gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
drakonis has quit [Quit: WeeChat 2.2]
drakonis has joined #nixos-chat
<jackdk> suppose I wanted to get djgpp into nixpkgs. I've tried lying down and the idea won't go away. Where would I start?
<jackdk> (as a cross-compiler)
<samueldr> not sure, but if I were to look at it, I'd start to look at mingw and emscripten; mingw we have support I'm sure, emscripten I think
<samueldr> (and I'd need knowledge about djgpp, but I assume you do :))
<jackdk> ah yeah that's a good place to start digging
qyliss^work has quit [Quit: bye]
qyliss has quit [Quit: bye]
qyliss has joined #nixos-chat
qyliss^work has joined #nixos-chat
endformationage has joined #nixos-chat
<drakonis> i'm making a nix for starters talk
<drakonis> gonna be my first talk ever aaaaaa
<drakonis> i have cold feet
<drakonis> i need some advice on what to cover
iqubic has joined #nixos-chat
<colemickens> isn't there a tool that is meant to help testing of PRs?
<colemickens> I'm finding relevant stuff in here, much automated, that I'd be happy to test and report results for.
<ldlework> drakonis: I wonder if the Nix Challenge is a good conceptual vector by which to introduce Nix
* colemickens is curious about the r13y.com footnote... very curious
<drakonis> ldlework: tell me about it
<drakonis> rather, what is the nix challenge?
<ldlework> drakonis: You start the talk by illustrating the frustrating typical experience with hard-drive failure of your daily workstation. Enumerate the real losses most people will incur if that were to happen right now. Quickly summarize how some people might mitigate this with dotfile management and backups. But we know that doesn't cover everything. Ask them to imagine what it would take to be able to
<ldlework> perfectly replicate that failed workstation without imaging. And then build Nix up conceptually by explaining how the store and profiles work.
<ldlework> drakonis: the nix challenge is where you purposefully dd your daily workstation, and then document yourself recovering from it, and the ease with which you did, because of your use of Nix
<ldlework> typically one would only be willing to do that if they're sure they wont lose any state
jasongrossman has joined #nixos-chat
<drakonis> ah alright
<drakonis> neat but i don't think it would work out here
<drakonis> i could try it once i have a laptop
<drakonis> i need to get a laptop :|
<drakonis> i'm going to get a dell xps at this point
<drakonis> aaaaaaaaaaaaa wtf this is crazy expensive
jasongrossman has quit [Read error: Connection reset by peer]
<drakonis> *eyepop*
MichaelRaskin has quit [Quit: MichaelRaskin]
endformationage has quit [Quit: WeeChat 2.3]
ottidmes has quit [Ping timeout: 268 seconds]
iqubic has quit [Ping timeout: 264 seconds]
disasm_ has joined #nixos-chat
jasongrossman has joined #nixos-chat
disasm_ has quit [Ping timeout: 245 seconds]
disasm_ has joined #nixos-chat
jackdk has quit [Ping timeout: 250 seconds]
disasm has quit [Quit: WeeChat 1.9.1]
drakonis has quit [Quit: WeeChat 2.3]
disasm_ is now known as disasm
<disasm> and server is back up!
<disasm> luckily only unrecoverable zfs dataset was /nix (which I didn't try super hard since I could rebuild that easily enough)
<clever> disasm: was able to read all other datasets? or did it recover via `zfs destroy tank/nix` ? i can see `zfs send` being tricky if you cant make a snapshot...
<disasm> clever: imported read only, was able to zfs send everything but zroot/nix which failed with I/O errors (tried twice) Didn't dig into it much further as a nixops deploy got it back where it needed to be with a fresh /nix :)
<clever> ah, maybe `zfs send` just works on a raw dataset, since its read-only, and doesnt need a snapshot
<disasm> yup, as long as it isn't mounted. That's how I backed up/restored my laptop that I sent back and had replaced
<clever> ah
<clever> i guess a snapshot/bookmark is only a hard-requirement if you want to support incremental sends
johanot has joined #nixos-chat
johanot has quit [Ping timeout: 246 seconds]
johanot has joined #nixos-chat
__monty__ has joined #nixos-chat
johanot has quit [Ping timeout: 245 seconds]
<sphalerite> hm, so can you send a dataset if it's mounted read-only?
johanot has joined #nixos-chat
<disasm> yeah
<sphalerite> what will they come up with next!
__Sander__ has joined #nixos-chat
johanot has quit [Ping timeout: 245 seconds]
johanot has joined #nixos-chat
aszlig has quit [Quit: Kerneling down for reboot NOW.]
aszlig has joined #nixos-chat
<lejonet> It makes perfect sense, as what you want with sending a dataset is that it doesn't mutate while sending, causing you to not have atomic state, ro mounting ensures that in a way
lopsided98 has quit [Quit: No Ping reply in 180 seconds.]
lopsided98 has joined #nixos-chat
jtojnar has joined #nixos-chat
johanot has quit [Quit: WeeChat 2.2]
das_j has quit [Remote host closed the connection]
das_j has joined #nixos-chat
<emily> on the subject of zfs, does anyone know how `zfs send`ing stuff from something with snapshots to something with them disabled work?
drakonis_ has quit [Ping timeout: 245 seconds]
ottidmes has joined #nixos-chat
<tilpner> By "with them disabled", do you mean "compiled zfs without snapshot support"?
<tilpner> Or "without auto-snapshotting"?
drakonis_ has joined #nixos-chat
endformationage has joined #nixos-chat
<disasm> lejonet: just to clarify, I'm not even sure if it works with a ro mounted volume. in this case I imported the pool ro (because it hung importing normally) and no volumes were mounted when I sent them.
<lejonet> disasm: ah, okay, I misunderstod then :) if the dataset is mounted, you'd usually have to make a snapshot that you send instead, but it was a while since I did any ZFSing :P
<lejonet> so thought that they maybe relaxed that if the dataset was mounted ro, then you could just send the dataset without taking a snapshot first
yl has quit [Ping timeout: 264 seconds]
<disasm> lejonet: yeah, haven't tested that use case. Maybe it works, maybe not!
<samueldr> following previous discussions https://twitter.com/chiakokhua/status/1102574053166968833
* samueldr wonders in what subtle way it will be incompatible with thunderbolt 3
<joepie91> ways*, I'm sure
<__monty__> They probably have no choice there though. I imagine thunderbolt's patent encumbered?
<samueldr> intel provided it to the usb-if, they can do whatever they want with their patents, including granting usage to usb-if members I guess
<__monty__> Hmm, didn't know intel was the principal on thunderbolt. Tied to apple in my head.
<mdash> ahaha USB4 based on thunderbolt :-|
<mdash> not too surprising I guess
<gchristensen> good grief
<mdash> "If USB is a single tire on fire, Thunderbolt is a small container of tires on fire. The phone number on the side is painted over and a lawyer has threatened you for looking at it too inquisitively"
<samueldr> may I remind you: don't shoot the messenger?
* samueldr ducks
<mdash> samueldr: i'm not upset at you! in fact having the protocol documented is a big step forward
<samueldr> :)
<samueldr> it was a plural you
<mdash> previously every thunderbolt device in existence was identical to an Intel reference design
<mdash> because they didn't share any other specs with manufacturers
<samueldr> yeah, there's that thing where *now* we might see it on non-intel hardware
<samueldr> hoping other implementations adhere to some standard of security
<__monty__> The standard where cables get periodical firmware updates?
<mdash> __monty__: and may or may not broadcast your data over wifi, store it on flash, etc
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 250 seconds]
Taneb has quit [Quit: I seem to have stopped.]
drakonis_ has joined #nixos-chat
Taneb has joined #nixos-chat
drakonis1 has joined #nixos-chat
johanot has joined #nixos-chat
drakonis has quit [Ping timeout: 255 seconds]
drakonis_ has quit [Ping timeout: 258 seconds]
drakonis has joined #nixos-chat
johanot has quit [Quit: WeeChat 2.2]
__Sander__ has quit [Quit: Konversation terminated!]
<ldlework> Anyone got some sweet dropbox nix expressions maybe including mounting some authenticated folder or something?
endformationage has quit [Ping timeout: 240 seconds]
yl has joined #nixos-chat
johanot has joined #nixos-chat
nckx has quit [Quit: Updating my GNU Guix server — https://guix.info]
jtojnar has quit [Read error: Connection reset by peer]
<infinisil> I've written an email to a professor at my university, and I haven't gotten a reply in a week, is it okay to send another mail to remind them?
<gchristensen> sure
jtojnar has joined #nixos-chat
<infinisil> I'm wondering if I should wait longer, they might be busy or something
<infinisil> Eh it'll be fine
nckx has joined #nixos-chat
<__monty__> As long as you're polite about it it's fine.
endformationage has joined #nixos-chat
<infinisil> Interesting problem: Proving to others that a number was chosen randomly and isn't some chosen value.
<infinisil> Ohh yes very
<__monty__> That comes up in blockchain gambling.
<__monty__> Wouldn't providing a pair (a, oneway a) be enough? The second being the chosen random number? Since we postulate the function's hard to invert?
<infinisil> Oh yeah you could totally have a trustworthy random number by saying it's the hash of the block mined an hour ago or so
<infinisil> __monty__: One could just brute force through lots of a's to find an exploitable `oneway a`
<infinisil> (unless I misunderstood what you meant)
<__monty__> No, I think you get it. But how realistic is bruteforcing a 256 or more bit space?
<__monty__> Or do you require things like being able to generate a die roll?
<__monty__> And proving it was "rolled" with a fair die?
<samueldr> maybe more "nothing up a sleeve" if the statement is somehow provably made at a moment in time, "the next block mined" or something like that? because maybe you could wait until a block is sleevy enough?
<gchristensen> depends if your adversary is the NSA, and also the NSA provided the number
<infinisil> __monty__: bruteforcing 256 bit isn't very practical if you need to find a single number, but it could be that many more numbers allow exploits
<infinisil> E.g. an exploit is possible if the number is dividable through 1056432
<infinisil> samueldr: Yeah that's what I'm thinking too
<infinisil> Wait, blockchains aren't useless???
<samueldr> they're not useless, but maybe we need to use less?
<infinisil> I now think there's these three valid uses for a blockchain: currency (bitcoin), name reservations (namecoin), proving that a number was chosen randomly
<samueldr> how about heating?
<gchristensen> how about funding R&D in to haskell and PLT?
<__monty__> I'm still hopeful blockchains will allow us to execute all notaries : >
<gchristensen> well if that is when my time comes, that is when my time comes
<__monty__> You're a notary? May be different in the US though. They're overpriced nepotist sons of bitches over here.
<andi-> At the current state I rather pay someone (lots) of money for attestating something then betting on some hyped language embedded in a hyped technology...
<gchristensen> pretty much anyone can be a notary: take a class, pay a fee, buy a seal.
<__monty__> There's a limit to how many notaries there can be here and in practice the title is transferred from father to son here more often than not.
<gchristensen> ah, so a bit protectionist then
<__monty__> s/bit/bit much : >
<gchristensen> :)
<gchristensen> where is here?
<__monty__> Belgium.
<andi-> Here (germany) it seems like only lawyers can become a notary.
<__monty__> That might be the case here too.
* disasm starts sending all his documents requiring a notary to gchristensen
<gchristensen> oh dear
<__monty__> Nixpkgs notarize when?
<andi-> just append your documents to nixpkgs git objects..
<gchristensen> lol
johanot has quit [Quit: WeeChat 2.2]
jackdk has joined #nixos-chat
<gchristensen> a fun thing to do is use genisoimage to smuggle data in to a VM without installing something like smb
Moredread[m] has joined #nixos-chat
<Moredread[m]> samueldr: I was thinking about spaceheaters that use cryptomining hardware to offset their cost. That's the only way it makes atleast a bit of environmental sense to have proof of work blockchains
<Moredread[m]> and there goes my businessplan :p
<gchristensen> I wonder if anybody has actually really bought them, or actually distributes work to them
<Moredread[m]> "5000 cores available" doesn't sound like to many are deployed
<andi-> and every 6 months you need a new sideboard because that older generation of asics isn't competitive anymore? :D
<andi-> I sometimes wish `nix-build … -K` would also work with remote builders :/