirely lacking departing-trains information
<pie_> you can check out but you can neer leave
__monty__ has joined #nixos-chat
lejonet has joined #nixos-chat
waleee-cl has joined #nixos-chat
ixxie has joined #nixos-chat
<__monty__> qyliss: Having a *graphical* login prompt. For those people who are afraid of terminals, which is still most of them.
<qyliss> Right, but you said that you had one for other reasons
<qyliss> I assume that you are not afraid of terminals
<__monty__> Reasons other than X "needing" it.
<qyliss> Right. But what are those reasons, _for you_?
<__monty__> Other less experienced users using the same machine.
<qyliss> ah, okay, fair enough.
<clever> one thought i had for hsdm, is to have a curses based login screen
<clever> and after you login, spawn X and your session
<qyliss> I've seen a curses DM in nixpkgs
<clever> effectively the same process, no terminal to deal with, but simpler to implement
yorick has joined #nixos-chat
<pie_> in hindsight maybe that got too long
<jackdk> I am glad someone is trying to unify that abstraction
<joepie91> pie_: currently focused on something else so I can't read the details, but as a general recommendation: start the post with the "why does this exist, why does it matter, what problem does it solve" and only delve into the details after that, so that people understand in what frame of reference they should read the details
<joepie91> (a link to an existing thing that describes the problem alongside a two-sentence summary blurb is fine too)
<joepie91> pie_: oh sorry, that's there, the link jumped to post 2 so I didn't see it
<joepie91> nevermind :)
<pie_> joepie91, oops :D
<pie_> also yeah im just dropping the link
<pie_> i probably did the pitch all wrong but its a start i guess
<joepie91> will give it a proper read at Some Later Point :P
* joepie91 is currently refactoring a years-old library
<pie_> joepie91, i can use the criticism later when you have time, i might actually be able to internalize it...eventually
* pie_ is a slow learner :C
<joepie91> pie_: right, can't make any promises about when though
<joepie91> too busy for $client atm to allocate working days to this, so it falls into my time off, and I don't schedule things in my time off :P
<joepie91> (for obvious reasons)
Myhlamaeus has joined #nixos-chat
ixxie has quit [Remote host closed the connection]
jasongrossman has quit [Ping timeout: 245 seconds]
<infinisil> Idea for a blog post series: It's called "Let's learn about Nix" which walks you through different things you might want to learn about Nix
<infinisil> E.g. "What are multiple outputs?" could include some experiments to find out how they work, what the manual tells you, finding out how they're used in nixpkgs
<infinisil> Or another post could be "What are Nix builtins?" in which one could show the auto-completion in nix repl, go to the nix source to find all of them, try to use some of them, etc.
ixxie has joined #nixos-chat
<ldlework> anything you do is gonna be of great value infini-chan
* ldlework swoons
<infinisil> Hehe, well I first have to figure out how to set up my hakyll blog, I've looked into it once, but this really isn't the thing I'm into
<infinisil> Being me I couldn't just use the template though, I had to start from a blank state
<gchristensen> so many blogs die at that step
<gchristensen> so many "I just redid my blog, hopefully it'll encourage me to blog more"s
ixxie has quit [Quit: leaving]
ixxie has joined #nixos-chat
<infinisil> Maybe I should really just start by copying some template then yeah
waleee-cl has quit [Quit: Connection closed for inactivity]
endformationage has joined #nixos-chat
<pie_> infinisil, i have read things along the lines of hakyll is overcomplicated i think
<pie_> infinisil, you could use that one site generator thats in nix i think? ;D
<pie_> (oh no dynamic typing)
<ajs124> I had the same problem with rolling something for my cv website. hugo academic looked really nice, but it's sooo overcomplicated
Drakonis has joined #nixos-chat
<infinisil> Well, I *tried* to set up my blog once again, but instead it turned into learning HTML
<infinisil> But this plus later then CSS will be a good thing to know imo
<infinisil> So no regrets :)
ixxie has quit [Remote host closed the connection]
ixxie has joined #nixos-chat
ixxie has quit [Remote host closed the connection]
<ajs124> Nono, everything is fine, thanks for asking.
<ajs124> Startup finished in 1min 20.784s (kernel) + 1h 5min 4.220s (userspace) = 1h 6min 25.004s
<ldlework> infinisil: yeah everyone should probably know how to build for the web
<ashkitten> ajs124: are you booting nixos on a commodore 64?
<ajs124> ashkitten: that machine actually has two quad-core xeons. the problem is it was trying to clean up /tmp on a hdd raid, which was rebuilding.
<ashkitten> oof
<ajs124> /tmp probably contained >500k files and directories
<ashkitten> hmm
<ashkitten> i wonder if i should take tmp off of tmpfs now that i have special allocation vdevs on ssds
<ajs124> I don't have enough RAM to keep it in /tmp -.-
<ashkitten> but then again i've got 32 gigs of ram
<ashkitten> so idk
<ajs124> yeah, that server also only has 32GB
<ashkitten> i should benchmark mass deletions to test, i think they should be sorta fast
<ajs124> they're definitely faster on tmpfs than anything else
<ajs124> and they're very slow on a degraded raid, which is resyncing with >100MB/s
<ashkitten> well yeah tmpfs is in ram
<ashkitten> and drops when you reboot
<ajs124> The only thing faster might be reformatting the fs. Like, if you have a separate lv or vdev or something and just nuke it on every boot.
<ashkitten> or a zfs dataset
<ajs124> ye. I guess you can drop and recreate those as well
<ashkitten> either way ive got enough ram for this desktop
__monty__ has quit [Quit: leaving]
Myhlamaeus has quit [Read error: Connection reset by peer]