gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
<pie_> samueldr, its already written in bash
<samueldr> oh
<pie_> wish i could remember who i got some hints about wine from...logs on other computer
<gchristensen> samueldr: more bash?
jasongrossman has joined #nixos-chat
<pie_> hmmm netdevsim
<pie_> joepie91, new repo who dis https://github.com/deliciouslytyped/nix-winnix
<gchristensen> oh you're deliciouslytyped?
Guanin has quit [Quit: Leaving]
aszlig has quit [Quit: Kerneling down for reboot NOW.]
aszlig has joined #nixos-chat
<pie_> wtf my wine suddenly got a start menu :D it didnt have that 5 minutes ago
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.2)]
<drakonis> pie_: wazzat
<drakonis> is this the new wine package?
<drakonis> a lutris replacement, shall we call it?
MichaelRaskin has quit [Ping timeout: 245 seconds]
jasongrossman has joined #nixos-chat
Myhlamaeus has quit [Ping timeout: 258 seconds]
drakonis has quit [Quit: WeeChat 2.4]
__monty__ has joined #nixos-chat
teto has quit [Ping timeout: 268 seconds]
jackdk has quit [Ping timeout: 258 seconds]
<sphalerite> Just checked the logs for #nixos because I have iqubic muted, and now I have renewed confidence in the decision to mute them…
jasongrossman has quit [Ping timeout: 250 seconds]
endformationage has quit [Ping timeout: 246 seconds]
jasongrossman has joined #nixos-chat
jasongrossman has quit [Ping timeout: 246 seconds]
jasongrossman has joined #nixos-chat
nh2[m] has joined #nixos-chat
<nh2> ajs124: I hope you've sent your email, as I hit all bugs eventually :) For example these two kexec EFI bugs https://github.com/nh2/kexec-tools/compare/find_mnt_by_fsname-proc-mounts
<infinisil> sphalerite: I'm with you there..
kisik21 has joined #nixos-chat
kisik21 has quit [Ping timeout: 255 seconds]
kisik21 has joined #nixos-chat
Synthetica has joined #nixos-chat
<gchristensen> should I talk to them?
<pie_> ok it feels like better infinite recursion debugging is really going to be necessary with pervasive use of overlay systems like i am
<pie_> hnix supposedly has better output right?
<gchristensen> what error do you get, pie_?
<pie_> just the usual infinite recursion at wherever stuff
<pie_> i fixed it by swapping some self to super but thats not very insightful
<gchristensen> can you be super specific, and paste the exact error?
<pie_> error: infinite recursion encountered, at /home/nixos/stuff/nixapps/wine/pls/winepackages/again/layers/3_packages.nix:17:9
<pie_> best directory structures ever :P
<gchristensen> cool, thanks :)
<pie_> i can do a --show-trace
<pie_> but the other problem is it basically goes through a lot of "internals" like customization.nix and im not sure how helpful one could be with stuff like that
kisik21 has quit [Ping timeout: 246 seconds]
<joepie91> pie_: https://transfer.sh/mD9xN/nix-error-report.7z -- run Nix with --show-trace, pipe trace output through that tool (`node index.js`)
<joepie91> code is awful right now
<joepie91> but I wrote that basically to address that exact same issue
<joepie91> it'll produce a much easier-to-read stacktrace with code snippets
<joepie91> makes it easier to understand why the recursion is happening
<joepie91> oh
<joepie91> that is missing a dependency I think, sec
<pie_> oh cool ill give it a shot
<joepie91> pie_: hold on, lemme package up the stacktrace parser
<pie_> though i would prefer a easy to run .nix :P :D
<joepie91> none of this is polished or published yet
<joepie91> so it's a bit of a mess
<pie_> yeah thats ok
<pie_> i wasnt very good with my first few wine links i gave you either
<joepie91> such is working in the weeds
<joepie91> :P
<pie_> about to push some changes to the repo that makes my package library automatically call stuff in the packages dir so you dont have to manage shit like all-packages
<pie_> its a bit messy but seems to work
<joepie91> pie_: https://transfer.sh/h1WoP/nix-stacktrace-parser.7z -- unpack into its own folder, run `npm link`, then in the unpacked nix-error-report folder run `npm link nix-stacktrace-parser`
<joepie91> (yarn link instead of npm link is fine too, if you have a preference)
<pie_> i have no prefernce
<pie_> whatever is easier lol
<joepie91> may need sudo on the first `npm link`` if your Node/npm are installed systemwide and you have not changed your global module dir, yarn does not have that requirement
<joepie91> (yarn stores dev links in ~ by default)
<joepie91> I don't actually know if sudo npm link works *at all* on NixOS lol
<joepie91> pie_: alternative solution if you have issues with any of the above, though this will be wiped out the first time you do anything with npm/yarn again, is to manually drop a `nix-stacktrace-parser` folder into `nix-error-report/node_modules` with the contents of the latter 7z
<joepie91> then no symlinks are involved at all
<andi-> sudo + npm on one line makes me dizzy.. that sounds wrong
<joepie91> andi-: it's because npm stores its development links in the global module dir
<joepie91> which is system-wide by default
<pie_> uhh tfw you run attrNames on a set and it has a "" key
<pie_> good thing i looked lol
<pie_> oh derp
<pie_> fm
<pie_> * fml, how do i not accidentally introduce bugs like this
<pie_> i have a file named .kkrieger and I ran basename on it
<joepie91> ha, .kkrieger!
<pie_> well not even basename
<pie_> renamer = name: builtins.head (super.pkgs.lib.splitString "." name);
<joepie91> been a while since I've seen that mentioned
<pie_> i should just strip the .nix from the end or use a name attribute from the import or something
* joepie91 should really package up this nix-error-report tool at some point
<pie_> <oiaohm> Some of the reason why Valve is interested in supporting wine/proton is that it will give them a means to sell some games they cannot sell to windows users any more because they don't work but they do work on wine so they can sell to Linux users.
<pie_> makes sense that its not from the bottom of their hearts :P
<joepie91> laziest inferred type annotations on earth: https://i.imgur.com/S2Wlsea.png
<pie_> is there a lib function for removing entries from the *end* of a list?
<joepie91> is there not a generic slice?
<pie_> joepie91, lol, you do what you can
<joepie91> pie_: yeah okay but "typeof <foo>" is like, one inch short of saying "lol fuck you, figure it out yourself"
<joepie91> :P
<pie_> :D
<pie_> idk if theres a generic slice
<pie_> ah i guess i could do take length - 1
<pie_> i still need to access the last result though to check if its .nix
<pie_> oh theres a lib.last for "everything except the last"
<pie_> nvm lol, it keeps tripping me up that the docstrings are above the function definition for some reason
<pie_> woo! lib.removeSuffix!
Jackneill has joined #nixos-chat
<pie_> joepie91, ugh i didnt feel like trying it right now but i will probably need to, somehow my autoimports stuff causes all kinds of infinite recursion problems if i try to use things the way i like
<gchristensen> any opinions on things like google forms, but OSS?
waleee has joined #nixos-chat
<gchristensen> ok I'll use google forms
<andi-> I haven't seen a good OSS version of google forms.. Been searching whenever I needed to so a survey-ish thing :/ There were a few things (limesurvey?) that looked a lot like overkill for just a simple one-time form...
Jackneill has quit [Read error: Connection reset by peer]
Jackneill has joined #nixos-chat
pie_ has quit [Ping timeout: 264 seconds]
Jackneill has quit [Remote host closed the connection]
Jackneill has joined #nixos-chat
Jackneill has quit [Read error: Connection reset by peer]
Jackneill has joined #nixos-chat
<manveru> gchristensen: for what purpose? just embedding in a website?
<gchristensen> I'd like to create a survey and email it out to a few dozen people
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.2)]
<manveru> yeah, hard to beat google on that... not sure how OSS would work without making it vastly more complex :|
pie_ has joined #nixos-chat
endformationage has joined #nixos-chat
<sphalerite> gchristensen: I don't think iqubic needs a talking to, I'd probably go for a friendly hint that several people have them muted and that their style doesn't make helping them fun. But I don't see that going anywhere productive. So I'd just say let the mutes roll :p
<gchristensen> that doesn't sound very good
<gchristensen> can you PM me some stuff about why you've muted them?
<sphalerite> helping people is rewarding in that it makes me feel good. iqubic is good at preventing that good feeling from happening, and I recognised the pattern of helping iqubic not being worth it. So I muted iqubic to avoid wasting time helping them.
<sphalerite> That's about it
<gchristensen> ok, cool
<gchristensen> could you PM me some examples? :)
<ar> .44
Synthetica has quit [Quit: Connection closed for inactivity]
* pie_ mumbles something about how the windows registry doesnt compose or something
<clever> andi-: ive been to a bussiness that was just handing out kindle ereaders, opened to a google forms page, lol
<andi-> well I guess that works if you have physical contact to all the people you want to question
drakonis has joined #nixos-chat
Myhlamaeus has joined #nixos-chat
drakonis has quit [Read error: Connection reset by peer]
drakonis_ has quit [Ping timeout: 255 seconds]
drakonis has joined #nixos-chat
drakonis_ has joined #nixos-chat
drakonis has quit [Read error: Connection reset by peer]
drakonis_ has quit [Ping timeout: 258 seconds]
<infinisil> Ugh, Medium now integrates with Google -.-
<__monty__> Do you mean, google now scrapes medium?
<__monty__> I'll take anything that means I can spend less time on medium though.
<infinisil> __monty__: When you go on a medium site, there's a google pop up on the top that pretty much automatically signs you into your google account
<__monty__> Re earlier discussion. I've usually found pointing out to new people that there is such a thing as help vampires works splendidly. People are usually more like "Oh, didn't even know I was being an ass." rather than mad at me for pointing it out.
<__monty__> infinisil: I still get the "We've seen you here before sign in with google/facebook" overlay.
<infinisil> Yeah that's what I mean
<__monty__> That's been there forever though.
<infinisil> Pretty sure it hasn't, also just found https://github.com/thebaer/MMRA/issues/49
<{^_^}> thebaer/MMRA#49 (by jhass, 3 weeks ago, open): Remove "Sign in with Google" popup
<__monty__> I could've sworn I've had it for ages.
<__monty__> Maybe A/B?
<infinisil> Screw Medium, and screw google
<__monty__> Not as if there was much quality content on medium.
<gchristensen> +1
waleee has quit [Quit: WeeChat 2.4]
<samueldr> I don't understand, isn't medium a pretty "open" (as in: takes in any kind of blogs) platform? so it would be like saying "not as if there was much quality content on the internet."?
<samueldr> (though some would argue that there is no quality content on the internet)
<samueldr> but yes: what an horrible platform looking at it from any viewpoint :/
<__monty__> samueldr: tumblr was open too. Never encountered quality content there either.
<__monty__> Not like I looked hard, just anecdotal.
<gchristensen> Medium did a good job of curating content at the start
<pie_> ok ...does building windows applications from source on linux make any sense?
<pie_> s/ok//
<sphalerite> yes, why wouldn't it?
<gchristensen> someone does cross compilation of on linux to windows
<andi-> IIRC firefox for Windows it cross compiled with LLVM last time I checked
<andi-> that article is interesting.. It is a good example for the limits of Rust if you want to do it "right"
drakonis has joined #nixos-chat
<gchristensen> I feel the take-away here is that wlroots is simply not authored in a way which is memory-safety-friendly, not an insufficiency in Rust's design
<simpson> Yeah. This was a sort of taming to fit within Rust's ownership model.
<pie_> sphalerite, idk, all the build tooling is on windows or something
<simpson> Also, the cost of a second-system API.
<pie_> im not talking about cross compilation
<sphalerite> pie_: huh? I don't understand in that case
<gchristensen> for me, wayland has been surprisingly better than X was
<pie_> sphalerite, im uhhhh doing stuff with WINE lol
<gchristensen> it is unfortunate that a library, core to many parts of the wayland ecosystem (wlroots) is not compatible with Rust
<pie_> and uhhh we like to build stuff from source with nix, right? :P
<__monty__> If you're looking at rust as a "safer C" it being hard to interface with an existing "bad library" isn't motivating though.
<gchristensen> while I agree with your premise, I don't see it that way
<gchristensen> wait, no, I think I do completely agree
<andi-> I didn't mean to judge the quality of either of the sides. It is just tough putting unffiting things together. Yet another language is not changing that.
<samueldr> https://xkcd.com/2143/ <- totally unrealistic, where's /nix/store ?
<gchristensen> I have also been a bit distressed over sway's maintainers :(
<simpson> gchristensen: I don't know what Rust needs in order to compete, but I think it's illustrative that there is a crate that does X11 bindings by wrapping XCB. Normally one would *generate* an "XRsB" by turning XML into Rust; I wonder why that hasn't materialized.
<gchristensen> right
<gchristensen> sway rejected an issue requesting documentation be more clear that `exec sway` is safer than just `sway`
<andi-> thats meh
<simpson> The thing that xproto and XCB were supposed to prevent: https://github.com/DaMrNelson/xrb/blob/master/src/protocol.rs
<__monty__> Why is exec sway safer?
<andi-> __monty__: the classical case is that you login on a TTY, run your i3/Xorg/sway and when that crashes it is back in your user session available to everyone that might come across the machine. With `exec ....` you will instead drop out of the shell and password input will be required.
<gchristensen> in particular this is a concern if your screen is locked and the user has a convenient way to crash your gui stack
<__monty__> Ah, hadn't considered.
<gchristensen> which elevates any crash-inducing bug from an annoyance to privilege escalation
<pie_> this needs to be the alt text <samueldr> https://xkcd.com/2143/ <- totally unrealistic, where's /nix/store ?
<__monty__> I thought the process somehow got more privileges. I'm not fooling myself, physical access to the machine means game over. But yeah, let's not make it easier than it has to be.
<pie_> wow that sounds stupid <gchristensen> sway rejected an issue requesting documentation be more clear that `exec sway` is safer than just `sway`
<pie_> i have run into a surprisingly large amount of screen locker crashes...
<gchristensen> yeah most of them are very bad
<gchristensen> in X11 they're not even part of the design of the system, so they depend on assorted techniques to be good
<gchristensen> I would imagine they're part of the protocol for Wayland, but I don't know
<pie_> write my own screen locker with blackjack and nix
<gchristensen> jwz has a bitter series of posts about how every screensaver but his own (xscreensaver) is trash
<ajs124> gchristensen, do you mean https://www.jwz.org/xscreensaver/toolkits.html ?
<gchristensen> that is one of them
<sphalerite> jwz is a bit of an a- unpleasant person imho
<sphalerite> gchristensen: what's trash about xscreensaver?
<pie_> sphalerite, every *other* screensaver
<sphalerite> oh whoops, mis-parsed
<gchristensen> I don't mean to imply I agree with his assessment
<simpson> In the 14yrs, another thing that happened is s/Xlib/XCB/g, which is a big improvement in both the amount of underlying code in the TCB as well as the overall correct-by-construction'ness.
<simpson> (Even if one still tries to use Xlib, they will be using "Xlib/XCB" instead of the older "Xlib/Xtrans".)
<gchristensen> yea
<simpson> I point this out mostly because I would politely assert that GTK+ and Qt have gone in the opposite direction~
<sphalerite> gchristensen: my cake hasn't been going as well as yours :p
<gchristensen> ooh what kind of cake?
<sphalerite> carrot cake
<gchristensen> nice!
<sphalerite> I started making it yesterday, before realising I didn't have any baking powder
<gchristensen> oops :D
<sphalerite> so I put everything in the fridge, dropped an egg in the process, and generally stuff wasn't going well
<sphalerite> today I bought some baking powder and finished it off
<__monty__> Baking powder has never worked for me.
<sphalerite> it didn't come out of the baking tin properly so now it's in several pieces
<gchristensen> oh dear :P
<sphalerite> delicious pieces though! :D
<monsieurp> how does that work?
<__monty__> Might as well not use it.
<gchristensen> sounds like your baking powder was old, or you overmixed it, __monty__
<sphalerite> monsieurp: it stuck to the bottom
<monsieurp> I mean, what's the recipe exactly?
<sphalerite> so, knowledge for next time: use baking paper!
<monsieurp> of course
<monsieurp> always my friend
<monsieurp> especially when baking
<sphalerite> lol
<sphalerite> also, is github slow for anyone else?
<monsieurp> as a Frenchman, I always have a baking paper roll handy
<__monty__> gchristensen: But it still reacted perfectly hilariously with vinegar. : )
<__monty__> The internet is slow for me today. I guess that doesn't count though?
<sphalerite> eeeh stumbled over ertes- for the third time in the past few days :(
<__monty__> He was part of the nix community too?
<monsieurp> can I post articles related to the economy and all that?
<sphalerite> __monty__: yeah
<sphalerite> monsieurp: sure. This is the off-topic channel. Just don't start any flamewars ;)
<monsieurp> I ran into this article on bloomberg
<monsieurp> that I wanted to share
<monsieurp> this is just... so lol
<gchristensen> off the top of my head, the only thing I would say is definitely off topic for -chat is things that are rude, mean, or disrespectful
<sphalerite> https://sphalerite.org/dump/github-nice.png A++ frontend work
<gchristensen> monsieurp:...ouch
<pie_> :D
<pie_> danger, meese crossing over this button
<monsieurp> gchristensen: :D
<monsieurp> gchristensen: bear in mind Uber is soon filing for IPO!
<monsieurp> there's going to be blood all over the place
qyliss^work has quit [Ping timeout: 259 seconds]
qyliss^work has joined #nixos-chat
<drakonis> post ipo uber isn't going to survive for long
qyliss has quit [Ping timeout: 258 seconds]
<drakonis> they're going to find new and exciting ways to jack up prices for users
<drakonis> make it worse for drivers
qyliss has joined #nixos-chat
<monsieurp> I don't use Uber
<monsieurp> never did, never will
qyliss^work has quit [Quit: bye]
qyliss has quit [Read error: Connection reset by peer]
qyliss^work has joined #nixos-chat
qyliss has joined #nixos-chat
drakonis_ has joined #nixos-chat
__monty__ has quit [Quit: leaving]
<pie_> but will uber use you
<drakonis> uber uses all
<drakonis> but none shall use uber
drakonis has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-chat
hedning_ has quit [Quit: hedning_]
<pie_> today, on enabling two factor auth: oh i cant log into this account a friend shared with me
* pie_ shakes fist at capitalism
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-chat
lopsided98_ has joined #nixos-chat
lopsided98 has quit [Ping timeout: 252 seconds]
jasongrossman has joined #nixos-chat
jackdk has joined #nixos-chat
<elvishjerricco> I keep thinking about having a transactional file system, where each file IO syscall takes a transaction descriptor argument in addition to the file descriptor. POSIX could be a layer over top this with some sane default transactions. The transactional semantics would be up to the file system driver (though I imagine one could write a wrapper for any POSIX file system driver)
<gchristensen> what would the transaction descriptor be for?
<elvishjerricco> gchristensen: All IO done on a given transaction descriptor is done in the same transaction, with the same transaction settings (i.e. read-committed vs repeatable-read etc.)
<gchristensen> ah
<elvishjerricco> The equivalent to close for them would be commit, except it'd sync everything to disk unless it was opened with some flag
<elvishjerricco> Wondering how much of POSIX would have to be reinvented for that.... probably most of it
<pie_> elvishjerricco, you mean so stuff doesnt end up in weird inconsistent states?
<elvishjerricco> pie_: Yea. But also just do applications can get transactional semantics without a bunch of manual work
<gchristensen> you could just use sqlite :P
<pie_> ah, well, would be nice xD