<colemickens>
I wonder if anyone has tried closed-source patreon releases for OSS software. Write a feature, release it to Patreon only, add it to the open source release after N weeks, or after some donation threshold was met.
<samueldr>
it gets hard when you mix in contributions from others AFAIUI
<ldlework>
colemickens: I know some people who are doing that for D&D modules
<samueldr>
because the license applies to their copyright ownership
<ldlework>
patreon's get exclusive access for a couple months, then everyone gets it for free
<samueldr>
so unless you have a CLA, which is not great for OSS, it's harder to do *outright* closed-source
<samueldr>
(or reject any external contribution)
<colemickens>
ldlework: nice, I hope that is going well for them; I haven't really come across that model in any of the things I patron so far. it's all "exclusive extra" content so far.
<colemickens>
samueldr: hmm
<colemickens>
samueldr: I guess maybe all CLAs bottom out on this, but it'd basically be a CLA that says "you have to give me exclusive non-OSS license for any code contributions" which feels... well, I'd be turned off.
<samueldr>
yeah
<samueldr>
wel
<samueldr>
well*
<samueldr>
CLA doesn't necessarily mean that
<samueldr>
but generally you end up handing-off control of the code to that other party
<samueldr>
which generally can include them just pivoting-it-off into closed-source outright
<samueldr>
depending on the license though
<samueldr>
you can do it
<samueldr>
and even not depending on the license
<samueldr>
you can do it in spirit
<samueldr>
all those that allow distribution without distributing sources (e.g. MIT) you can just do it
<samueldr>
and GPL, you could do it, and distribute the source to those who pay
<samueldr>
hopefully no one leaks it
<samueldr>
but if they do, it's what the license is there for
<samueldr>
"hopefully", as in "hopefully for your support scheme"
<samueldr>
though ideally you're doing it not relying on secrecy, but relying on people being good
kalbasit_ has quit [Remote host closed the connection]
danderson has quit [Remote host closed the connection]
danderson has joined #nixos-chat
<lovesegfault>
Huh
<lovesegfault>
God, ini is cursed
Guest88372 has quit [Read error: Connection reset by peer]
jared-w has quit [Ping timeout: 260 seconds]
sorear has quit [Ping timeout: 260 seconds]
Guest88372 has joined #nixos-chat
sorear has joined #nixos-chat
jared-w has joined #nixos-chat
<ldlework>
argh, after figuring out this jack stuff I sort of feel a mortal obligation to write it all down
waleee-cl has quit [Quit: Connection closed for inactivity]
danderson has quit [Remote host closed the connection]
<clever>
, i said!
<clever>
,
<{^_^}>
Special commands: find tell locate expand inclusive-language random-pr - Commands sorted by use count, page 0 (use ,<n> to view page <n>): help libraries ping channels paste escape'' -a escape" library overlay ifd pinning profiling botsnack unfree ask which-channel xml xy fancy-uninstall imperative wololo nixlang++ cache pills pointers tofu home-manager invite runtimedeps stateversion thesis commands exec matrixbridge smart-questions tias
<{^_^}>
Invalid command syntax
<clever>
?
<clever>
there it goes
danderson has joined #nixos-chat
<ldlework>
woahhhh, i just figured out how to dynamically connect hardware devices to jack at runtime O_O
<ldlework>
I wonder what the general mechanism is in linux to respond to USB devices being plugged in and what not.
<ldlework>
I wonder if I could write a thing such that when you plugin a USB mic, it automatically adds a new source/sink to jack, and then wires them up to the pulseaudio fakes.
<ldlework>
then....
<ldlework>
/you could like just plugin a device and have it operational/
<ldlework>
Something is super screwy with recording audio
dadada_ has quit [Ping timeout: 264 seconds]
__monty__ has joined #nixos-chat
<gchristensen>
doing io intensive things like "reading the disk" on AWS is reminding me that bare metal is pretty sweet and the cloud isn't quiiiite so good as I remembered
dadada_ has joined #nixos-chat
<supersandro2000>
you mean someone others computer?
<gchristensen>
I don't buy that perspective. the cloud is super good for a lot of things
<gchristensen>
but doing traditional linuxey tasks that you maybe wouldn't normally bother doing in the cloud... not great!
<joepie91>
gchristensen: depends a lot on the provider. the big cloudycloud providers just have bad perf :)
<gchristensen>
yeah
<joepie91>
with Azure topping the list in terms of utterly unusable disk I/O, AFAIK
<gchristensen>
they have nice prices, apis, geodistribution, and services though :P (even when I fundamentally disagree with their business)
<gchristensen>
yesterday I tried to `dd` one 2T drive to another, and it was going to take 4 days
<joepie91>
can't agree on the prices, geodistribution is okay but not meaningfully different from 'a handful of reputable VPS providers' - the only real selling point there IMO is the API
<gchristensen>
and one vendor
<joepie91>
that's the thing that other providers are usually behind on
<joepie91>
eh, I don't consider "one vendor" a feature, personally :P
<joepie91>
baskets are bad
<gchristensen>
depnsd on who you are :)
<gchristensen>
not to mention their security posture, which basically no other provider can match
<gchristensen>
anyway this morning is brought to you by `dd` running at 7.4MB/s, reading from an EBS volume and writing to /dev/null
<joepie91>
lol
<joepie91>
nice
<joepie91>
'compiling' xkcd goes here, but dd'ing
<Ashy>
don't forget to crank the ebs volume type and iops up
<Ashy>
io2 goes up to 64k iops on a single volume
<gchristensen>
it took the machine like 9 hours to boot
<Ashy>
you can change it on the fly without rebooting
<gchristensen>
Ashy: it is snapshot based, so not sure even going to io2 w/ 64k iops will improve it in the short term
<Ashy>
as always test on some other instance first of course
<Ashy>
ah yeap
<eyJhb>
gchristensen: 7 MB/s! That is faster than like, less than 7 MB/s.
<eyJhb>
I actually see that often on some of my systems and get mad :p
<eyJhb>
But mostly when I think it should be 100 MB/s (home file transfer), but I forget my default ssh config will go waaay out to connect to my server next to me
<gchristensen>
5400rpm rust can hit 250MB/s, how are you hitting 7? :) a hard disk over USB 2.0 would be faster at 60MB/s
<joepie91>
250? huh? afaik 100-130MB/sec is typical sequential for 5400rpm
<__monty__>
: (, stuck on USB2 and 100 Mb networking.
<eyJhb>
gchristensen: Sometimes I get slow speeds with USB 2, but it is mostly network. Soo HOME -> Luxembourg -> OpenVPN to home
<gchristensen>
joepie91: ah yeah whoops
<aminechikhaoui>
gchristensen did you just say nice things about Azure ? :o
<__monty__>
I *think* he was talking about AWS?
<gchristensen>
I'm talking about AWS this whole time
<aminechikhaoui>
ah lol
<aminechikhaoui>
I'm just trolling
<gchristensen>
speaking of which, our AMI uses an MBR partition table so it is impossible to make /nix/store bigger than 2T without surgery
<aminechikhaoui>
yeah I got bitten by that one time
<gchristensen>
(:
<gchristensen>
maybe we can sneak that fix in to the 20.09 AMIs
<eyJhb>
I am having a harder and harder time, to realise that 2000 is 20 years ago, and that people form 2000 are not 10 years old... Like movies from 2005, etc. seems like fairly new...
pie_ has quit [Quit: pie_]
pie_ has joined #nixos-chat
<eyJhb>
Any Makefile experts regarding for C++ projects?
<eyJhb>
I have a Makefile that I would just want some eyes on real quick, before I add it to the wiki
<drakonis>
i'd buy it just to escape my nvidia hell
<cole-h>
^^^^^^^^^^^^^^^^^^^^^^^
<joepie91>
I've refused to buy nvidia cards for years now specifically because of the driver crap
<cole-h>
I bought my 1080 before I was on Linux x)
<joepie91>
and I honestly can't think of anything to complain about in amdgpu
<drakonis>
same
<drakonis>
i bought my dinky ass 960 before linux
<joepie91>
it just... works, no bullshit, no crashes, no weird errors
<drakonis>
i'm still paying for it
<joepie91>
no mysterious card resets
<drakonis>
ah yes
<joepie91>
UNLIKE A CERTAIN OTHER BRAND
<drakonis>
i hate that i cant hibernate my machine...
<cole-h>
👀
<drakonis>
because the driver wont recover
<joepie91>
that appears to work fine for amdgpu also
<cole-h>
I'm probably going to keep using my 1080 + a 5800 in tandem until all the bugs are ironed out, however
<drakonis>
hmm
<cole-h>
see: Navi reset bug
<drakonis>
i see people caring too much about perf
<drakonis>
i see not having to worry about the horror show that is dealing with nvidia
<drakonis>
plus reasonably competitive perf
<cole-h>
Well, when you're playing video games, you do want the best performance...
<drakonis>
i dont even play anything that needs all this performance
<joepie91>
eh
<joepie91>
disagree
<joepie91>
I want my games to actually, y'know, run
<joepie91>
and with that I mean the game I selected to play, not the "what's broken with the drivers this week" game
julm has quit [Ping timeout: 265 seconds]
<drakonis>
yes
<joepie91>
performance is secondary to me :P
<cole-h>
Fair enough. But in Windows, I never had any issues with Nvidia. Considering that's where the majority of "gamers" are... seems about right that they'd go for Nvidia
<cole-h>
gamers -> desktop gamers
<drakonis>
lol
<drakonis>
yes
<joepie91>
back when I still ran Windows, which was admittedly quite some time ago, nvidia was a shitshow there too
julm has joined #nixos-chat
<drakonis>
its still kind of bad there
<cole-h>
Really? Huh.
<drakonis>
in some minor ways of course
* cole-h
runs a win10 VM and hasn't had any issues
<drakonis>
i've had some issues with cursors on full screen with flux and night light
<cole-h>
tbh, the only issue I ever had with Nvidia was how the control panel takes 10 million years just to change tabs
<drakonis>
the cursor would disappear or not be affected by the tint
<cole-h>
OK, I don't think that's Nvidia related since there's a registry tweak to fix the "unaffected by tint"
sphalerite has quit [Quit: nixos 20.09, here I come!]
cjpbirkbeck has joined #nixos-chat
vesper has quit [Ping timeout: 256 seconds]
vesper11 has joined #nixos-chat
vesper has joined #nixos-chat
<jD91mZM2>
sphalerite: ma27 is correct, my username is indeed confusing. One of the many points of it is that when someone IRL that I don't like asks for my username, I just power through it and leave
<jD91mZM2>
They'll be like "yeah, uh, I didn't... oh... never mind"
<jD91mZM2>
it's a flawed plan, a single pen and paper foil it
<jD91mZM2>
s/foil/\0s/
vesper11 has quit [Ping timeout: 272 seconds]
<lovesegfault>
jD91mZM2: your bio on github is one of my favorite things
<lovesegfault>
literally every time I read it I laugh
<jD91mZM2>
:D
<jD91mZM2>
It's incredibly, incredibly old. I'm not sure why I haven't changed it, but now I'm pretty happy that I didn't
<lovesegfault>
Please never change it
<jD91mZM2>
I guess I won't, lol
<jD91mZM2>
I'm getting Deja Vú, have you praised my bio before, or was that someone else maybe?
<lovesegfault>
Hopefully I have!
<jD91mZM2>
Hmm, not finding anything in the chat logs (thank God (samueldr) for those btw), I guess this is the start of my path towards insanity. I knew it would happen eventually ¯\_(ツ)_/¯
<samueldr>
,jdwhat
<{^_^}>
jD91mZM2
<samueldr>
it sums up what you said earlier
<V>
"what's your username?" "V." "how do you spell that?" "V." "like, v e e?" "no, V."
<jD91mZM2>
LOL
<samueldr>
wee
<jD91mZM2>
How many platforms have you been able to hog that name on, V?
<V>
hog? hardly
sphalerite has joined #nixos-chat
<V>
IRC only, pretty much
<jD91mZM2>
Aww
<V>
but it doesn't matter all that much, because I don't care for having a username
<jD91mZM2>
Makes sense. I'm starting to think I should've separated my usernames across services, just like everyone should separate their passwords
<V>
since I also get to have conversations like "what's your name IRL?" "V." "no, like, what's your *real* name?" "it's V."
<V>
and then people get very confused
<jD91mZM2>
It is?
<V>
yes
<jD91mZM2>
That is cool, did you change name or were you actually born as V?
<jD91mZM2>
I mean I don't mean to be disrespectful, just confused and amazed
<V>
is anyone born as anything? :)
<jD91mZM2>
I'm pretty sure I was born as a human...
<jD91mZM2>
But yeah I mean did your parents give you that name?
vesper11 has joined #nixos-chat
<V>
nah. but they kind of forfeited any right to call me anything
<V>
my point is really just that names are common identifiers given to something by people, and there's really no reason why anyone should keep a name given to them before they had a personality
<jD91mZM2>
Absolutely, no, you're right.
<gchristensen>
it is your name, do whatever you want with it
<V>
I quite like the system some cultures have/had where a child gets an arbitrary name and then gets to choose their own at coming of age
<V>
much more respectful
<gchristensen>
that sounsd cool
<V>
but there's no reason why you can't just do that regardless ^^
<jD91mZM2>
Some day I want to get really rich and legally get all the names. Like a crap ton, until I hit some kind of strange limit and break all systems
<jD91mZM2>
And then, at least how it works in Sweden, I can change between my first names for free I think. Well actually, I did hear there's a new law that makes your first name always be your *first* first name, but idk if it applies to me since I was born before this
<gchristensen>
just take the name Bo and you'll already break a bunch of systems
<V>
but yeah. I feel that mononyms are a concept that people can grasp, or that they're plausibly going to have come across at some point, but a single character? that usually blows peoples' minds :)
<__monty__>
Is there a way to protect a directory from `rm -r` without taking away read permission, don't wanna miss out on ls?
<sphalerite>
__monty__: -w
<sphalerite>
__monty__: if the directory is write-protected, it can't be deleted from.
<sphalerite>
Though you'll need to apply that recursively if you want to protect the whole tree.
<sphalerite>
of course, root can still do whatever it wants.
<__monty__>
sphalerite: No, I want to be able to delete files in it, add files to it, look at the files in it, but not accidentally delete the directory.
<__monty__>
I don't need protection from root, just from myself.
<sphalerite>
alias rm='rm -i'
<sphalerite>
or zfs snapshots :p
<__monty__>
Nah, that's a recipe for blindly saying y.
<colemickens>
we need some nix community based musicians to contribute tunes for zimbatm's next stream :P
<sphalerite>
hm, simpson seems to be away right now.
<V>
[21:45] <jD91mZM2> Makes sense. I'm starting to think I should've separated my usernames across services, just like everyone should separate their passwords <- I do this. for anything where is isn't something publicly identifying me (e.g. GitHub), I just generate a random one and keep that in my password manager as well. comes with the nice side effect of not requiring something untaken across all the services
<colemickens>
I wonder how they make the random bit. I like to sha256 silly things and then take substrings that look nice for stuff like that
<jD91mZM2>
sphalerite: Another solution: Switch to using trash-cli. I mean it's kinda silly, but I've gotten the habit to never use rm anymore, and it doesn't often save me I'm really happy when it does
<immae>
__monty__: you can alias rm = rm -i
<__monty__>
I mostly only use rm when I need the space.
<__monty__>
immae: That just trains you to press y with muscle memory.
<immae>
right
<jD91mZM2>
sphalerite: TIL MichaelRaskin is 7c6f434c, I had no idea lol
<jD91mZM2>
V: Nice
<sphalerite>
colemickens: base64 for a wider variety of characters ;)
<sphalerite>
base64 </dev/urandom | head -c 10
<V>
tr -cd a-z </dev/urandom | head -c10
<V>
for more controlled selection
<V>
(then you can vary the set and such)
<samueldr>
no *you* are an ofjevfikznko
<sphalerite>
pwgen -0 for things that are halfway pronounceable
<andi->
How terrible is faking `curl` in a nix build with a script that just copies from /nix/store? Finally building coreboot...
<infinisil>
Acceptable!
<samueldr>
as terrible as you want it to be
<samueldr>
I think it's not that bad of an idea though, as a way to move forward, leaving a huge breadcrumb to come back to and fix later
nckx has quit [Ping timeout: 264 seconds]
nckx has joined #nixos-chat
bqv has quit [Quit: WeeChat 2.9]
bqv has joined #nixos-chat
__monty__ has quit [Quit: leaving]
rajivr has joined #nixos-chat
* colemickens
is so intimidated trying to jump into the middle of C++ codebase
<{^_^}>
nix#4193 (by colemickens, 23 hours ago, open): nix: flakes: allow overriding flake inputs in bulk with extra nix file
<worldofpeace>
colemickens: ooh, you're suggestion of an input override file is a great idea
<colemickens>
cole-h: you use flakes and h-m and use the h-m module directly right? do you understand the who thing about nixUnstable being unsupported? Is that only if h-m is used "standalone" ?
<energizer>
my APC UPS makes a *plink* noise every minute or so. is this normal?
<infinisil>
colemickens: That seems a bit too ad-hoc to me, I feel like there has to be a better solution than that
<colemickens>
infinisil: hm I'm a bit surprised at that reaction. I have a hard time imagining a good alternative (but that's likely my limited imagination).
<ashkitten>
ugh that reminds me i need to replace my UPS
<colemickens>
I thought it seemed elegant-ish or at least soemwhat simple? it would be the same schema as flake.inputs and the tooling could load the override file, parse it the same way and then update the inputs it'd already loaded.
<ashkitten>
i even replaced the battery but it doesnt handle power outages at all
<colemickens>
Just curious what you're thinking :)
<infinisil>
colemickens: Ahh I didn't realize it's the same format as the lock file
Raito_Bezarius has joined #nixos-chat
<infinisil>
s/lock file/flake.nix
<colemickens>
infinisil: I should be more explicit!
<ashkitten>
my uninterruptible power supply is quite interruptible
<infinisil>
colemickens: Maybe this could be generalized to a full second flake.nix file
kalbasit has quit [Ping timeout: 240 seconds]
<infinisil>
Well, one you could pass with --flake-override
<infinisil>
And it would temporarily override any values in flake.nix
<colemickens>
infinisil: yes, maybe, I was just wondering how far it makes sense to go with the override idea.
<colemickens>
and then I also added a point about maybe calling it `--impure-overrides`? idk, trying to emphasize that this won't update the lock file and is considered a one-off-ish thing?
<cole-h>
colemickens: That's my understanding -- if it's used as a package rather than the module