<lovesegfault>
vim $(rg -o --pcre2 '(?<=start\ \")\/nix\/store\/.*?-git' "$(nix-build -A foucault)")/activate
<lovesegfault>
el master hacko
parsley936 has quit [Remote host closed the connection]
<gchristensen>
lovesegfault: no
<lovesegfault>
gchristensen: it's hyperscale
<lovesegfault>
synergy
<gchristensen>
lol I mean the zpool.cache thing
<lovesegfault>
gchristensen: Ah, okay, I thought you were hack-shaming me
<lovesegfault>
I stopped using zfs for / and removed the rollback; I just mount it as a tmpfs now
<gchristensen>
my best code is a hack
<gchristensen>
maybe that doesn't say much about my code
<lovesegfault>
My best code is a hack too
<iqubic>
So, I know I'm going to doing something like this: "mkdir pkgs/applications/audio/castersoundboard", but I have no idea where to put the hyphens in that name.
<DigitalKiwi>
caster-soundboard matches the upstream capitalization
<iqubic>
DigitalKiwi: I'll do that then.
<iqubic>
I'm going to need to clean this up a bit:
<{^_^}>
nix-community/impermanence#3 (by lovesegfault, 10 seconds ago, open): impermanence: bind mount with correct permissions
<iqubic>
Is the version field required? I'm asking because I can't seem to find any version numbers here.
<iqubic>
Oh wait... I can't just do a fetchFromGithub can I? The source is one level deep in the github repo.
<cole-h>
lovesegfault: nit: to make the bash scripting more readable, needs more whitespace
<cole-h>
My rule of thumb is basically "newline before a comment"
<iqubic>
How do I find the correct rev for this?
<iqubic>
I want to know what rev to pull from for fetchFromGithub
<cole-h>
rev = git hash
<iqubic>
I know that. I just don't know how to find that.
<gchristensen>
(this would be better for #nixos)
<gchristensen>
my touchpad has a bootloader version?
<iqubic>
gchristensen: Are you saying that my questions would be better for #nixos?
<gchristensen>
yea
<samueldr>
gchristensen: sounds plausible, now is it running nixos, and if not, why aren't you working on that?
<gchristensen>
very good question
<samueldr>
it's likely you have an embedded controller that is "hella smart" that handles not only the touchpad
<samueldr>
(thinking keyboard and *some* stuff like leds and such)
<samueldr>
aaaand, a closed source component of most machines
<gchristensen>
:(
<samueldr>
sums up my feelings when I realised as much
<samueldr>
annoying when you know that fact, there is a firmware, that you probably cannot fix, that needs fixing
<lovesegfault>
cole-h: editing
<cole-h>
ty
<lovesegfault>
cole-h: done :)
<cole-h>
lovesegfault: Another nit: the `unset`s should be slightly separated from the rest of the stuff (e.g. `previousPath=...\n\n unset ...` rather than just `previousPath=...\n unset ...`)
<lovesegfault>
fair, I mostly unset out of paranoia
<lovesegfault>
done
<cole-h>
lovesegfault: Final nit (more personal style, so feel free to disregard) I prefer control flow separated as well (e.g. `previousPath="/"\n\n for pathPart...` and `targetPath=...\n\n if ...`)
<gchristensen>
you canunset by creatinga new (...) scope
<gchristensen>
avoid unset*
<cole-h>
separated from variable declarations*
<lovesegfault>
gchristensen: Hm, maybe
<gchristensen>
also I wonder what it'd take to get shellcheck running on those scripts
<cole-h>
copy it out and run it manually? :P
<lovesegfault>
cole-h: I lest previousPath attached to the for b/c it's the accumulated state of the for
<gchristensen>
cole-h: or break it out ot a script which is executed with arguments
<lovesegfault>
gchristensen: I generated them, copied them from the activation file, and pasted into a .sh file and ran shellcheck
<lovesegfault>
came out clean
<gchristensen>
yah but I mean like forever
<cole-h>
lovesegfault: I understand. But I don't usually do that in e.g. my Rust code (`let i = 0;\n\n for x in i..100 { ... }`). Just looks cleaner IME
<lovesegfault>
Ah, yeah, the only way is to extract them
<cole-h>
lovesegfault: Why not extract them and use @file@ and stuff with the substitute family?
<lovesegfault>
cole-h: I'm not against that, I just don't want to do it :P
<cole-h>
Heh
<lovesegfault>
After this is merged that can be a task
<lovesegfault>
talyz: Are you around?
<cole-h>
Hmm, I wonder how shellcheck handles @foo@
<cole-h>
Doesn't appear to dislike it
<gchristensen>
for args?
<cole-h>
Yeah
<cole-h>
e.g. `sourcePath="@persistentStoragePath@@file@"`
<cole-h>
Doesn't appear to dislike it
<cole-h>
(Oh I sent that already oops)
<gchristensen>
why not just pass them in as arguments?
<cole-h>
That works too
<lovesegfault>
Alright, fully bind-mounted system take one
andi- has quit [Remote host closed the connection]
andi- has joined #nixos-chat
Valodim has quit [Ping timeout: 260 seconds]
drakonis has quit [Read error: Connection reset by peer]
Peetz0r has quit [Quit: Doej!]
slack1256 has quit [Ping timeout: 256 seconds]
Peetz0r has joined #nixos-chat
<etu>
lovesegfault: oh, wow, good job! That's a pile of bash :) Then we only want to wrap the users API right as a last step. Should be fairly easy now with the creation that flexible and permissions taken care of :)
<etu>
lovesegfault: I'll test it in... a bunch of hours or so, like in 5 hours when I have time free :)
<etu>
✨ lovesegfault
<{^_^}>
lovesegfault was put on Santa's "nice" list
AluisioASG has quit [Read error: Connection reset by peer]
AluisioASG has joined #nixos-chat
<lovesegfault>
etu: Ah, you're up :D
<lovesegfault>
I'm using it right now and it seems to work!
<lovesegfault>
I've learned you can't make zsh_history a bind mount b/c zsh tried to clobber it by moving zsh_history.new into zsh_history
<lovesegfault>
So I just changed HISTFILE to be $XDG_DATA_DIR/zsh/history and I bind mount the full zsh dir
<lovesegfault>
did the same for bash
<lovesegfault>
gchristensen: You were right, doing this whole thing gives the computer an incredible "new computer smell"
<JJJollyjim>
Lmao
<etu>
Yeah, it's nice to reboot and know that there's no garbage state that you didn't want it expected :)
<etu>
s/it e/or e/
<lovesegfault>
Yeah, it's super nice
cole-h has quit [Quit: Goodbye]
<sphalerite>
Why am I considering buying an LTO drive :|
<samueldr>
oops, just cleaned for ~55GiB of left-over gc roots I don't really need or want
<sphalerite>
what's oops about that?
<samueldr>
oops I was wasting 55GiB
<samueldr>
out of ~225 for the drive with the store
waleee-cl has quit [Quit: Connection closed for inactivity]
<lovesegfault>
sphalerite: What's an LTO?
<ar>
lovesegfault: tape
<lovesegfault>
OH
<lovesegfault>
those like hundreds of TB in a single tape things?
<ashkitten>
ugh, my cosmo hasn't been locked yet but planet's last announcement would seem to state that all contributions should be?
<samueldr>
ashkitten: get in touch with them right now
<ashkitten>
already sent them an email yesterday, they havent responded
<sphalerite>
locked?
<ashkitten>
i dont have enough of a brain
<sphalerite>
what?
<sphalerite>
lovesegfault: you planning on getting one now? :p
<lovesegfault>
sphalerite: You tempted me
<lovesegfault>
Whether or not I buy something is an eternal struggle between what people here tempt me with and how ridiculous my wife will think I am if I actually buy it
<sphalerite>
ahaha
<sphalerite>
I don't think I'd recommend actually getting a tape drive
<sphalerite>
afaik you also need to store the tape in controlled conditions for the data to last
<lovesegfault>
I could get a fridge
<ashkitten>
samueldr: is it reasonable to email them additionally to hello@planetcom.co.uk despite having already sent a message on indiogogo
<samueldr>
not unreasonable, that's for sure
<ashkitten>
ok
<sphalerite>
what is this announcement?
<JJJollyjim>
Anyone seen anything about running nixos on a Planet device btw? Would love to have something other than a hacked-together old debian on my Gemini
<JJJollyjim>
Though tbh the battery life is so much better on android, it might not be worth it :/
<ashkitten>
planning to port to the cosmo when i get it
<ashkitten>
really annoyed rn with planet not having locked my contribution
<JJJollyjim>
Awesome
<JJJollyjim>
Yeah that's annoying :/
<ashkitten>
idk if theyll reply to my email over the weekend
<ashkitten>
but at least they have that in 2 places now
<ashkitten>
gonna be really annoyed if the damn thing doesnt arrive before i move
<lovesegfault>
What did y'all get from indie gogo?
FRidh has joined #nixos-chat
<lovesegfault>
talyz: o/
<talyz>
lovesegfault: \o
<talyz>
lovesegfault: to reproduce the issue all you have to do is move the real file out of the way and reboot
<talyz>
lovesegfault: that creates an empty file in place of the real one, as expected, but systemd doesn't like that
<lovesegfault>
talyz: Oh, I see
<talyz>
lovesegfault: and I would expect that behaviour to be quite common, which means file bind mounts aren't going to work very well with uninitialised state :/
<lovesegfault>
Interesting
<lovesegfault>
I wonder if the right behavior would be to issue an error during activation saying the source doesn't exist?
<lovesegfault>
this then allows the file to be created by $whatever and the user can move it to the state dir
<lovesegfault>
Another alternative is to fallback to symlinking, but not sure how that could be impl'd
<talyz>
I think the solution, sadly, is sticking with symlinking, since the works fine in the majority of cases for files. Bind mounting is still viable and makes a great difference to directories.
<lovesegfault>
You might be right :/
<adisbladis>
JJJollyjim: I've been running NixOS on a gemini
<adisbladis>
It's still a hella lot of work to be done before it's usable though
<JJJollyjim>
oh awesome!
<JJJollyjim>
usable like the debian image, or actually usable :P
<adisbladis>
No UI running (needs fully working libhybris)
<adisbladis>
Basically, it boots up and you can use the serial console
<JJJollyjim>
ah yeah
<JJJollyjim>
is that with a mainline kernel or the provided 3.18?
<adisbladis>
3.18
<JJJollyjim>
right
<JJJollyjim>
i think you can get about that far on mainline too
<JJJollyjim>
but no further :P
<adisbladis>
Ahh, that wasn't the case when I was hacking on this
<eyJhb>
`How Crash Bandicoot Hacked The Original Playstation | War Stories | Ars Technica`
<eyJhb>
Update {^_^} to inclue links
<eyJhb>
youtube*
<__monty__>
Oh, yeah, pretty cool how they managed to pack that onto a single disc.
<eyJhb>
Not much the disk, more the memory limitations :D But prop also the disc after they figured this out
viric has joined #nixos-chat
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos-chat
<eyJhb>
I love theese war stories WAAAY too much
<eyJhb>
Same with all the hacking videos of the old consoles
<viric>
Yesterday I knew about bitmessage.
<viric>
so many years and only yesterday I knew it
<__monty__>
Heh, I read about it yesterday too.
<__monty__>
Not the most efficient distribution strategy.
bqv has quit [Quit: WeeChat 2.8]
<viric>
bah, it's not for video
<eyJhb>
I have used it before, but forgot about it
<viric>
anonymising and hiding metadata is never efficient
<eyJhb>
Also, marked broken...
<__monty__>
It's more an email replacement than anything else.
<viric>
email or not, it depends on the threshold of PoW
<__monty__>
Had a compromising bug in a recent version.
<eyJhb>
`A remote code execution vulnerability has been spotted in use against some users running PyBitmessage v0.6.2.`
<eyJhb>
GG
<__monty__>
viric: Yeah, to combat spam.
<viric>
in use - great. Practical results.
<eyJhb>
How the F do you get RCE in Python?
<eyJhb>
You have to write some really bad code for that
<viric>
haha
<gchristensen>
is there a trivial CLI program to spawn a gui prompt with two buttons: "yes please" and "nah", and then the caller knows what was picked?
<viric>
It has some 'crypto' in C I guess
<gchristensen>
eyJhb: it is super easy!
<gchristensen>
`pickle.load`
<eyJhb>
Like, I have done it on purpose, but if you read just a LITTLE when you use subprocess etc. you won't
<gchristensen>
viric it calls arbitrary functions and passes arbitrary data
<viric>
eval. :D
<viric>
eval is better than that
<eyJhb>
That is just stupid
<viric>
well, more than practicity, I liked the broadcast mechanism
<viric>
Tox lacks the public channels thing
<__monty__>
It does?
<viric>
or BBS, if want want
<__monty__>
The problem with Bitmessage's broadcast is there's no way to restrict who's listening.
<viric>
crypto should make it irrelevant
<__monty__>
Subscriber anonymity cuts both ways.
<viric>
__monty__: what is that?
<__monty__>
Does Tox have channels much like IRC does? Why doesn't that count as broadcast?
<viric>
tox doesn't, does it?
<__monty__>
Pretty sure it does.
<viric>
ah I thought it only had group chats with invitations
<__monty__>
Might be an auto-invite thing. But I'm pretty sure their dev channel is hosted on the tox network.
<viric>
Ah, maybe they have bots that accept all friends and invite people there on request?
<viric>
hm I read "New DHT-based groupchats"
<__monty__>
viric: Afaiui bitmessage's broadcast works by distributing an address. Anyone with that address can decrypt the messages. That's great if you're a company and you want to allow people to subscribe to a newsletter anonymously. It's extremely error-prone if you want to share things among a group of friends. As soon as the address travels over an insecure channel it could be compromised and there's no
<__monty__>
way of detecting that.
<viric>
bitmessage doesn't have any PFS or anything similar either.
<eyJhb>
This also meaaans, holy hell, that if you have logged all messages I guess you can now decrypt them because of the RCE `Bitmessage developer Peter Šurda's Bitmessage addresses are to be considered compromised.`
<eyJhb>
Someone should make a pimessage
<eyJhb>
:D
<viric>
broadcast and no PFS, that's bad
<__monty__>
viric: Probably lacks repudiation as well?
<viric>
well, once the original message can be decrypted & authenticated at any future time...
<__monty__>
All of this and it still manages to be kind of better than PGP >.<
<viric>
do you think tox group chats can do deniabilitY?
<viric>
often in tox all goes about unknown people not able to link private key with dht key
<viric>
but friends know the link
<manveru>
is there an easy wrapper for applying resource limits and sandboxing to a single executable?
<gchristensen>
is systemd-run a thing that could work?
<manveru>
hm, possibly
<manveru>
would be nice to have it independent of systemd though :)
<gchristensen>
maybe firejail can do it
<viric>
manveru: nsjail?
<manveru>
that sounds cool, yeah :)
<manveru>
i'm just writing a wrapper for hashicorp nomad
<manveru>
their normal sandboxing relies on predefined chroot paths in the worker... and they hardlink them all, can't really do that with the nix store
<manveru>
well, you can, but only if you have infinite time :P
<manveru>
now i just have to figure out what's different between nsjail and firefail
<viric>
manveru: doesn't a bind mount work in chroot?
<manveru>
i think because you cannot do bind mounts on macos or something
<__monty__>
viric: Oh, based on a clique though so I doubt it scales.
<viric>
__monty__: that's all end-to-end, isn't it? What is a clique?
<viric>
ah yes, that, a clique.
<manveru>
was just gonna use chpst, but i kinda like the idea of limiting CPU and bandwidth too :)
<viric>
__monty__: for what I remember from OTR between A and B, deniability meant that only B tell that a message comes from A because B knows they didn't write it. But an observer cannot tell whether it comes from A or B.
<viric>
is that repudiability?
<viric>
(I'm reading about it)
<viric>
so, in short, I thought that axolotl and all those group chat things may be implemented *over bitmessage* flood communication.
<viric>
but maybe freenet would be a more clever option than bitmessage, for transport
<viric>
for that kind of transport
<__monty__>
Repudiability is the ability to come out and say you didn't write a mesasge.
<viric>
and who would believe that? an outsider to the chat or an insider?
<viric>
I'm trying to look for high-latency anonymizing networks/overlays and all that comes out is "low latency" :)
<viric>
(attempts)
<sphalerite>
manveru: don't bother with firejail, last I checked it was a hot mess
<sphalerite>
manveru: idk about nsjail though
<manveru>
yeah, already found that out :)
<__monty__>
They could be the same thing. I'm familiar with the repudiation terminology, not deniability so I can't really say they're exactly the same thing.
<sphalerite>
__monty__: I think what you're referring to is usually called deniability
<manveru>
i really didn't feel like making an nsjail config generator right now, so going with systemd-run and systemd-nspawn instead... i assume they mostly do the same under the hood anyway and i can replace it later
<sphalerite>
eh nah I should just shut up I don't _really_ know what I'm talking about when it comes to cryptography
<__monty__>
My crypto classes used the repudiation terminology so that's what I'm comfortable using.
<eyJhb>
manveru, sphalerite Michael something something uses nsjail
neeasade has joined #nixos-chat
<sphalerite>
eyJhb: Michael something something?
<pie_>
viric: idk the details, there was pond but idk what happened to that
<eyJhb>
sphalerite: cannot remember the nick :p
<eyJhb>
`MichaelRaskin`
<gchristensen>
can nix-store create a gc root for a .drv?
<sphalerite>
gchristensen: I don't think so, but you can create it manually or with nix-instantiate?
<__monty__>
pie_: Didn't pond suffer from debilitating DHT vulnerabilities?
<__monty__>
viric: Have you looked at GNUnet?
<gchristensen>
sphalerite: I'm creating one with nix-instantiate, and then I want to create another to the same drv with a new name without evaluating again
<gchristensen>
I guess that means I should create it by hand
<eyJhb>
lovesegfault: my teacher is annoying regarding this
<eyJhb>
Her solutions are just matlab
<eyJhb>
Without comments or any explanation
cole-h has joined #nixos-chat
<pie_>
11/10
waleee-cl has quit [Quit: Connection closed for inactivity]
waleee-cl has joined #nixos-chat
Jackneill has quit [Ping timeout: 260 seconds]
<eyJhb>
Hate it... Use python god damn it
<eyJhb>
I haven't found any reason to use Matlab in a very long time
<viric>
__monty__: I read that textsecure also uses the clique for the update of ratchet keys
<eyJhb>
I love how if a want to play the next song, and use my media keys, some random tab in chromium will start the next video, but I CAN go to the previous song without any issues
<eyJhb>
Also, play/pause are random. Spotify it times, youtube other times
<{^_^}>
swaywm/sway#5430 (by grahamc, 5 hours ago, open): swaynag: allow specifying more buttons which execute and dismiss
<DigitalKiwi>
according to the tweets; CVEs
<eyJhb>
Spotted two RCEs. which is impressive!
<eyJhb>
\s
<gchristensen>
lmao
<DigitalKiwi>
there was only one line of code!
<eyJhb>
I want to find SOMETHING wrong now, but there is simply too little code
<eyJhb>
More gchristensen :p
<eyJhb>
Throw in some eval
parsley9366 has joined #nixos-chat
<iqubic>
I have question that might be polarizing: My mother has a an old IPhone 8 she's no longer using that she wants to give to me, for free. If I take this, I'd be switching from an old Alcatel One Touch Evolve Android phone to an IPhone 8. Should I take the IPhone?
<gchristensen>
I don't write C if I can avoid it
<gchristensen>
iqubic: if you like it, sure :)
<eyJhb>
gchristensen: join the CVE side!
<iqubic>
But don't you worry about IPhones not being open source?
<eyJhb>
Only if you worry about it iqubic
<DigitalKiwi>
the only android i like more than iphones are my blackberry keyone and that's not because of android that's because of the keyboard
<gchristensen>
eyJhb: no thanks
<gchristensen>
eyJhb: I already had to take a shower after that patch
parsley936 has quit [Ping timeout: 256 seconds]
<DigitalKiwi>
and the headphone jack...
<samueldr>
iqubic: you do you
<samueldr>
iqubic: your alcatel phone is "not open source" if you haven't actually gone and verified you can have the sources
<samueldr>
and alcatel, in my experience, sometimes don't provide the sources as legally required
<eyJhb>
gchristensen: I have a PR in some IPMI tool
<samueldr>
then, that's only about the open source ideals
<samueldr>
think about security and security updates
<samueldr>
did your old one touch even get updates? when was it the last time?
<samueldr>
compare with the updates cycle and security of iphones, and I mean it, *you* compare those, and make up your mind
<DigitalKiwi>
my keyone is stuck on 7 :(
<samueldr>
any decisions, make them with knowledge and understanding of your own
<samueldr>
DigitalKiwi: I need to verify, I'm not sure the keyone was or wasn't made by the same OEM as alcatel
<samueldr>
the later android blackberry devices were made by TCL
<samueldr>
yep
<samueldr>
TCL
<samueldr>
though AFAIUI with the software from blackberry, including their secure (AFAIK) boot
<risson>
DigitalKiwi: about the NixOS artwork you did for gchristensen, are you willing to do something as spectacular as this for me?
<samueldr>
quick search confirms it
<DigitalKiwi>
risson: quite possibly! maybe even more better lol
<DigitalKiwi>
now i have *experience*
<risson>
DigitalWiki++
<samueldr>
fruit, not quick
<risson>
Man I don't know how to type
<samueldr>
:)
<risson>
DigitalKiwi++
<{^_^}>
DigitalKiwi's karma got increased to 11
<DigitalKiwi>
i don't quite have an encylopedia of knowledge
<risson>
Well, let me know if you're interested for another project :D
<samueldr>
>> Cunningham came up with the name WikiWikiWeb because he remembered a Honolulu International Airport counter employee who told him to take the Wiki Wiki Shuttle, a shuttle bus line that runs between the airport's terminals
<iqubic>
I'm just glad that the custom keyboard I use (MessageEase) will still work on Apple.
<gchristensen>
iqubic: messageease looks cool
<ldlework>
Does anyone want a unit-test framework for Nix?
<gchristensen>
I made a tiny one, once. it'd be cool to have one for real
<ldlework>
Did you publish it?
<gchristensen>
hrm. maybe as a gist? I don't even re
<gchristensen>
member what project it was in
<ldlework>
One problem I've run into is that when my test files go to import the local code they want to test, I get things like: error: getting status of '/nix/store/src/options.nix': No such file or directory
<ldlework>
I'm not sure how I'm finding my way into the store lol
<sphalerite>
gchristensen: ooooooooooh. lol
<pie_>
<ldlework> Does anyone want a unit-test framework for Nix?
<pie_>
ldlework: yes i even started making a bad one
<ldlework>
I created #nixos-nixt
<ldlework>
come help
<pie_>
well gee sounds like people keep reinventing this
<ldlework>
unless you wanna publish yours
<ldlework>
i'd rather just use something
<pie_>
i dont think a serious implementation can exist until we have tryEval and better notions of equality
<pie_>
but what do i know
<pie_>
tryeval is currently useless
<pie_>
well * pure nix
<ldlework>
I just need something to help with the development of Styx
<pie_>
whatever gchristensen has is probably better
<energizer>
i have a /root/.ssh/id_rsa that's encrypted with a passphrase, but i can't remember what i was thinking when i created it or what it could be useful for, since nearly everything that would use a root ssh key is non-interactive. i think i should delete it and recreate one that's not passphrase-encrypted. am i missing something?
<cransom>
you can remove passphrases.
<energizer>
ah that's a good tip
<energizer>
is there some reason to have a root key that's passphrased?
<samueldr>
root is just another user :)
<samueldr>
it would be weird to enforce or suggest not passphrasing on an arbitrary user
<cransom>
the keys that i spread around the world,they have pass phrases attached. if it's an automated process, there's no passphrase but it's also only for the machines that need it.
<cransom>
meaning, the public keys i send out that i, as a human, have a local private passphrase.
<energizer>
samueldr: the nix daemon can't unlock it
<samueldr>
sure, and you're right that passphrases on keys for automation is not useful
<samueldr>
I don't know enough about ssh, ssh agents and nix, but I guess there must be a way to make it ask
<energizer>
related: i'm thinking about setting up a CA for my personal computers, wondering how much of a hassle it is and if it's worthwhile.
<energizer>
to replace authorized_keys
<energizer>
any opinions?
<__monty__>
Think gchristensen has such a setup.
<ldlework>
A really cool thing would be to be able to create simple CLI applications with Nix
<energizer>
ldlework: what is the advantage of doing it in nix vs a general-purpose programming languaeg plus nix for the packaging?
<ldlework>
like, my nixt unit testing thing will be mostly nix
<ldlework>
it would be cool do have some kind of way to declare the cli app to go with it
<ldlework>
what commands it has, their args, a little bash snippet or something for the body, i dunno probably a stupid idea
<pie_>
ldlework: idk ask profpatsch
<pie_>
ldlework: does my thing look like its going to help any
<energizer>
i guess if you're gonna write software in nix, you need a test runner, and that test runner might be best written in nix?
<ldlework>
energizer: yeah and then with a cli tool ontop
<pie_>
because the output it takes actually gets evaluated as nix code
<ldlework>
use it for what?
<pie_>
idk, the cli stuff
<pie_>
i didnt really think through this
<ldlework>
i was thinking along the lines of
<ldlework>
a nixlang api where you can describe a cli interface
<ldlework>
and then it would generate some bash
<energizer>
what's with languages having unsafe functions built in and with short names
<ldlework>
but like, it would be annoying to then have to switch to bash for the function implementations or whatever
<joepie91>
random idea: a "Things you can do with NixOS" blog post, that just shows a bunch of neat things you can do under NixOS, that would appear completely impossible / pie-in-the-sky to someone not familiar with Nix... like wrapping some part of your system in a container by moving a few lines, or having containers with a shared Nix store, or rolling back on boot
<pie_>
energizer: well technically its not exposed by default, thats what the --option is for
<pie_>
joepie91: that would be cool
<pie_>
though ive never done the whole trust fall thing
<ldlework>
if only Nix were Lisp and I could have a chance at converting some basic forms to bash
<joepie91>
energizer: I've had to spend a not-insignificant amount of my remaining lifespan trying to kill the proposal in Node.js to call the unsafe, uninitialized-memory-leaking Buffer allocation method `.fastAlloc` or such
<ldlework>
Heh what if cli subcommands were actually pure nix
<joepie91>
energizer: it is now called `.unsafeAlloc`
<pie_>
nix has so many weird hacks i wish the BDFL would acknowledge it as a proper language
<pie_>
and stop neglecting the interpreter
<energizer>
joepie91++
<{^_^}>
joepie91's karma got increased to 14
<ldlework>
and the actual wrapper just ran nix repl or something
<joepie91>
as it turns out, a SIGNIFICANT chunk of people do not even recognize that naming matters where safety is concerned
<joepie91>
and will downplay it when told this
<pie_>
joepie91: "but its fast!"
<pie_>
&thred
<ldlework>
I guess an actual Nix backed CLI wouldn't be able to interact with anything
<ldlework>
Besides with activation scripts then you're back to bash
<ldlework>
damn!
<joepie91>
energizer: unfortunately the `child_process` module is from before my time, and so it has a `.exec` that is vulnerable to shell injection and an `.execFile` that is not...
<pie_>
joepie91: id jsut say dont take those people seriously but thats not how any of this works ;_;
<joepie91>
these damn things are everywhere
<joepie91>
pie_: unfortunately they are all too often the people working on language cores
<ldlework>
infinisil: could this read/write directly from the current working directory instead of going through the store, etc?
<joepie91>
and once it's in core, it ain't never coming out again
<ldlework>
the problem with nix cli stuff is you can't do any real work
<ldlework>
i wish i was smart enough to understand parsing and compilation and all that and make a toy language
<pie_>
ldlework: nix only puts stuff in the store afaict
<joepie91>
pie_: re: trust fall, I have repeatedly rescued my system via GRUB rollback :P
<pie_>
you only get symlinks on the outside
<energizer>
"20.09pre226148.0f5ce2fac0c (Nightingale)" what is "226148"?
<joepie91>
including a remote server once
<pie_>
joepie91: that was an overly veiled joke at pie in the sky :p
<infinisil>
ldlework: Nix can already read directories without going through the store
<joepie91>
lol
<ldlework>
Yeah but you can't do any work
<joepie91>
right :P
<joepie91>
ldlework: parsing is surprisingly easy
<joepie91>
compilation is trickier, especially optimization-wise
<__monty__>
ldlework: Since you like lisp just read the wizard book.
<joepie91>
you can pretty much think of parsing as a while(true) loop that consumes byte by byte and updates internal state based on a combination of the currently-parsed byte + previous state
<infinisil>
ldlework: The nix-rts thing can do stuff by calling builtins.exec :P
<ldlework>
When can I train a transformer model on a few examples and get what I want.
<joepie91>
depending on the parsing model it may backtrack, keep a stack of stuff, etc.
jtojnar has quit [*.net *.split]
CRTified[m] has quit [*.net *.split]
CRTified[m] has joined #nixos-chat
jtojnar has joined #nixos-chat
zimbatm has quit [*.net *.split]
Taneb has quit [*.net *.split]
obadz has quit [*.net *.split]
colemickens has quit [*.net *.split]
ldlework has quit [*.net *.split]
johanot has quit [*.net *.split]
savanni has quit [*.net *.split]
raboof has quit [*.net *.split]
davidtwco has quit [*.net *.split]
liszt has quit [*.net *.split]
emilazy has quit [*.net *.split]
Taneb has joined #nixos-chat
zimbatm has joined #nixos-chat
savanni has joined #nixos-chat
raboof has joined #nixos-chat
ldlework has joined #nixos-chat
davidtwco has joined #nixos-chat
johanot has joined #nixos-chat
obadz has joined #nixos-chat
emilazy has joined #nixos-chat
liszt has joined #nixos-chat
mutantmell has quit [*.net *.split]
lovesegfault has quit [*.net *.split]
jD91mZM2 has quit [*.net *.split]
infinisil has quit [*.net *.split]
bridge[evilred] has quit [*.net *.split]
clever has quit [*.net *.split]
makefu has quit [*.net *.split]
lukegb has quit [*.net *.split]
aszlig has quit [*.net *.split]
clever has joined #nixos-chat
bridge[evilred] has joined #nixos-chat
infinisil has joined #nixos-chat
jD91mZM2 has joined #nixos-chat
lovesegfault has joined #nixos-chat
mutantmell has joined #nixos-chat
aszlig has joined #nixos-chat
makefu has joined #nixos-chat
lukegb has joined #nixos-chat
<__monty__>
I'm not so sure I agree parsing's easy. People use parser compilers for a reason : )
ky0ko has quit [*.net *.split]
leonardp has quit [*.net *.split]
hexa- has quit [*.net *.split]
thefloweringash has quit [*.net *.split]
thefloweringash has joined #nixos-chat
hexa- has joined #nixos-chat
leonardp has joined #nixos-chat
ky0ko has joined #nixos-chat
hexa- has quit [Max SendQ exceeded]
Shados has quit [*.net *.split]
balsoft has quit [*.net *.split]
tokudan[m] has quit [*.net *.split]
danielrf[m] has quit [*.net *.split]
leons has quit [*.net *.split]
emily has quit [*.net *.split]
aranea has quit [*.net *.split]
JJJollyjim has quit [Ping timeout: 244 seconds]
metheflea has quit [Ping timeout: 244 seconds]
davidtwco has quit [Ping timeout: 246 seconds]
CRTified[m] has quit [Ping timeout: 244 seconds]
* ldlework
puts his hands in the air
leonardp has quit [Ping timeout: 252 seconds]
thefloweringash has quit [Ping timeout: 252 seconds]
<ldlework>
Wheeeeee!
liszt has quit [Ping timeout: 246 seconds]
Irenes[m] has quit [Ping timeout: 260 seconds]
jtojnar has quit [Ping timeout: 244 seconds]
pistache has quit [*.net *.split]
monsieurp has quit [*.net *.split]
ajs124 has quit [*.net *.split]
qyliss has quit [*.net *.split]
leah2 has quit [*.net *.split]
sphalerite has quit [*.net *.split]
talyz has quit [*.net *.split]
leah2 has joined #nixos-chat
ajs124 has joined #nixos-chat
qyliss has joined #nixos-chat
monsieurp has joined #nixos-chat
pistache has joined #nixos-chat
talyz has joined #nixos-chat
sphalerite has joined #nixos-chat
leah2 has quit [Max SendQ exceeded]
Shados has joined #nixos-chat
eyJhb has quit [*.net *.split]
betawaffle has quit [*.net *.split]
mog has quit [*.net *.split]
nckx has quit [*.net *.split]
aleph- has quit [*.net *.split]
genevino has quit [*.net *.split]
liszt has joined #nixos-chat
aleph- has joined #nixos-chat
mog has joined #nixos-chat
nckx has joined #nixos-chat
betawaffle has joined #nixos-chat
eyJhb has joined #nixos-chat
aranea has joined #nixos-chat
yegortimoshenko has quit [Ping timeout: 244 seconds]
nocent has quit [Ping timeout: 244 seconds]
aterius has quit [Ping timeout: 244 seconds]
davidtwco has joined #nixos-chat
puzzlewolf has quit [Ping timeout: 244 seconds]
aanderse has quit [Ping timeout: 244 seconds]
vaibhavsagar has quit [Ping timeout: 244 seconds]
betawaffle has quit [Max SendQ exceeded]
raboof has quit [Ping timeout: 246 seconds]
liszt has quit [Changing host]
davidtwco has joined #nixos-chat
liszt has joined #nixos-chat
davidtwco has quit [Changing host]
genevino has joined #nixos-chat
hexa- has joined #nixos-chat
ohhaimark[m] has quit [Ping timeout: 260 seconds]
arcnmx has quit [Ping timeout: 260 seconds]
raboof has joined #nixos-chat
betawaffle has joined #nixos-chat
worldofpeace has quit [Ping timeout: 260 seconds]
rycee has quit [Ping timeout: 260 seconds]
leah2 has joined #nixos-chat
<samueldr>
ah, I have join/parts hidden, so I didn't really understand until I read the global notice
<ldlework>
hehe
yegortimoshenko has joined #nixos-chat
puzzlewolf has joined #nixos-chat
aanderse has joined #nixos-chat
ohhaimark[m] has joined #nixos-chat
arcnmx has joined #nixos-chat
worldofpeace has joined #nixos-chat
rycee has joined #nixos-chat
__monty__ has quit [Quit: leaving]
CRTified[m] has joined #nixos-chat
colemickens has joined #nixos-chat
aterius has joined #nixos-chat
leons has joined #nixos-chat
leonardp has joined #nixos-chat
JJJollyjim has joined #nixos-chat
vaibhavsagar has joined #nixos-chat
thefloweringash has joined #nixos-chat
danielrf[m] has joined #nixos-chat
tokudan[m] has joined #nixos-chat
emily has joined #nixos-chat
metheflea has joined #nixos-chat
Irenes[m] has joined #nixos-chat
jtojnar has joined #nixos-chat
balsoft has joined #nixos-chat
nocent has joined #nixos-chat
<energizer>
`--builders ssh://foo` says "Copying /nix/store/... from ..." but when i look at my system bandwidth i see 18Kb/s Rx. if i manually pipe data over ssh from that host it's in the MB/s. what could be going wrong?
<samueldr>
something better suited to the on-topic channels of nix/nixos
<energizer>
ok
parsley9366 has quit [Remote host closed the connection]