<elvishjerricco>
Then again, they are working in extremely cold temperatures. I know they spend like two thirds of the helicopters power just on heating the electronics to a usable temperature
<samueldr>
that's the exact thing AFAIUI
<samueldr>
I would assume 500hz is a typo
<samueldr>
or uh, mistake
<elvishjerricco>
Yea I can't find any claims on clock speed in any other article talking about the Ingenuity+Snapdragon
<elvishjerricco>
I wouldn't be surprised if it is severely underclocked though. I bet they have pretty extreme stability requirements
<samueldr>
would it be enough clock cycle to autonomously fly?
<samueldr>
I would have epected 500mhz
<samueldr>
not 500_hz_
<elvishjerricco>
Oh, yea I'm agreeing there's no way it's 500hz
<elvishjerricco>
They must have misreported 500Mhz or something
<samueldr>
500mhz could entirely make sense here given those chips may run at that clock rate depending on the governor used
<elvishjerricco>
Unfortunately, the Perseverance itself is running some proprietary OS. But the Ingenuity could have been running NixOS if you'd been working for Nasa samueldr!
<samueldr>
haha, booting mobile nixos even
<elvishjerricco>
Hell, they could have put a raspberry pi on that thing and it would have been about the same. That's kinda wild.
<samueldr>
kind of... but that's ~2015 era hardware for a reason
<samueldr>
it wasn't designed yesterday
<samueldr>
>> The Raspberry Pi 3 Model B was launched in February 2016
<samueldr>
that might have been too new even
<elvishjerricco>
Yea, I'm sure the 801 was an extremely carefully made decision
<samueldr>
and AFAIUI it's because the qualcomm thing is not just the CPU, but the whole flight platform
<samueldr>
designed from the ground up (hah!) for drone applications
<{^_^}>
#113090 (by cole-h, 6 days ago, open): nixos/zrepl: init
<supersandro2000>
elvishjerricco: it actually works
<supersandro2000>
not like bashtop
<supersandro2000>
the only other way to do that without bash would be to patch readline or bash so
omnd has quit [Ping timeout: 272 seconds]
kini has quit [Quit: bye]
kini has joined #nixos-chat
kini has quit [Quit: bye]
<lovesegfault>
What is it about physical exercise that makes getting waster easier?
<lovesegfault>
I worked on the field all day
<lovesegfault>
and now I had like 2 beers
<lovesegfault>
and I am _beyond_ wasted
<lovesegfault>
what is the cause of this
kini has joined #nixos-chat
slack1256 has quit [Ping timeout: 240 seconds]
Dotz0cat has quit [Read error: Connection reset by peer]
Dotz0cat has joined #nixos-chat
BaughnLogBot has joined #nixos-chat
<patagonicus>
Remember to turn off debug logs after stuff starts working. I've just deleted a 28GB Gluster client log file, which prevented me from starting chrome (I guess chrome does something in /tmp?).
<patagonicus>
Also, I know that filesystems can have different amounts of overhead for small files, but I'm copying 1.8T from ext4 to glusterfs on xfs and it's already taking up 3.4T and it's not even done.
cole-h has quit [Ping timeout: 240 seconds]
<ar>
patagonicus: btw, why gluster, which will have inherently more overhead because of it sits on top of another filesystem, and not ceph?
<patagonicus>
Good question, I haven't really looked into Ceph yet. But I'm in the early stages of this project and it's basically just "I want to play around with a distributed filesystem" and "I want a few TB accessible over the network". Performance is going to be horrible compared to a medium tier amd64 tower with the same four hard drives in it anway …
<ar>
ceph typically used to sit on top of another filesystem (and back in those times, xfs was recommended), but it's using its own block device storage layer - bluestore
<patagonicus>
To be fair, Gluster basically stores the data raw on the underlying filesystem, so I think the problem here is the xfs tuning options that the manual suggest and don't work well for my use case (which, to be fair, isn't really a standard Gluster setup). But I want to try the restore anyway, so I'll recreate one brick with default xfs settings to
<patagonicus>
see if that's closer to the ext4 values. And if it isn't I might just switch back to ext4.
tilpner_ has joined #nixos-chat
tilpner has quit [Ping timeout: 265 seconds]
tilpner_ is now known as tilpner
<patagonicus>
Reformatted one brick and told gluster about it, now it's doing something. :D It has recreated some small files, but no big ones yet AFAICT.
kini has quit [Remote host closed the connection]
kini has joined #nixos-chat
kini has quit [Ping timeout: 260 seconds]
kini has joined #nixos-chat
<patagonicus>
Yeah, so, it seems to be doing something along the lines of 300KByte/s network traffic, so rebuilding a few TB could take a bit. Hmm.
Dotz0cat has quit [Ping timeout: 264 seconds]
<patagonicus>
ar: So, ceph says minimum for ceph-mon is 24GB RAM + 4GB+ per ceph-osd (though for ceph-osd 2GB may work). The machines I'm running that on only have 2GB RAM, so they seem to be the bare minimum for ceph-osd with nowhere to run ceph-mon. Red Hat Gluster docs on the other hand seem to recommend 4GB RAM for each node, although that seems to include
<patagonicus>
some monitoring as well.
<ar>
patagonicus: it's kinda sorta true if you run fast (>10G) network and expect some decent performance
<patagonicus>
I mean, if Gluster performance stays where it is, I'll try ceph instead, but I know someone else who's running a 6 node cluster on the same hardware. I'll have to ask if they've tried rebuilding (although that one is 3 replicas instead of my 2 + 1 arbiter, I think). Otherwise the performance is acceptable for my use case, but it's so far rebuild ~1
<patagonicus>
GB, which I don't really consider ok.
<ar>
edef: sorry to bug you, but would you mind bumping solvespace a bit? new versions have "Helix" and "Revolve" tools. did a lazy local bump to try them out (https://dpaste.com/FLNGTXXN5) and it seems to work
Guest84 has joined #nixos-chat
<Guest84>
,locate libstdc++.so.6
<{^_^}>
Found in packages: libgccjit, robomongo, gcc-unwrapped.lib
jD91mZM2 has quit [Read error: Connection reset by peer]
<infinisil>
Guest84: You can spam the bot in #bottest, or you can install nix-index locally to not have to use the bot :)
<infinisil>
,locate
<{^_^}>
Use ,locate <filename> to find packages containing such a file. Powered by nix-index (local installation recommended) https://github.com/bennofs/nix-index
<Guest84>
I come here only for the bot. `nix-index` locally takes 30 minutes and often segfaults
<Guest84>
will use bot now
<__monty__>
You can also spam the bot in queries.
<infinisil>
Oh yeah
<joepie91>
`nix-locate` seems to just work for me?
<joepie91>
don't recall ever running a manual index
<joepie91>
@ Guest84
<Guest84>
Yeah well that's a one time thing :)
<Guest84>
joepie91: You probably don't remember. Thing is I need to clean ~/.cache often and that makes things messy (I should save nix-index maybe)
<infinisil>
joepie91: It definitely doesn't work without having run nix-index first once
<__monty__>
Aren't nix-index and nix-locate different tools?
<__monty__>
Or am I thinking of nix search?
<infinisil>
__monty__: nix-index and nix-locate are separate binaries but the same tool
<infinisil>
nix-index created the database, nix-locate queries the database
<infinisil>
s/created/creates
<infinisil>
,locate nix-locate
<{^_^}>
Found in packages: nix-index
<infinisil>
,locate nix-index
<{^_^}>
Found in packages: nix-index
<infinisil>
:P
<__monty__>
Maybe joepie91 is using a nix-index module that regenerates the DB when nixos-rebuilding?
<infinisil>
Oh yeah I think there is something like that
<infinisil>
Or at least I've seen a PR
* joepie91
has no idea
<joepie91>
also, is there any reason why it couldn't be precomputed by Hydra?
<__monty__>
Maybe it takes into account overlays?
<Guest84>
At least for me nix-index is generated in ~/.cache, (maybe it can look in a series of locations?). That said, who would want to wait for every rebuild-switch to add 3 0mins?
<__monty__>
And maybe nix-locate runs nix-index for you if the DB's missing?
<Guest84>
my nix locate needs are fulfilled for the day, thanks guys.
<Guest84>
bye
Guest84 has quit [Quit: Connection closed]
<__monty__>
I wouldn't really mind. I don't spend my time rebuilding my system.
<__monty__>
I don't search for files and packages all that often either though.
<ar>
hm. does anyone know of something to set xinput properties per-application window (especially under kwin/plasma5)? i would love to be able to disable trackpoint middle mouse button scroll in, for example, solvespace
Dotz0cat has joined #nixos-chat
<gchristensen>
I wish there was a tool for one-off zfs sends that just figured out all the options you want
BaughnLogBot has quit [Ping timeout: 260 seconds]
BaughnLogBot has joined #nixos-chat
ekleog has quit [Quit: WeeChat 2.9]
<gchristensen>
anyone have opinions on sha256 vs. skein vs. edonr?
<samueldr>
because humans have been "gaming" the algorithm into giving it what it wants and not the truth
<samueldr>
similarly "select shop store fronts" with text in scripts I cannot read... maybe it's an ad? how can I know which of those is a shop vs. something else???
<colemickens>
I wish opting out were a legitimate choice.
<colemickens>
Getting tired of answering those things _for services I pay for).
<samueldr>
__monty__: good point
<V>
Hey, on the bright side we're feeding Google's machine learning algorithms a bunch of garbage
<samueldr>
colemickens: I wouldn't mind having those on _public data sets_
<colemickens>
Just let me enable 2FA and then stop with these theatrics.
<samueldr>
yeah
<colemickens>
samueldr: sure, yeah, I think that's hcaptcha or something?
<colemickens>
hm maybe not, this looks ... more productized than I remember it being.
<V>
Isn't hcaptcha the one that supports authoritarian regimes
<V>
By CF
<__monty__>
Anyone else tries to do as bad as job as possible on the categorizing captchas?
<colemickens>
__monty__: I've just assumed that they cross-check and then require you to complete substantially more since your answers differ from others
<__monty__>
colemickens: That's the price to pay for rebelling ; )
* colemickens
oh they have a token too, now I feel bad I mentioned them without looking at it more
aleph- has quit [Quit: WeeChat info:version]
aleph- has joined #nixos-chat
ece has joined #nixos-chat
srk has quit [Ping timeout: 268 seconds]
srk has joined #nixos-chat
aleph- has quit [Quit: WeeChat info:version]
aleph- has joined #nixos-chat
<genevino>
i might have weird ideas, but is it currently possible to use the AUR on nixos?
<genevino>
__monty__: i'm exceptionally bad at solving captchas for a human.
aleph- has quit [Quit: WeeChat info:version]
aleph- has joined #nixos-chat
aleph- has quit [Quit: WeeChat info:version]
aleph- has joined #nixos-chat
slack1256 has joined #nixos-chat
slack1256 has quit [Remote host closed the connection]
<supersandro2000>
10 times and you need to have 100% right
<lovesegfault>
cole-h, bbigras: Does say start for y'all on the latest nixos-unstable?
<colemickens>
looking for creative solutions. I have a headless NixOS box with a webcam. Looking to easily plumb it into a Jitsi conference or a Discord conference, etc.
<colemickens>
I've seen theres an OBS-Studio fork that uses WebRTC all over internally that can supposedly broadcast to a Jitsi room. I don't quite understand why there isn't more documentation on interop at the signaling edge so that webrtc stuff can interoperate better.
<colemickens>
alternatively I know there's some sort of simple rtmp server packaged in nixpkgs
<lovesegfault>
I can't get it to work with newer _latest kernels
<lovesegfault>
I reverted back to lts kernel and it works again
<bbigras>
lovesegfault: it works for yesterday's nixos-unstable
<lovesegfault>
bbigras: did you reboot?
<bbigras>
I'm on it right now on my laptop
<bbigras>
I think so. brb I'll try again
__monty__ has quit [Quit: leaving]
<lovesegfault>
👍
<lovesegfault>
ashkitten: is pipewire 0.3.22 working for you?
<ashkitten>
we don't have that yet in nixpkgs
<lovesegfault>
ashkitten: yeah we do, it landed
<ashkitten>
we have 0.3.21 last i checked
<lovesegfault>
shows 0.3.22 here
<lovesegfault>
(broken, fwiw)
<bbigras>
lovesegfault: I just rebooted. It's fine
<lovesegfault>
bbigras: intel gpu?
<bbigras>
yep
<lovesegfault>
I suspect it's an AMD-only issue
<lovesegfault>
colemickens: did you have an AMD gpu or is memory failing me?
<lovesegfault>
bbigras++
<{^_^}>
bbigras's karma got increased to 6
<ashkitten>
lovesegfault: if you have 0.3.22 then you've done something wrong because there is no PR for that in nixpkgs and it has some config changes that require our module to be tweaked
<bbigras>
I'm on 9816b99e71c3504b0b4c1f8b2e004148460029d4