<gchristensen>
not sure if that smell is my legs cooking or the banana bread in the oven
<cole-h>
I didn't know... cooked human legs smelled similar to banana bread...
waleee-cl has quit [Quit: Connection closed for inactivity]
lopsided98 has quit [Remote host closed the connection]
lopsided98 has joined #nixos-chat
<cole-h>
gchristensen: Building pgtk right now, thanks for the tip. Interested to see how well it works :)
<gchristensen>
:)
<cole-h>
`nix-top` is pretty nifty, being able to see what's building and how
<cole-h>
Thank goodness for FF reader mode
<cole-h>
150 char long lines :puke:
<drakonis1>
dark raeder is real good if you arent using that one
<cole-h>
I am, and it is very good. The colors are almost always spot on, too
<cole-h>
No more real need for Stylus anymore
<colemickens>
drakonis1: that makes me think of guix+shepard
<drakonis1>
its good stuff.
<drakonis1>
i'm back on nix, mind you, because it has spent more time on the cooker
<drakonis1>
so much to be taken from it
<cole-h>
gchristensen: OMG THIS IS WONDERFUL
<cole-h>
gchristensen+++++++++++
<cole-h>
Wayland-native Emacs is all I need. I can die happy now.
<ashkitten>
qyliss++
<{^_^}>
qyliss's karma got increased to 33
<cole-h>
(gchristensen++)
<drakonis1>
i envy y'all with your waylands
<cole-h>
gchristensen++
<{^_^}>
gchristensen's karma got increased to 210
<cole-h>
I'll get there eventually
<cole-h>
Enjoy all those pings
<ashkitten>
drakonis1: stuck on nvidia with me?
<cole-h>
I'm on Nvidia, too, though
<drakonis1>
yeah
<cole-h>
Just use nouveau :-)
<gchristensen>
it isn't perfect, but it is pretty good :)
<ashkitten>
cole-h: i do gaming on my computer so i can't use nouveau
<drakonis1>
and kill off all my perf? no can do.
<drakonis1>
no cuda here
<cole-h>
ashkitten: I have a passthrough VM for my gaming needs :P
<drakonis1>
i think there's no vulkan driver for nouveau right now
<cole-h>
That is correcet
<ashkitten>
cole-h: i don't even want to deal with windows in a vm lol
<cole-h>
One of its shortcomings
<cole-h>
s/correcet/correct
<drakonis1>
i'm still waiting on nvidia to release the power management firmware and raise the stakes on supporting their hardware on linux
<cole-h>
Sweet, I just got a fifo fault and now sway is dead
<drakonis1>
sweet.
<ashkitten>
drakonis1: have they stated they'd do that?
<cole-h>
At least jack didn't die, so I can still jam out
<drakonis1>
they have a pending talk regarding increasing support for linux
<ashkitten>
drakonis1: doesn't mean anything until they say it explicitly
<cole-h>
So, what do you guys do when your keyboard doesn't respond to switching VTs?
<drakonis1>
they actually did.
<drakonis1>
"We'll report up-to-the-minute developments on NVIDIA's status and activities, and possibly (depending on last-minute developments) a few future plans and directions, regarding our contributions to Linux kernel; supporting Nouveau (the open source kernel driver for NVIDIA GPUs, that is in the Linux kernel), including signed firmware behavior, documentation, and patches; and NVIDIA kernel
<drakonis1>
drivers."
<ashkitten>
hm
<ashkitten>
well i'll believe it when it happens
<cole-h>
^
<drakonis1>
same.
<drakonis1>
its probably born out of a desire to not be driven out of the linux market by amd
<drakonis1>
cant wait for it tho
<drakonis1>
sadly its just a cool month way.
<drakonis1>
s/way/away
cole-h has quit [Quit: WeeChat 2.7]
das_j has quit [Remote host closed the connection]
ajs124 has quit [Remote host closed the connection]
ajs124 has joined #nixos-chat
das_j has joined #nixos-chat
cole-h has joined #nixos-chat
das_j has quit [Quit: killed]
ajs124 has quit [Quit: killed]
ajs124 has joined #nixos-chat
das_j has joined #nixos-chat
<cole-h>
gchristensen: I guess one downside is it doesn't appear to like emacsclient
drakonis1 is now known as drakonis
<qyliss>
gchristensen: oh wow, yeah, I just updated pgtk for the first time in ages and it's so much better
<pie_[bnc]>
omg theres a "participating" filter! great!
<pie_[bnc]>
ive been using commented:myname and author:myname or stuff like that
<pie_[bnc]>
"Notifications Beta" aha
evanjs has joined #nixos-chat
<andi->
It reminded me of outlook 365 where they clearly had text on the screen I just wasn't able to comprehent it since it was all tiny, grey on white etc..
malSet has quit [Ping timeout: 265 seconds]
drakonis has quit [Ping timeout: 246 seconds]
<evanjs>
but yeah the contributor/uniq package count on repology for nix -- 1 to 34 :D -- 43030/1250
<evanjs>
AUR in comparison -- 8683/22772 -> 2.62 to 1
<evanjs>
1 to 34 and 1 to 2.62. Man I am still tired. Rx vyvanse and kickstart aside :D
<Taneb>
But if you get caught you can be detained for weeks
<DigitalKiwi>
well the first rule is don't get caught
<Taneb>
Easiest way to do that is to not do the thing
<DigitalKiwi>
buncha normies out here jeez
<emily>
as a drug taker and law breaker I would recommend against trying to smuggle stimulants into japan unless you want to be behind bars for the next few decades.
<emily>
evanjs: fwiw, if concerta would work for you then your psych would probably be willing to prescribe it for the duration of a trip. i hear you need paperwork signed in triplicate for it though
<emily>
and might have to fill the prescription locally rather than taking it on the plane? not sure
<evanjs>
emily: oh for sure. I was on it when I was younger so it's not a _terrible_ option, but blah
<evanjs>
either concerta or straterra reallllly messed with my appetite :D like I felt sick just thinking about food. Might be different now, though
<eyJhb>
Not sure which kind of drug taker emily is now
waleee-cl has quit [Quit: Connection closed for inactivity]
<emily>
eyJhb: :)
* emily
isn't actually on stimulants, though she probably should be.
KeiraT has quit [Remote host closed the connection]
KeiraT has joined #nixos-chat
andi- has quit [Ping timeout: 260 seconds]
andi- has joined #nixos-chat
<elvishjerricco>
Other than file systems that include their own device manager, is there anyway to solve the write hole issue with software RAID-like systems on Linux? Even lvmraid doesn't have a solution since it's just mdraid underneath.
andi- has quit [Read error: Connection reset by peer]
andi- has joined #nixos-chat
evanjs has joined #nixos-chat
andi- has quit [Ping timeout: 246 seconds]
andi- has joined #nixos-chat
<gchristensen>
elvishjerricco: isn't the write whole a problem only on RAID-5 for filesystems which don't have atomic updates (like the zfs block pointer)?
<elvishjerricco>
gchristensen: No. E.g. ext4 with journaling can *almost* finish writing its data such that only some devices in the raid device don't have the complete write. Since it writes in-place, this leads to the write hole.
<gchristensen>
"whole" *shame*
<elvishjerricco>
CoW file systems would be closer to safe, but the uber / super block could still experience write hole
<gchristensen>
but still the write has happened or not happened
<elvishjerricco>
gchristensen: I guess you'll only experience problems with CoW if you fail to finish writing a new uber / super block before another improper shutdown, since then your array may still return the old new block, which would point to corrupt data at that point
<gchristensen>
it would not point to corrupt data, though
<gchristensen>
because the old block pointer is still valid
<elvishjerricco>
It would
<gchristensen>
why would it be corrupt?
<elvishjerricco>
start with uber block 1, totally safe and persistent
<elvishjerricco>
try to write uber block 2, but fail halfway and only complete the write on some disks
<elvishjerricco>
boot back up, the system experiences write hole and gets back uber block 1. This is fine since the data is there
<elvishjerricco>
Attempt to write a new uber block 2', fail part way through.
<elvishjerricco>
Now your uber blocks are a mix of 1, 2, and 2'
<elvishjerricco>
The data from uber block two is likely to have been overwritten in the transaction corresponding to 2'
<elvishjerricco>
If you read from uber block 2, you'll find corruption
<elvishjerricco>
which is totally possible because of write holes
<gchristensen>
I'm not sure this is true ...
<elvishjerricco>
What's the error in that hypothetical?
<gchristensen>
I'm still thinking through it :P
<elvishjerricco>
ZFS is only able to beat this because it can check the uber blocks of ALL physical disks and figure out what the deal is
<elvishjerricco>
I guess you could do the same with mdraid if you read your uber block from physical disks in stead of the mdraid device
<gchristensen>
I don't think the uberblock is written at all until all the data in its tree is already written
<elvishjerricco>
that's correct
<elvishjerricco>
But if 2' wrote data, it will likely have overwritten data from 2
<elvishjerricco>
So getting the 2 uberblock on next boot will let you read 2' data instead
<gchristensen>
and it writes 4 copies per disk, 2 at the start and 2 at the end, and it picks the highest txg id of them all
<elvishjerricco>
Yea, and refuses to write any new uberblocks until all the disks are synced
<elvishjerricco>
which isn't guaranteed with mdraid because writehole
<gchristensen>
gotcha
<gchristensen>
traditional filesystems / raid are scary :P
<elvishjerricco>
Yea, though this is an exceedingly unlikely hypothetical :P
<elvishjerricco>
And it's the only thing I can think of that would kill CoW on mdraid
<elvishjerricco>
And even then data loss would be one transaction
<gchristensen>
and I suppose is exactly the reason ZFS says don't use {hw,md}raid
<elvishjerricco>
It'd be fun to see someone try my mdraid+physical uberblock idea
<elvishjerricco>
That would be able to implement a ZFS-level fix
<elvishjerricco>
Completely pointless though when stuff like ZFS exists :P
<gchristensen>
I have a fix for you :P
<gchristensen>
ln -s /dev/zvol/mdraid /dev/md127
<elvishjerricco>
I wonder if this issue is why btrfs still technically has a write hole...
<elvishjerricco>
lol
<elvishjerricco>
gchristensen: Also worth noting this issue can only happen with mirrors of 3 or more disks. raid5/6 will always prefer real data over parity data, so it's more like raid0 until you need to replace a disk. raid0 would not have this problem at all.
<gchristensen>
raid0 lol
<gchristensen>
it would not have this problem for unrelated reasons
<elvishjerricco>
So I guess CoW on mdraid raid5/6 would basically be totally safe
<elvishjerricco>
Worst case, 2 and 2' end up being mixed together, failing checksum validation, forcing it to go to 1
<gchristensen>
out of curiosity, why are you thinking through this?
<gchristensen>
and, why not just use zfs :P
<elvishjerricco>
gchristensen: I keep trying to think of a way to get ZFS level integrity using traditional block layers so I can tell people what to implement when they complain that ZFS is a layering violation :P
<elvishjerricco>
But mostly it's just interesting
<gchristensen>
hmmm seems I have problems with this new emacs
<samueldr>
is it that it's not vim?
<samueldr>
(sorry)
<gchristensen>
nah
<gchristensen>
`emacs ./myfile` doesn't ... open that file
<samueldr>
sounds problematic
<samueldr>
bad wrapper not passing args?
<gchristensen>
I think my config is busted
<gchristensen>
there was something wrong with it, so I comment part of it out, and probably not enoug
malSet has joined #nixos-chat
<gchristensen>
oh no and my forward delete key doesn't work
psyanticy has quit [Quit: Connection closed for inactivity]
<gchristensen>
welp my emacs config is broken enough to be annoying and not enough for me to figure out how to fix it
waleee-cl has joined #nixos-chat
slack1256 has joined #nixos-chat
drakonis has quit [Ping timeout: 272 seconds]
drakonis has joined #nixos-chat
__monty__ has quit [Quit: leaving]
claudiii has quit [Quit: Connection closed for inactivity]
<cole-h>
Dangit Firefox, stop giving me OCSP errors >:(
<emily>
security.OCSP.enabled = 0 (...but while OCSP is broken, you probably shouldn't be getting OCSP errors, so maybe don't)
<emily>
unless you mean OCSP-stapling-related stuff in which case that's a cert problem and there's nothing you can do about it
<cole-h>
It just happens every once in a while for no discernable reason, on known-good websites (like just now, nixos.org)
<cole-h>
I probably monkeyed with my config too much
<emily>
it could be your connection just failing intermittently as firefox attempts to contact the ocsp server
* emily
has OCSP off. revocation is pretty broken and Chrome defaults to not bothering anyway, so might as well avoid the extra requests.
<emily>
hopefully ocsp must-staple takes off...
drakonis has quit [Quit: WeeChat 2.7]
drakonis has joined #nixos-chat
slack1256 has quit [Remote host closed the connection]