<ashkitten>
and i guess fsync might land once the associated kernel patches are landed
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
lukegb has quit [Quit: ~~lukegb out~~]
lukegb has joined #nixos-chat
<infinisil>
Idea: Have an editor plugin that inserts fake sha256 hashes
<infinisil>
But it encodes the location of that hash in the hash itself
<infinisil>
Such that when Nix displays "Expected <hash>", you can provide some integration that automatically finds that hash and replaces it!
<clever>
infinisil: previously, i would just `grep -r` for the given hash, but now with base16, base32, and base64 representations, it aint simple
<infinisil>
Although, I guess there's not enough bits in a 256-bit hash to encode the full location..
<infinisil>
clever: Yee
<colemickens>
infinisil: nix-prefetch doesn't encode the location or anything, but it does automated hash replacements by sending 00s in: https://github.com/msteen/nix-prefetch
<colemickens>
I use it to update src and vendor/cargoShas in my overlays
<infinisil>
"by sending 00s in"?
<colemickens>
oh right, I store the revs in an external file -_- oops that's how I get away with it
<colemickens>
infinisil: it runs the fetcher with 000s for the hash and then parses out the error message to tell you the new hash
<infinisil>
Ah
<colemickens>
but I forgot that I am not updating htem in nixpkgs, I cheat and store them in a json file I control :(
<colemickens>
I guess the same broad find/replace of oldhash/newhash should still work in theory though
<infinisil>
Honestly, maybe Nix should just allow you to interactively update the hash
<infinisil>
Like, maybe if you have `interactive-hash-update = true` in /etc/nix/nix.conf
<clever>
that reminds me, i saw a really confusing error months ago
<infinisil>
For hash mismatches where the hash is in a writable file, it just asks you "Do you want me to update the hash?"
<clever>
2 different nix files, with the same url, one with hash A, the other with hash B
<clever>
the error says expected A, got B
<clever>
so you change the 1st file, A->B
<clever>
then the error says expected B, got A
<clever>
(from the 2nd file)
<clever>
but you dont notice the 2nd file, so you change the 1st back
<clever>
repeat, in circles :P
<infinisil>
lol nice
<colemickens>
yes.
<colemickens>
and then you're doing something else so you don't even realize you're doing it until you've done it 4 times
<samueldr>
it shouldn't be normal for software written after 1990 to just fail silently without printing why
<samueldr>
but here I am, debugging a build issue in a completely foreign language and ecosystem
Emantor has joined #nixos-chat
<samueldr>
where exit(1) is the best they do for error reporting it seems
<pie_>
c: im so sorry
<pie_>
id say use more rr but im not even sure thats relevant
<evanjs>
*sees gping on Reddit* Oh, great it's already in tree. *demos* wait, this looks diff--gah. still using the Python version
<evanjs>
lol I guess this would technically be a downgrade? from 1.1 (python) -> 0.1.6 (rust)
<pie_>
the idea of allvm has been nagging me
<pie_>
subconsciously
<pie_>
now im just like, can we just do the JVM thing for everything
<pie_>
(which probably isnt actually what theyre saying)
<pie_>
give me rust but with introspection
<pie_>
and i might not even need to want an interpreted language?
<pie_>
actually i need to sit down and figure out at some point what i actually want out of an interpreter and where the line between that and compiled is...
<pie_>
joepie91: youre me right but you have a head start, what is is that i want here? :P ^
<evanjs>
pie_: I was sort of scared about what you were saying for a second haha
<pie_>
evanjs: i know just enough to be dangerous :^)
rajivr has joined #nixos-chat
<evanjs>
well, that was "difficult" /s. wonder if the maintainer wants off after the oxidization, though :P Guess I'll ask in the PR
<evanjs>
Mrm, I guess we're still blocked on the license, though
<evanjs>
It only appears in Cargo.toml right now, soooooo....
<joepie91>
pie_: yeah so llvm sounds like the closest thing, with a runner-up being language-agnostic debugger protocols like that used by the VS Code debugger
tilcreator has quit [Quit: Idle for 30+ days]
<eyJhb>
When did this `rm -rf *` became illegal to do?
<eyJhb>
become*
<eyJhb>
Ohh, of course. I had a file named --help :p
<etu>
:D:D
<joepie91>
eyJhb: that seems like a fun troll
<joepie91>
that would break so many shell scripts I bet
<eyJhb>
joepie91: I am just sitting pondering, if you can do some fun stuff regarding how it handles unicode chars
FRidh has quit [Quit: Konversation terminated!]
<MichaelRaskin>
I wonder if it is a good idea to make nix-daemon configurable to set some extra groups to builders during fetching. Would allow rooted Android to use unmodified Nix, at least inside a chroot…
<philipp[m]>
Somebody here was talking about some weird keyboard a while back and wanted it for android. Is this it?
<joepie91>
that may have been me
<joepie91>
it looks like something in vaguely the same family as what I'm thinking of
<joepie91>
but not quite the same
<joepie91>
the one I'm thinking of was a 3x3 grid with 3x3 subgrid
<philipp[m]>
it's fun.
<philipp[m]>
But I think I'll always be faster with a normal keyboard.
<philipp[m]>
I always think there should be more grid. Too few possibilities this way.
ninjin has joined #nixos-chat
<eyJhb>
Miss the old T9
<joepie91>
eyJhb: I actually have a USB remote control that implements T9 :P
<joepie91>
it's really neatly done
<philipp[m]>
Do we miss t9 or do we miss not typing anything longer than 160 chars in a phone?
<joepie91>
eyJhb: registers as a standard keyboard, just types the initial letter on the first press, and then if you re-press the same button within the timeout it does backspace + the next letter for that button, etc.
<joepie91>
very clever
<eyJhb>
philipp[m]: NO LONGER THAN 160 CHARS! I typed essays into that thing :p I miss the physical buttons
<eyJhb>
Cool!
<eyJhb>
Also, how the hell does 8vim work?
<philipp[m]>
Depends in what type of pain you like I guess.
* joepie91
still regularly types T9
<joepie91>
(I use a dumbphone and occasionally people text me)
<philipp[m]>
I can see that muscle memory could be better with time but starting out the way I'm doing it right now is painfully slow.
<MichaelRaskin>
joepie91: I think what you describe is not T9 though?
<MichaelRaskin>
I am pretty sure T9 included dictionary-based guessing even if you only pressed each digit once
<joepie91>
MichaelRaskin: annoyingly T9 means different things depending on who you talk to
<joepie91>
either the numpad->letter mapping or the predictive model
__monty__ has joined #nixos-chat
<joepie91>
it's one of those terms where common usage doesn't really match official meaning, I guess it's a weird form of genericide
<joepie91>
or well, formerly-common, because noone talks about T9 anymore :D
<MichaelRaskin>
I believe the featurephones themslelves by now are mostly in agreement about the latter choice. At least in Europe+Russia…
<philipp[m]>
I only know what MichaelRaskin describes as t9 and I remember my first phone with it.
<__monty__>
Tap 2 thrice for C, 3 twice for e, etc?
<MichaelRaskin>
That's non-predictive
<MichaelRaskin>
I think nowadays Wikipedia is pretty confident T9 is predictive
<philipp[m]>
>This article needs additional citations for verification.
<MichaelRaskin>
Look at the Talk, they are debating whatever but not whether non-predictive is also T9
<eyJhb>
Not sure if it makes sense to type like that
<__monty__>
The patents linked do seem to specifically describe the predictive systems.
<philipp[m]>
German Wikipedia is sure that t9 needs to be predictive.
<__monty__>
I guess the non-predictive letters-on-digits is "just" what american phone keyboards have been like since forever?
<joepie91>
MichaelRaskin: my dumbphone doesn't call anything T9
<MichaelRaskin>
Does it even have predictive mode?
<joepie91>
predictive mode is just the capitalization state + a "goes fast" swoosh icon next to it
<joepie91>
yes
nckx has joined #nixos-chat
<philipp[m]>
This is the kind of stuff, people would have duelled over 200 years ago, right?
<joepie91>
:P
<__monty__>
My T303 doesn't mention T9 either. A samsung flip phone did though.
<eyJhb>
damn it philipp[m] , I am super intrigued by 8vim now....
<philipp[m]>
Then try it :D
dada_da has quit [*.net *.split]
aleph- has quit [*.net *.split]
monsieurp has quit [*.net *.split]
vesper11 has quit [*.net *.split]
pie_ has quit [*.net *.split]
makefu has quit [*.net *.split]
lucus16 has quit [*.net *.split]
srk has quit [*.net *.split]
hyperfekt has quit [*.net *.split]
hyperfekt has joined #nixos-chat
monsieurp has joined #nixos-chat
<eyJhb>
I have it installed and used it a little philipp[m] , it is quite nice!
srk has joined #nixos-chat
lucus16 has joined #nixos-chat
aleph- has joined #nixos-chat
<eyJhb>
But I fail to see, that it will ever be faster than omji board swiping across the keyboard :p
<eyJhb>
Let me find the correct word for it
makefu has joined #nixos-chat
<eyJhb>
ouija board! There it was. But basically, and swiping keyboard :D Unless you code, then maybe it could be cool
pie_ has joined #nixos-chat
dada_da has joined #nixos-chat
bbigras has quit [*.net *.split]
CRTified[m] has quit [*.net *.split]
feepo has quit [*.net *.split]
steveeJ has quit [*.net *.split]
liszt_ has quit [*.net *.split]
kgz has quit [*.net *.split]
johanot has quit [*.net *.split]
talyz has quit [*.net *.split]
CRTified[m] has joined #nixos-chat
feepo has joined #nixos-chat
bbigras has joined #nixos-chat
talyz has joined #nixos-chat
johanot has joined #nixos-chat
kgz has joined #nixos-chat
liszt_ has joined #nixos-chat
steveeJ has joined #nixos-chat
kgz has quit [Max SendQ exceeded]
kgz has joined #nixos-chat
thefloweringash has quit [Ping timeout: 244 seconds]
philipp[m] has quit [Ping timeout: 244 seconds]
JJJollyjim has quit [Ping timeout: 244 seconds]
artturin has quit [Ping timeout: 244 seconds]
worldofpeace has quit [Ping timeout: 240 seconds]
crazazy[m] has quit [Ping timeout: 240 seconds]
emily has quit [Ping timeout: 240 seconds]
pinage404[m] has quit [Ping timeout: 240 seconds]
DavHau[m] has quit [Ping timeout: 240 seconds]
rycee has quit [Ping timeout: 240 seconds]
arcnmx has quit [Ping timeout: 240 seconds]
Hm7000 has quit [Ping timeout: 260 seconds]
danielrf[m] has quit [Ping timeout: 244 seconds]
Irenes[m] has quit [Ping timeout: 244 seconds]
puzzlewolf has quit [Ping timeout: 244 seconds]
manveru[m] has quit [Ping timeout: 244 seconds]
Ox4A6F has quit [Ping timeout: 240 seconds]
kraem has quit [Ping timeout: 240 seconds]
aterius has quit [Ping timeout: 268 seconds]
nicolas[m] has quit [Ping timeout: 268 seconds]
jtojnar has quit [Ping timeout: 268 seconds]
ma27[m] has quit [Ping timeout: 268 seconds]
balsoft has quit [Ping timeout: 268 seconds]
aanderse has quit [Ping timeout: 268 seconds]
leonardp has quit [Ping timeout: 268 seconds]
aaronjanse has quit [Ping timeout: 268 seconds]
colemickens has quit [Ping timeout: 246 seconds]
leons has quit [Ping timeout: 264 seconds]
bbigras has quit [Ping timeout: 246 seconds]
Armin[m] has quit [Ping timeout: 240 seconds]
danderson[m] has quit [Ping timeout: 240 seconds]
CRTified[m] has quit [Ping timeout: 246 seconds]
<pie_>
joepie91: during morning shower thoughts i figured out that half the thing is that these are easy to decompile, so i guess really i want source (even if slightly harder to get)
<pie_>
thats on one side, now i just gotta figure out what it is about *using* interpreters for writing code
<pie_>
though probably its just the repl style experimentation
wildtrees has joined #nixos-chat
danielrf[m] has joined #nixos-chat
emily has joined #nixos-chat
manveru[m] has joined #nixos-chat
thefloweringash has joined #nixos-chat
worldofpeace has joined #nixos-chat
rycee has joined #nixos-chat
arcnmx has joined #nixos-chat
pinage404[m] has joined #nixos-chat
philipp[m] has joined #nixos-chat
Irenes[m] has joined #nixos-chat
puzzlewolf has joined #nixos-chat
crazazy[m] has joined #nixos-chat
DavHau[m] has joined #nixos-chat
aaronjanse has joined #nixos-chat
wildtrees has quit [Remote host closed the connection]
wildtrees has joined #nixos-chat
<gchristensen>
I'm looking at the output of `objdump -d` ... is there a nice resource on learning how to read it properly? specifically it has "callq a200 <LivenssDetection@plt>" and I want to know exactly what the <...> bit is saying
JJJollyjim has joined #nixos-chat
<JJJollyjim>
okay time to 20.09 my servers
<gchristensen>
oh @plt is the .pltsection, and it is calling LivenssDetection in .plt
Hm7000 has joined #nixos-chat
<JJJollyjim>
yeah
<JJJollyjim>
just the address of that symbol
<gchristensen>
but then why is it saying "undefined symbol LivenssDetection" when this .so is loaded ... a200 is indeed .plt's LivenssDetection ...
<JJJollyjim>
hey i recognize that file lmao
<gchristensen>
yeah? :)
<JJJollyjim>
is that libfprint-tod-goodix?
<gchristensen>
yes...!
<JJJollyjim>
lmao
artturin has joined #nixos-chat
<JJJollyjim>
i've been REing the goodix driver to write a native linux one
<leonardp>
joepie91: good to know :)... short version: how do i cache the /nix/store between builds/stages when using nixos/nix:latest running in a gitlab/gitlab-runner container?
<eyJhb>
Seems like a Gitlab question?
<eyJhb>
Else build a image that have all you need, and use that as a image :p
<gchristensen>
JJJollyjim: I don't suppose you know where LivenssDetection is supposed to come from
<joepie91>
JJJollyjim: it has a cutoff of N lines (configurable in the bridge) beyond which it auto-pastes, to prevent spamming IRC by someone who is unfamiliar with IRC not having multi-line messages
<joepie91>
the substitute message really should have a first-line preview though
<leonardp>
eyJhb: that was my first approach but my upload speed is incredibly slow... -.-
<__monty__>
joepie91: And shorter urls would be nice tbh.
<joepie91>
__monty__: not realistically possible without a separate URL shortener
<joepie91>
so it could be an optional add-on at best
<__monty__>
I don't really care about the how. The urls always overflow a line. Messes with readability and clickability.
<eyJhb>
leonardp: Don't really think you have any other option. Build on a remote (DigitalOcean,etc) and upload from there?
<MichaelRaskin>
Now I want to try all my terminals to see if overflow breaks clickability anywhere
<eyJhb>
Is your speed in the Kbit/s, or just a huge image?
<MichaelRaskin>
(Human messages routinely overflow 8 characters, so readability impact of multiline is something I am just used to)
<leonardp>
eyJhb: huge image.. i think i'll run it overnight..
<leonardp>
and pray my isp doesn't reset the connection
<eyJhb>
How huge is huge leonardp ?
<eyJhb>
Huge is like, 10+ GB
<JJJollyjim>
gchristensen oh hmm no
<JJJollyjim>
lemme have a look
<JJJollyjim>
is this -53xc-0.0.4?
<leonardp>
eyJhb: well it's more like 1.6 GB :)
<__monty__>
MichaelRaskin: Irssi at least starts the url on the next line even if there's only a single character on the current line and it flows urls from column 80 to column 0 rather than indenting to align with the previous line for clickability. This means the timestamps are sometimes included in the url when clicking. I know weechat *does* align even when wrapping urls but that makes it impossible to
<gchristensen>
JJJollyjim: and 0.0.6
<__monty__>
click them without an url copying mode.
<eyJhb>
leonardp: That is less than LaTeX! :p
<leonardp>
haha true!
<eyJhb>
Compressed or as a docker image size?
<leonardp>
oh.. i forgot about compression
<MichaelRaskin>
__monty__: hmm, indeed, I am too used to my ii + less -F + screen setup
<eyJhb>
Because, LaTeX was around 3.6 GB in tar.gz.
<leonardp>
that should speed up things a lot!
<__monty__>
I'm not saying these are unfixable problems btw. Just that shorter urls would be *way* better.
<eyJhb>
Hopefully
<eyJhb>
But you can always spin up a VM and then do it there
<leonardp>
maybe i'll get the host server to comply in building it...
<leonardp>
but he's not exactly well-tempered
<leonardp>
eyJhb: i'm stupid... i can just run the gitlab runner on my NixOS raspberrypi and upload the artifacts (which are small)
<JJJollyjim>
gchristensen: yep, it definitely doesn't exist
<gchristensen>
how does it work on Ubuntu ... :/
<JJJollyjim>
dlopen is called with RTLD_LAZY, so it succeeds anyway
<gchristensen>
it fails here -- let me take a look ...
<JJJollyjim>
i expect it will segfault if it tries to call it
<JJJollyjim>
what fails?
<JJJollyjim>
on arch i can run libfprint fine
<JJJollyjim>
with the aur package, and it happily loads it
<gchristensen>
interesting, mine doesn't use dlopen
<JJJollyjim>
oh?
<JJJollyjim>
libfprint-tod should be loading .so files from a directory, right?
<eyJhb>
leonardp: Yeah, that is a a way of doing it as well!
<gchristensen>
I'm setting FP_TOD_DRIVERS_DIR to the directory containing the .so, it says "Opening driver ...." and then "Impossible to load module ...: undefined symbol: LivenssDetection"
<eyJhb>
What are you building?
<JJJollyjim>
ah, how do you turn on debug logging?
<gchristensen>
G_MESSAGES_DEBUG=all G_DEBUG=all
<JJJollyjim>
oh yeah wtf
<JJJollyjim>
i swear it worked for me a second ago
<JJJollyjim>
yeah it worked that time
<JJJollyjim>
uhhh
<JJJollyjim>
G_DEBUG breaks it
<JJJollyjim>
if you run it with just G_MESSAGES_DEBUG, it works
<gchristensen>
.....!
<JJJollyjim>
lmao wtf
<JJJollyjim>
im gonna reverse engineer this, i bet it's an obfuscation thing
<JJJollyjim>
they don't want me getting their proprietary livenss detection algorithm
weeqw has joined #nixos-chat
cole-h has joined #nixos-chat
<JJJollyjim>
defence in depth: 1. don't run with G_DEBUG, 2. misspell the name of all your functions
<JJJollyjim>
(the later is a common theme in all goodix driver code lol)
<gchristensen>
hilarious.
<JJJollyjim>
oh lol
<JJJollyjim>
G_DEBUG turns on various things that make debugging easier
<JJJollyjim>
such as
<JJJollyjim>
bind-now-modules
<JJJollyjim>
All modules loaded by GModule will bind their symbols at load time, even when the code uses %G_MODULE_BIND_LAZY.
<JJJollyjim>
<JJJollyjim>
i.e. exactly what makes this work lol
<JJJollyjim>
ok maybe this is just a bug, not malice
<gchristensen>
ah ha!
weeqw has quit [Quit: Leaving]
<cole-h>
srhb: Is it also built against thrussh 0.30.1? I didn't have to do anything special to get it to work...
rajivr has quit [Quit: Connection closed for inactivity]
<gchristensen>
JJJollyjim: thanks!
<gchristensen>
JJJollyjim: is there a way to visualize the data file it produced?
<JJJollyjim>
the fingerprint image?
<gchristensen>
yea
<cole-h>
srhb: If you built it yourself, maybe you need to `cargo update`
<gchristensen>
`file` tells me it is just "data"
<JJJollyjim>
oh hmm what's the file path/name?
<srhb>
cole-h: I just noticed the server still doesn't acknowledge me. I assume they have to fix something on their end.
<cole-h>
Just a sanity check, but you added your key to the nest, right?
<srhb>
cole-h: I did, yes :)
<cole-h>
Technology 🤯
<srhb>
Indeed. :D
<cole-h>
srhb: I'd run it with RUST_LOG=debug and see what it spits out/where it gets stuck/whatever
<JJJollyjim>
gchristensen: if it's using standard libfprint methods, there's a gtk debug UI you can get with the gtk-examples meson flag
<JJJollyjim>
but i suspect it's using its own proprietary thing
<srhb>
cole-h: No need I think, until ssh itself works..
<srhb>
Unless the server is available somewhere for testing?
<JJJollyjim>
if it's like the windows one, that file is encrypted
<cole-h>
Nest is closed-source at the moment
<JJJollyjim>
cos the windows one does stuff in Intel SGX
<srhb>
cole-h: Bleh.
<JJJollyjim>
but i assume it's not encrypted here
<JJJollyjim>
cos i dont think sgx works on linux
<cole-h>
srhb: Can you (try to) ssh into the nest with debug stuff turned on and see if authentication succeeds?
<cole-h>
`ssh srhb@nest.pijul.com -vvvv` is what I did
<JJJollyjim>
gchristensen: also fun fact your fingerprint sensor is communicating with that library over TLS-over-USB lmao
<gchristensen>
JJJollyjim: awesome, thank you a lot! the filename is var/lib/fprint/grahamc/goodix-tod/0/7
<gchristensen>
hilarious.....
<gchristensen>
I did notice the private keys in `strings`
<JJJollyjim>
yeah lol it's very Engineered(tm)
<JJJollyjim>
the theory being that a remote server can trust the SGX's attestation that you just scanned your fingerprint
<JJJollyjim>
but there are a lot of holes in it
<eyJhb>
Debugging adb like, is it the cable or my computer? Because 99% of the cables I have, do not have the data connected...
<cole-h>
If that works (you'll probably need to do the "Enter -> ~." trick to exit the connection), I'd try to get in contact with Pierre with logs. (Sorry I can't be of more help)
<gchristensen>
that is a cool idea
<gchristensen>
okay so the pam integration works great. swayidle's screen locker doesn't care about it yet... todo :)
<gchristensen>
adventures in online grocery shopping: almost bought 20 pounds of oranges
<eyJhb>
gchristensen: Are you sure that you do not secretly want 20 pounds of oranges?
<MichaelRaskin>
That's only twice how much I sometimes end up transiently owning for my own consumption
<MichaelRaskin>
Presumably Graham could get some help with oranges, making the doubled amount also perfectly doable
<srhb>
cole-h: Yeah. that's what I was testing, and no, it does not succeed.
<gchristensen>
I mean it'd be fine, but I wanted to just by 5
<eyJhb>
> lbs
<{^_^}>
undefined variable 'lbs' at (string):359:1
<eyJhb>
We need a converter
<eyJhb>
Seems doable
<MichaelRaskin>
Hook up the entire units to the bot?
<JJJollyjim>
> "${20 / 2.205} kg"
<{^_^}>
cannot coerce a float to a string, at (string):359:2
<MichaelRaskin>
Correction: only 1.5× times as much as I sometimes have
<infinisil>
I've written a unit library before, maybe I should do the same in Nix :P
<MichaelRaskin>
OK, just make sure to import at least the entire units database of units maybe then?
<infinisil>
That would be nice yeah
<infinisil>
But tbh, Nix is pretty bad for implementing unit calculations..
<infinisil>
Unless..
<MichaelRaskin>
Oh, their database is a nice text file
<infinisil>
> let __mul = a: b: "hello"; in 10 * 20
<{^_^}>
"hello"
<infinisil>
Nice, this might work
<infinisil>
MichaelRaskin: Afaik there's no standard unit database, so which one are you talking abuot?
<cirno-999>
what is nixusers' favorite music player?
turlando has joined #nixos-chat
<srhb>
> let __mul = f: g: x: f (g x); addTwo = x: x+2; in (addTwo * addTwo * addTwo) 42
<{^_^}>
48
<srhb>
Well well well how the turntables
<srhb>
Function composition at long last.
<srhb>
I'll be over here mutilating every nix repo in sigh.
<srhb>
sight*
<MichaelRaskin>
Are you _sure_ the first version of the phrase is not the correct one?
<srhb>
No. :D
<srhb>
> let __mul = f: g: x: f (g x); addTwo = x: x+2; __div = lib.dk; in addTwo * addTwo * addTwo / 42
<{^_^}>
attribute 'dk' missing, at (string):361:56
<srhb>
> let __mul = f: g: x: f (g x); addTwo = x: x+2; __div = lib.id; in addTwo * addTwo * addTwo / 42
<{^_^}>
48
<srhb>
*ASCENDANCE*
<cole-h>
srhb: I read in one of the discussions that there were IPv6 troubles -- what if you try to ssh into srhb@51.210.158.166?
<srhb>
cole-h: Same deal. As far as I can tell, the server is just refusing my key. It's also difficult to test because it appears to ban me for a while afterwards. :P
<eyJhb>
Is there any nice reads about pijul, to get an idea of why it is "nicer"?
<gchristensen>
I'm interested because 1. it actually uses patches, 2. it supports patching more than just lines
<__monty__>
eyJhb: I think you'd find more info if you look into why darcs is nice and why it's not so nice. Because pijul basically started as darcs with the performance problems fixed.
<MichaelRaskin>
Of course, performance was not the only gotcha with darcs…
<__monty__>
What are you thinking of? I never used it.
<infinisil>
Y'all, we have units in Nix now!
<infinisil>
> 20 * pounds / kg
<{^_^}>
undefined variable 'pounds' at (string):397:6
<infinisil>
> 20 * pound / kg
<{^_^}>
"9.071847"
<infinisil>
That's how many kgs in 20 pounds :)
<MichaelRaskin>
A lot of time, what people really want to store is a series of patches as they happenned, and another as they _pretend_ all of this has happenned until they are into deep debugging and ask what was ever tested and need the first history.
<infinisil>
eyJhb: ^
<sphalerite>
> 20 * km / kg
<{^_^}>
undefined variable 'km' at (string):397:6
<sphalerite>
> 20 * kilometre / kilogram
<{^_^}>
undefined variable 'kilometre' at (string):397:6
<sphalerite>
> 20 * kilometee / kilogram
<{^_^}>
undefined variable 'kilometee' at (string):397:6
<sphalerite>
> 20 * kilometer / kilogram
<{^_^}>
undefined variable 'kilometer' at (string):397:6
<infinisil>
No prefixed versions yet
<infinisil>
> km = 1000 * m
<{^_^}>
km defined
<sphalerite>
> 20 * metre / gram
<{^_^}>
undefined variable 'metre' at (string):398:6
<sphalerite>
> 20 * meter / gram
<{^_^}>
undefined variable 'meter' at (string):398:6
<infinisil>
Well, except kg, which is the standard SI unit instead of grams..
<sphalerite>
oh, just m >_<
<sphalerite>
20 * m / kg
<sphalerite>
> 20 * m / kg
<{^_^}>
"20.000000 m kg^-1"
<sphalerite>
nice
<infinisil>
> g = 0.001 * kg
<{^_^}>
g defined
<MichaelRaskin>
Aaaand now we have a conflict between gram-g and freefall-g
<infinisil>
Oh no
<__monty__>
Meh, acceleration of gravity is just 10 anyway.
<__monty__>
MichaelRaskin: Doesn't tagging of patchsets fix that? If every released set of patches is available isn't that enough?
<infinisil>
> earthRadius
<{^_^}>
"6378100.000000 m"
<lassulus>
maybe {^_^} is living in a very high tower?
<MichaelRaskin>
I think people doing fine measurements ask whether the equipment is for Munich or London and set it up differently
<MichaelRaskin>
__monty__: you mean, let's tag every moment that has ever happenned?
<MichaelRaskin>
That kind of works
<__monty__>
No, I'm not suggesting reimplementing git on top of darcs : )
<MichaelRaskin>
It won't be git
<MichaelRaskin>
It's not easy to mess up the model _that_ bad
<MichaelRaskin>
Although, both Git and Darcs kind of succeeded
<__monty__>
I mean, does *every* step really matter?
<MichaelRaskin>
Either or
<__monty__>
Because you can bisect between patches afaik.
<MichaelRaskin>
Either you admit that Git stores too much data still, it is just a backup of recent changes and then ancient releases
<MichaelRaskin>
Or you start thinking in terms «has any human being ever thought that serialising this specific state onto a storage unit is a good idea»
<MichaelRaskin>
And when you start desperate debugging, well
<eyJhb>
Nice job infinisil !
<eyJhb>
Now we just need all currencies to point to DKK
<infinisil>
Hehe
<eyJhb>
You guys love it!
<eyJhb>
> 10 DKK
<{^_^}>
attempt to call something which is not a function but an integer, at (string):399:1
<eyJhb>
> DKK 10
<{^_^}>
"34650.100000 VND"
<eyJhb>
> VND 34650.1
<{^_^}>
"1.316704 EUR"
<infinisil>
You know, for currencies we just need to define all of them in terms of USD
<eyJhb>
Don't you dare! :( Then I am screwed once more just like pounds
<MichaelRaskin>
I think given the project demographics, EUR
<infinisil>
Which would allow using `10 * DKK / EUR` to convert from DKK to EUR
<eyJhb>
Do it in regards to EUR :p
<eyJhb>
Do it for EUR :p
<infinisil>
(but of course, money changes its value, so we should probably fetch the rates from somewhere to have them stay up-to-date)
<__monty__>
You should use a novelty reference currency. Way more fun that way.
<sphalerite>
infinisil: use gold as the reference :p
<sphalerite>
__monty__: you mean like vnd?
<sphalerite>
I mean, given the demographics, as MichaelRaskin said.
<infinisil>
Hmm, gold actually might be even worse, because the price of it differs all over the world, depending on manufacturer and whatnot!
<infinisil>
(afaik at least)
<MichaelRaskin>
Price of banking ignots is probably more unified
<MichaelRaskin>
But it just oscillates too much
<sphalerite>
or oil :p
<MichaelRaskin>
EEA currencies kind of have to have inertia w.r.t. EUR.
<infinisil>
which I used as a reference for the (partial) Nix implementation
cjpbirkbeck has joined #nixos-chat
neeasade has quit []
<sphalerite>
I want to (ab)use a mini-PC I have here as a wireless AP. However, the wifi driver doesn't allow adding the wifi interface to a bridge (my idea was to create a bridge containing the wifi interface and a VLAN interface on the ethernet link). Is anyone aware of a way around this?
<sphalerite>
re-asked in #nixos-on-your-router.
avn has quit [Ping timeout: 260 seconds]
dada_da has quit [Ping timeout: 240 seconds]
dada_da has joined #nixos-chat
<eyJhb>
I just installed Android 10, and I had to use the nano GAPPS bundle. Now I don't have a danish dictionary in my ASOP keyboard, and I can't download any. Any clue what to do? I am even unsure, if it because of the none gaaps
<__monty__>
Who needs a danish dictionary anyway?
<samueldr>
pastries?
vesper11 has joined #nixos-chat
<joepie91>
surprise surprise. the JS package author who is notorious for cheating on benchmarks and generally being overly full of himself... is a Trump supporter
<eyJhb>
:( Well then, I guess I must just use the Frenche one from now on out
<joepie91>
who could have expected that...
<eyJhb>
joepie91: Links?
<eyJhb>
To the some NPM stuff?
<eyJhb>
To the some <-
<joepie91>
eyJhb: to which part of what I described? :P
<{^_^}>
jonschlinkert/repeat-string#7 (by RealDolos, 4 years ago, closed): The benchmarks are somewhat pointless
<joepie91>
basically it's benchmarking "returning a value from a variable", not the actual module logic
<joepie91>
because it caches the args + result for the latest invocation, and then repeats the same invocation thousands of times in the benchmark, but no actual code would do that
<drakonis>
lmao
<drakonis>
amazing
<joepie91>
I think I ran into another case of this from the same guy a while ago, but I forgot which that was
<joepie91>
his code is also usually "clever" code
<joepie91>
in the sense of code that seems written to show off his knowledge of obscure language features
<eyJhb>
I want the deleted part
<eyJhb>
Why do they remove all the good things...
<joepie91>
close to code golfing
<joepie91>
anyway basically his code sucks to audit
<eyJhb>
But hey, at least you are paid to do it?
<joepie91>
I've been keeping him out of my dep tree for a while now due to the code quality issues, and apparently there's more than one reason to do that now...
<joepie91>
eyJhb: heh, sometimes.
<eyJhb>
No Trump supporters allowed?
<joepie91>
eyJhb: I don't trust maintainers who can accept and actively support such a lack of grounding in reality
<joepie91>
using a dependency is a trust relationship
<eyJhb>
Depends on the reality you believe in
<joepie91>
and that usually goes fine, but this is a point where someone lands in the "uhhhhhhh" bucket
<__monty__>
You figured out he supports Trump from his code? Yikes o.O
<eyJhb>
// this code caches the thing, and might be the very best, just like TRUMP TRUMP TRUMP <--- Might be why?
<joepie91>
display name also speaks volumes
<joepie91>
eyJhb: wait, does it actually say that now?
<eyJhb>
No no, but one can hope
<joepie91>
ah ok :p
<eyJhb>
I am pondering where to put meta data in nix atm...
<joepie91>
anyway, the reason this does not surprise me is because people who are unreasonably full of themselves tend to seek out other people who are unreasonably full of themselves...
<eyJhb>
Normally it goes into each mkDerivation, but... I return config and build, on each "thing"
<joepie91>
and in all of jon's behaviour I'm seeing similar traits to how Trump behaves
<joepie91>
like in that benchmark issue I linked
<eyJhb>
Jon for president in 2024?
<eyJhb>
2025?
<joepie91>
rather not :)
<joepie91>
techbros are doing enough damage as-is
<joepie91>
don't need them in office
<eyJhb>
Some of them could be an improvement I think
<__monty__>
"I blocked you because you posted a similarly critical issue on a project by a user I've blocked." What? That's some weird reasoning.
<joepie91>
whee, rsync completed, time to retry my server migration...
<joepie91>
__monty__: yeah, that's the sort of thing I mean. sidestepping the actual issue, looking for some personal attack towards the criticizing person, however much of a stretch it is
<joepie91>
that's one of the things where I'm seeing similarity between Trump and jonschlinkert
<joepie91>
well, Trump + following
<eyJhb>
joepie91: from what to what?
<joepie91>
eyJhb: Debian, Gogs, old PostgreSQL -> NixOS, Gitea, new PostgreSQL
<eyJhb>
Why no more Gogs? :(
<samueldr>
isn't gitea basically gogs, but with more contributors?
<samueldr>
or have they stopped following gogs?
<eyJhb>
Running tests for software/programs is something, running test to see if a crypto challenge will work with a new flag, is just spiteful to your computer.
<eyJhb>
Wasn't it the other way a around?
<samueldr>
gitea is a fork of gogs (just verified)
<eyJhb>
Well okay, maybe gitea is the way to go..
<eyJhb>
Seems cool as well
<eyJhb>
The `tea` in it, just makes me think of Java, and then I want to run
<eyJhb>
Whereas gogs, as Go in it
<eyJhb>
And therefore it gets hugs. - And before anybody starts, no I have not done that much Java. Only for algorithm courses
<samueldr>
>> The `tea` in it, just makes me think of Java, and then I want to run [it]
<samueldr>
;)
<eyJhb>
Which do not add love to the source
<eyJhb>
NO!
<eyJhb>
Bad samueldr ! :(
<joepie91>
eyJhb: Gitea is the way to go yeah, it forked off because the Gogs maintainer wanted to keep strict control over the project
<joepie91>
so Gogs is the version from the official maintainer and Gitea is the community-led version
<jtojnar>
I do not see them coming very far with their current knowledge
<jtojnar>
though they did manage to install NixOS, apparently
<eyJhb>
fyi. sommerfugl -> butterfly
<eyJhb>
How the hell did they install NixOS and make such a config without having used ls much?
<eyJhb>
Point taken jtojnar
<MichaelRaskin>
I guess some people are good at following and adapting the instructions from a manual! It's just that they never had a reason to read any manual describing how to use basic POSIX utilitiesP
<pie_>
i just heard a pretty good verdict of nix/os
<pie_>
"Hi x! I've used nix for a bunch of projects now and it's really a mixed bag. It's one of those bridge technologies that forces the user to become an expert in multiple disparate systems. In the case of nix, I feel like I need to be an expert not only in the nix language, but also in the arbitrary and changing ways nix's haskell rules function. In addition, the user needs to understand how cabal works. Also, there don't a
<pie_>
o be any reasonable ways to debug nix other than caveman debugging. But, when nix works, the results are great!"
<eyJhb>
Only a year later, and I already have Docker builds breaking because of bad Dockerfiles....
<gchristensen>
a whole year?
<eyJhb>
Granted, I have not tried building it in-betweeen
<eyJhb>
in-between* so it might have been a couple of months :p
<eyJhb>
Luckily it is 1 out of 25 Dockerfiles
<MichaelRaskin>
Nix doesn't force the user to become an expert. It's perfectly fine serving users having previously achieved experthood
<eyJhb>
Anyways, night ! :(
<eyJhb>
:) **
<infinisil>
Just created a script called "pot"
<infinisil>
Stands for "pulseaudio output toggle" :P
<infinisil>
Changes the default audio output to the next device just by calling it
<infinisil>
Saves me always having to open pavucontrol and manually clicking the output I want