<colemickens>
I'm trying to think of a qt5+py+qml app that is already packaged in nixpkgs that I could use for an example...
<waleee-cl>
qutebrowser? no qml though
drakonis_ has quit [Ping timeout: 272 seconds]
drakonis1 has joined #nixos-chat
<gchristensen>
I feel old: I can't seem to learn how to use the new reddit UI (I rarely use reddit anyway.) there seem to be a ton of mysteriously clickable places to take me away from the thing I'm looking at
<ashkitten>
it's bad ui
<gchristensen>
and the new one seems to hide waaaay more stuff for some reason
<andi->
I have the same issue with reddit which ultimately means I just use it less...
<ajs124>
gchristensen: the approach to the new reddit ui is to just not use it
<ashkitten>
i wish more interfaces were keyboard friendly
<ashkitten>
i don't want to use a mouse
<ajs124>
if and when they discontinue the old one, I'll just use some command-line reader or something. the new ui is literally unusable.
<ashkitten>
honestly i've been looking for a good mastodon cli client
<ashkitten>
but there's not one that supports inline images
<samueldr>
old.reddit.com to the rescue
<samueldr>
if you want to use reddit
<gchristensen>
who here has used zoom recently?
<samueldr>
the moment they remove it, they'll lose at the very least me as a content consumer
<ajs124>
my nuclear fallback for reddit is redreader in an android appvm. that assumes they don't discontiue their API, but if they do that, they're basically dead, anyways.
<gchristensen>
samueldr: yeah! I just found that, so good
<samueldr>
if you have an account it's also an account option
<MichaelRaskin>
gchristensen: I have
<ajs124>
gchristensen: here, me... I need it for university.
<samueldr>
it can sometimes end up being turned back to new reddit, but didn't happen for a while
<pistache>
samueldr: I think they'll lose a lot of users when they remove it
<samueldr>
that new reddit UI is... pretty bad
<samueldr>
pistache: I think so too
<MichaelRaskin>
Three clicks on «no, your app is broken» still work
<pistache>
there are also browser extensions to force the good UI
<samueldr>
but I don't know if it's a majority or a vocal minority
<samueldr>
it may be that they want to get rid of those users anyway
<bqv>
Tbh I refuse to use reddit from anything but a decent android app
<bqv>
Slide is nice, web pages are hideous
<gchristensen>
clarification: anyone using Zoom's app client on Linux?
<ajs124>
Is anyone here using lvmthing or other weird lvm features? And if so are you monitoring them in any way?
<lopsided98>
gchristensen: I use it
<gchristensen>
I was worried about this: $ grep Untrusted ~/.config/zoomus.conf => blockUntrustedSSLCert=false but it turns out to be reasonable
cjpbirkbeck has joined #nixos-chat
andi- has quit [Ping timeout: 244 seconds]
andi- has joined #nixos-chat
waleee-cl has quit [Quit: Connection closed for inactivity]
<ldlework>
I wish I could remember how shell.nix and default.nix fit together
drakonis_ has joined #nixos-chat
drakonis1 has quit [Read error: Connection reset by peer]
drakonis2 has joined #nixos-chat
drakonis_ has quit [Ping timeout: 260 seconds]
drakonis has quit [Quit: WeeChat 2.8]
<colemickens>
bravo to square. I'm on their checkout page and every digit of my CC is doubled when I type it on my mobile keyboard.
maxdevjs has quit [Ping timeout: 272 seconds]
<colemickens>
says "pickup til 10pm" I placed an order, now it says it will be ready at 4:30pm
<colemickens>
I swear how?
<adisbladis>
Yay
<adisbladis>
I let forge-pull run over night
<adisbladis>
Subsequent pulls are much faster
soju has joined #nixos-chat
drakonis2 has quit [Read error: Connection reset by peer]
drakonis_ has joined #nixos-chat
drakonis_ has quit [Read error: Connection reset by peer]
drakonis_ has joined #nixos-chat
cole-h has quit [Quit: Goodbye]
<sphalerite>
colemickens: does backspace also get doubled?
<sphalerite>
eyJhb: you could use vfat with the shortname=lower option to avoid any uppercasing of file/dirnames #LifeProTips
KeiraT has quit [Read error: Connection reset by peer]
ninjin has quit [Write error: Connection reset by peer]
<colemickens>
<sphalerite "colemickens: does backspace also"> That's actually exactly what I did, ha. There was another hitch, but a burger was secured in the end.
KeiraT has joined #nixos-chat
<sphalerite>
colemickens: \o/
ninjin has joined #nixos-chat
parsley936 has joined #nixos-chat
<adisbladis>
Does anyone know if there is some per-stream compressor for pulseaudio?
<adisbladis>
My use case is for calls where the other end is noisy and "trebly"
<srhb>
adisbladis: "per-stream" as opposed to having a separate sink for it?
<adisbladis>
srhb: Yeah
<adisbladis>
I mean, I don't really know what I'm asking here..
<adisbladis>
But I don't want the compressor enabled "globally"
__monty__ has joined #nixos-chat
<adisbladis>
I guess that's already possible by having a separate sink
<adisbladis>
And just redirecting the browser to that sink with the compressor
<srhb>
I can't think of another way to do it, yeah.
<adisbladis>
Duh
<adisbladis>
Alright, disregard me
<adisbladis>
I'm obviously an idiot :)
<srhb>
Never. :-)
<adisbladis>
Now maybe I can stop having headaches from all the high treble \o/
<srhb>
Yeah that does sound horrible. :P
<srhb>
I have some coworkers who are like reeeeaaaally bad at making sure their sound quality is at least decent with all the self-isolation going on
<adisbladis>
All of my headphones are completely unforgiving
<srhb>
Sudden burst eardrums is a phenomenon these days.
<adisbladis>
Even the slightest noise in the mix makes listening to a song unbearable
<adisbladis>
OTOH listening to a good mix is _fantastic_
* adisbladis
outs himself as an audiophile
<adisbladis>
srhb: It's pretty much impossible to get decent quality with laptop mics...
<adisbladis>
Supposedly my laptops built-in mic does noise cancellation, but it's pretty bad
<srhb>
adisbladis: :D Yeah, no kidding. I need a better mic, to be honest, I am one of the sinners in this regard
<srhb>
It's _just_ good enough that I don't feel terrible about it.
<adisbladis>
I got a lav mic pretty recently
<adisbladis>
I'm really happy with it :)
<srhb>
I've never tried any of their products, gotta check it out.
<adisbladis>
Rode Smartlav+
<adisbladis>
RØDE =)
<srhb>
Oh actually I've used them for drum kits, but that doesn't count
<srhb>
Their overheads are crazy.
<srhb>
I tried bringing home a dpa headset once (stage mic) that was kind of nice. :P
<gchristensen>
sphalerite: that response is good & thoroguh, though leaves me wondering if another tool than borgbackup could do essentially the same, but with the insight of what changed
<{^_^}>
borgbackup/borg#5194 (by lheckemann, 6 hours ago, closed): Wishlist/suggestion: external source of change information
leah2 has quit [Ping timeout: 260 seconds]
<__monty__>
Thanks
leah2 has joined #nixos-chat
cole-h has joined #nixos-chat
<cole-h>
sphalerite: Too bad, that would have been really cool (re: your borgbackup + `zfs diff` issue)
<cole-h>
Anybody who uses GPG (not a security key), what's the best way to transfer your secret key to your new setup? (e.g. Arch -> NixOS as I'm doing)
<cole-h>
Next question: what is the fastest way to copy an SSD to an HDD? I'm hesitant to use dd because apparently both disks need to be identical (or so I read from an SE answer).
<__monty__>
dd has no advantage to cp or rsync anyway.
<__monty__>
No reason to use such an old tool unless you *really* have a reason.
<__monty__>
Most iso copying tutorials just cargocult it.
<cole-h>
Oh OK. I've also heard `cat` is better.
<emily>
does cp actually set a reasonable blocksize by default
<emily>
cat definitely doesn't
<__monty__>
I'd just use rsync because it's what I'm used to, has nice progress and continuing interrupted transfers. I don't really care about any overhead.
<__monty__>
I'm sure it uses a more reasonable blocksize than most people pass to dd : )
<cole-h>
I more want a full clone, partitions and all
<cole-h>
So if anything goes wrong I can just boot to that disk without any hassle
<__monty__>
Ah
<cole-h>
For context: I'm doing a reinstall of NixOS on ZFS to test my knowledge soon, and then I'll do it for real on my SSD -- but I want its contents available to me until I'm comfortable I've migrated *everything*
drakonis1 has joined #nixos-chat
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 246 seconds]
<pie_>
you can copy block devices with cp?
<pie_>
oh
<pie_>
ou mean copying the file system contents
<pie_>
cole-h: i think the main size issue is if the target is too small oops you just truncated your partitions, if its bigger you have to resize them
<cole-h>
Disks are ~the same size.
<cole-h>
I haven't filled up my SSD, so no matter how close the HDD actually is to 1TB, it's more than 900GB :P
<pie_>
yeah but idk about block allocations
<cole-h>
True.
<pie_>
just beause its not full doesnt mean its filled from the beginning
<cole-h>
Well, I do have a backup thanks to borg, so I could just mount that if I ever need to grab something....
<__monty__>
pie_: According to the man page you should be able to with -R. Don't think you can both read a block device *and* write one.
<pie_>
silently truncating stuff does not generally sound like a good idea unless you can do something like zfs scrub
<cole-h>
I think I'll just `cat < /dev/whatever > /otherdisk/dev-whatever.img`
<pie_>
<emily> does cp actually set a reasonable blocksize by default
<pie_>
<emily> cat definitely doesn't
<emily>
if you use < > then it's definitely not going to do the right thing
<qyliss>
I think I would use ddrescue for this
<emily>
__monty__: re "more reasonable block size": I just checked and cat defaults to 128 KiB which is terrible
<pie_>
my default would be dd, but idk about the details
<cole-h>
Well, I'm getting conflicting messages here... Apparently both dd and cat are bad :')
<emily>
it looks like it also looks up a per-FS default but I don't know if that would work for block devices and especially not if you < > them rather than specifying the path
<pie_>
not sure what SO answer you were reading but this sounds like something there should be at least one good answer for
<qyliss>
ddrescue is designed for exactly this operation
<qyliss>
at least AIUI
<emily>
dd is fine, albeit awkward, I don't know what __monty__ is talking about. ddrescue is also fine
<qyliss>
dd has a bad block size default for modern drives, idk about ddrescue
<emily>
and as far as file copying programs go (which, while they might be fine to copy block devices, it's not quite the same task really), cp is not a very good file copying program, rsync is much better
<pie_>
qyliss: what bs *sould* be used?
<emily>
git outperforms cp sometimes...
<pie_>
im not comfortable with cp for this because i dont know about setting up the flags properly to copy things like metadata correctly
<qyliss>
pie_: 4K or 8K depending on drive
<pie_>
permissions, extra attrs, etc
<emily>
you usually want bs > actual block size because overhead
<emily>
I just set bs=1M and don't think about it
<qyliss>
oh, then what emily says :)
<emily>
sometimes even higher can be faster iirc
<pie_>
yeah i usually do something like 1M or 10M or whatever and that seems to be ok
<qyliss>
But you definitely do not want 512B, which is the dd default
<qyliss>
idk what ddrescue does because it seems to pick a sensible value and I've never had to think about it
<emily>
I'm sure someone has written a fancy dd replacement in Rust that automatically picks optimal block sizes and has reasonable CLI arguments and stuff
<emily>
maybe it's even in nixpkgs
<pie_>
qyliss: ill consider using ddrescue next time
<emily>
yeah, I don't think `cat` will do the right thing:
<ajs124>
hm... this cp talk makes we wonder if we still carry this patch against coreutils to use copy_file_range
<pie_>
emily: ooh
avn has quit [Ping timeout: 256 seconds]
<ajs124>
ah, no. it was only in another project. that's a free protip, btw: if you want to do COW copies on XFS across mount namespaces, use the copy_file_range syscall.
avn has joined #nixos-chat
<emily>
cp not defaulting to --reflink is sad
<__monty__>
emily: Don't disks usually lie about blocksizes? Re, the stat call. I've seen that said here several times.
<ajs124>
emily: yup. and their --reflink implementation iirc doesn't even work on XFS -.-
<emily>
yes, most report as 512, a few report as 4k, almost nothing admits to being 8k
Jackneill has joined #nixos-chat
<emily>
iirc even spinning rust has some kind of internal >512-byte block thing going on these days
* ajs124
sitting here with his SMR HDDs which have 32MB regions or something stupidly large
<emily>
zfs also doesn't support --reflink at all
<ajs124>
yup, but that's a filesystem thing (iirc). cp can't do anything about that.
<emily>
yeah
<__monty__>
> bcp.meta.description
<{^_^}>
undefined variable 'bcp' at (string):312:1
<__monty__>
,locate bin bcp
<{^_^}>
Couldn't find in any packages
<emily>
it's not packaged
<cole-h>
Thanks for all the help guys. Looks like `ddrescue` is indeed what I want (or rather, what I will use).
<__monty__>
Does bcp do any cleverness to figure out a good blocksize?
waleee-cl has quit [Quit: Connection closed for inactivity]
<pie_>
i didnt read the source but it just looks like a two page rust file
<__monty__>
I think it uses rust's write_all? I have no idea whether that figures out a blocksize though. Don't really read rust.
xd1le has quit [Ping timeout: 260 seconds]
<emily>
it defaults to 1M and lets you specify one
<emily>
as per the HEAD commit
<__monty__>
Thank you.
<emily>
it's a reasonable default since the ideal block size relates a lot more to crappy kernel/buffering/libc/... overhead than it does to the underlying block size
<ashkitten>
i just always use `dd [...] bs=4M oflag=sync status=progress` and have never had any issues
<emily>
looks like 1M is at least far from the worst choice, though the test conditions seem to use an actual FS
xd1le has joined #nixos-chat
* ornxka
always thought bs was supposed to be your device block size...
<ornxka>
ive been using bs=4096 for years
<ashkitten>
if you suspect the drive may be dying then use ddrescue, but i don't know why else you would need it
<ashkitten>
it doesn't need to be your device block size, it does need to be a multiple of your device block size
<emily>
the good news is that powers of two are really good at being multiples of each other
<ashkitten>
trying to write smaller than your device's native block size results in write amplification
<emily>
it's like they trained their whole life for it
<ornxka>
lol
<__monty__>
Write amplification sounds great! How strong can we make these writes?
<ashkitten>
lol
<ashkitten>
write amplification, for anyone who doesn't know, is when you write something in such a way that the underlying device has to read, copy, and write it back as a full block
<ashkitten>
it can also happen when writing records smaller than the stripe size on say, a raid5 array
<__monty__>
Sounds like a bad time. Wouldn't disks cache writes in hopes of gathering a full block though?
<ashkitten>
that would be clever, but not to my knowledge
<ashkitten>
i believe normally the kernel buffers writes like that, though[citation needed] so it's not an issue for most programs that don't use block devices directly
xd1le has quit [Remote host closed the connection]
xd1le has joined #nixos-chat
KeiraT has quit [Ping timeout: 240 seconds]
KeiraT has joined #nixos-chat
xd1le has quit [Read error: Connection reset by peer]
<colemickens>
(tbf, I get annoyed when apps don't store config files according to xdg dirs)
<emily>
life would really be a lot easier in a lot of ways if nix had used /var/lib/nix or something from the start
<emily>
though it's tricky because it's half /var, half /usr, half ...
<gchristensen>
as if FHS was a good idea
<emily>
picking /var/lib sounds great because it's guaranteed mutable, but then apple introduce a policy banning execution from /var, because that's for mutable data and executable code goes in /usr
<cransom>
it's not like the sprawl between /bin, /usr/bin/, /usr/local/bin, then having /opt was perfect. even when i was heavily in the freebsd side, i was swearing at linux people for assuming and hardcoding a /bin/bash
<ornxka>
i make directories in / all the time, whenever i make a container i just put the bind mounts in there
<ornxka>
it seems absurd that making a top-level directory should be so difficult
<__monty__>
Hmm, is the 128 KiB blocksize of cp tied to kernel buffers maybe?
<emily>
cransom: tbf, even mainstream linux distros have moved to linking {,/usr}/{,s}bin together
<__monty__>
What I don't understand is why users don't get to override apple's constraints. It's a fine default for most people but why take away options completely?
<emily>
nixos still having a sbin separation surprised me
<gchristensen>
emily: we do?
<cransom>
any sbin separation i usually see is because installers still do it. i usually see sbin symlinks to bin if at all
<emily>
gchristensen: diff -r /run/current-system/sw/{,s}bin gives results at least
<gchristensen>
huh
<armin>
i actually had to use systemd timers today for the first time.
<armin>
wow is that terrible when you spend an hour or so trying to understand what's going on, entirely conscious that what you want to do would have taken less than 1min with a cron job.
<gchristensen>
yeah it is hard to learn new things sometimes
<__monty__>
That's not been my experience. Timers are a lot less write-only.
<bqv>
there's a compatibility layer, isn't there?
<bqv>
systemd-cron?
<armin>
__monty__: oh i totally disagree with that one.
<armin>
yup there is, on various distributions at least.
<gchristensen>
<3 timers
<__monty__>
For me timers are a lot more readable even though they can be a lot more complex.
<gchristensen>
way more transparenti mo
<cransom>
i fully embraced timers when i needed to schedule something more than every minute.
<gchristensen>
hah
<cransom>
and output/environment did sane things with systemd that i didn't have to fix $PATH and friends.
<gchristensen>
yeah
<bqv>
to me, timers are the same as every other aspect of systemd. i theoretically despise it, but it's so much more practically convenient that i never use anything else
<gchristensen>
and the logs, and the next / left / last / passed / unit / activates output is all so good
<cransom>
i can't really hate systemd. it was different than i'd done in the past, but... not dealing with logrotate because journald existed. or specific cron tweaks so you'd get mail when it failed. hacking up init scripts? no thank you.
<gchristensen>
+1
<__monty__>
cransom: Do you rely on checking status now? Or do you have mail warnings setup?
<__monty__>
Thanks. Maybe I should look into free monitoring platform offerings.
<cransom>
pushover is just message delivery. i could probably nextcloud or homeassistant things easily too, but since it's all at home, i'd rather have it hosted.
<cransom>
i used to do the more exhaustive home monitoring, but i spent more time setting up and maintaining then ever saving by having monitoring.
<gchristensen>
oh cool cransom
<__monty__>
I'd just like to be able to get a quick overview of backup health without getting pinged for every failure.
<gchristensen>
I collect the age of all my ZFS snapshots on m ybackup host and alarm if any dataset's latest snapshot is between 1 day old and less than 1 week old
<cransom>
if you have a robust backup environment, you probably want a more robust monitor than a systemd unit failing. it works for me though. if i was doing something at work, i'd be using the real $monitoring and it woudl go to pagerduty, etc etc.
<__monty__>
Can't describe it as robust. But I want something between no warnings at all and a warning every 10 minutes.
<__monty__>
gchristensen: That's some sort of monitoring service, right?
<gchristensen>
prometheus and alertmanager with pushover
wildtrees has joined #nixos-chat
<gchristensen>
only thing that cost was a one-time fee to buy the pushover app
<cole-h>
Alright my friends. It is time to reinstall NixOS and test how robust my setup is...
<__monty__>
Nice knowing you.
<__monty__>
; )
<cole-h>
:D
<__monty__>
gchristensen: Thanks, might look into it.
<cransom>
the few bucks i spent on pushover has been worth it. hooked printer status into it, transmission. nice notes on the phone that tell me about things i want to know about.
<gchristensen>
same
<gchristensen>
little helper scirpts on various systems like `doathing ; tellme`
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos-chat
cole-h has quit [Client Quit]
<samueldr>
:| syncthing has a rough edge in that even if you configure one folder to have a lower minimum free space required, the global setting applies... and nowhere does it explain why it's not syncing until I manually "rescanned"
<samueldr>
1% of 1TB is quite a lot of space for a folder of text notes
waleee-cl has joined #nixos-chat
cole-h has joined #nixos-chat
<__monty__>
You're just not taking enough notes!
<cransom>
use a bigger font
cole-h has quit [Client Quit]
<gchristensen>
I don't think I can, my font already takes like 3h to build and 2+GB
<samueldr>
that's comically bad
<samueldr>
(though I'm sure it's not comic sans, but I wanted the pun to be out there)
<gchristensen>
:D
<gchristensen>
comic sans takes no building. this is probabl ylike twemoji or osmething
<samueldr>
yeah
<gchristensen>
though I did have to pay for comic mono
<samueldr>
and some say crime doesn't pay
<gchristensen>
the crime being creating comic mono/
<samueldr>
interpretation left to the reader
<gchristensen>
the crime being creating comic mono.
clever has quit [Ping timeout: 240 seconds]
cole-h has joined #nixos-chat
cole-h has quit [Client Quit]
clever has joined #nixos-chat
drakonis_ has quit [Ping timeout: 256 seconds]
drakonis_ has joined #nixos-chat
<gchristensen>
I wish there was a postgres cli option like "please don't let me change anything"
<MichaelRaskin>
This sounds like a role with readonly access?
<gchristensen>
yeah but also
<MichaelRaskin>
I agree that .* needs better and more convenient sandboxing.
<clever>
i think mysql also has a cli flag, to stop you from doing an update/delete without a where clause
<clever>
so you cant make a typo and blast an entire table
<gchristensen>
nice
<clever>
and the scary thing with sql, `update tbl set a=1` is how you start something with a quere clause
<gchristensen>
in general it'd be nice to have a mode which said like "don't let me do damage" and it limited my queries to not be insane or destructie
<clever>
and you could accidentally hit enter maybe, and it take effect
<gchristensen>
"only give a given query like 2 seconds of cpu time"
<MichaelRaskin>
Wait what??
<MichaelRaskin>
Enter would not work without semicolon, no?
<joepie91>
(it wouldn't in the postgres CLI)
<clever>
MichaelRaskin: depends on the mode i think, and ; is fairly close to enter
<clever>
its one cat away from not having any more data :P
<MichaelRaskin>
I think I have seen some cats where no syntactical change in SQL would save data withing the reach of its claws
<joepie91>
to be fair, almost everything is one cat away from not having any more data
<joepie91>
I don't think cats are a reasonable threat to be expected to model against
<joepie91>
:P
<__monty__>
Only allowing non-destructive updates, i.e., ones that are invertible functions would be cool.
* gchristensen
is now known as PokaYokeShill
dadada_ has quit [Quit: WeeChat 2.7.1]
dadada_ has joined #nixos-chat
cole-h has joined #nixos-chat
cole-h has quit [Client Quit]
cole-h has joined #nixos-chat
cole-h has quit [Ping timeout: 260 seconds]
cole-h has joined #nixos-chat
<sphalerite>
well, this is a rabbit hole I did not expect to end up in: making changes to coreboot so that it supports the itb images generated by u-boot.
cole-h has quit [Ping timeout: 265 seconds]
<samueldr>
and how's it going?
<sphalerite>
samueldr: better than I would have thought, though I haven't got u-boot loading yet :D
<eyJhb>
Finally, the day is almost done. Took 6 hours to change the breaks on my car, because of rust. Broke a T55 (torque), because of that... :(
<eyJhb>
But first brake change, and I am not dead, for now at least
<eyJhb>
talyz: but now I need you again! I will ping you again tomorrow. I think I am using the persistence module wrong :p
<__monty__>
Are you sure this has saved you any money compared to a garage?
<eyJhb>
__monty__: 165 DKK! :p
<eyJhb>
But that was because I needed all the tools. And I need to factor in my wage at 0 DKK
<eyJhb>
But always worth it, I like being able to do such things myself
<samueldr>
I can only think about donkey kong kash
<eyJhb>
samueldr: how would 165 DKK then equal in donkey kong kash? :p
__monty__ has quit [Quit: leaving]
<MichaelRaskin>
> DKK 165
<{^_^}>
"571726.650000 VND"
<MichaelRaskin>
> EUR 16.5
<{^_^}>
"16.500000 EUR = 123.750000 DKK"
<MichaelRaskin>
samueldr: it is Danish krone, ~ 7.5 DKK per 1 EUR
<samueldr>
yep, I knew what it stood for, though didn't for the exchange rate
parsley936 has quit [Remote host closed the connection]
parsley936 has joined #nixos-chat
cole-h has joined #nixos-chat
cole-h_ has joined #nixos-chat
cole-h has quit [Quit: Goodbye]
cole-h_ has quit [Client Quit]
cole-h has joined #nixos-chat
ninjin has quit [Ping timeout: 240 seconds]
ldlework has quit [Ping timeout: 272 seconds]
ldlework has joined #nixos-chat
ninjin has joined #nixos-chat
<cole-h>
Good news: system is mostly operational after a full wipe and reinstall. Bad news: "mostly".
<gchristensen>
oh?
<cole-h>
jack audio isn't working.
<ldlework>
cole-h: did you run that command
<ldlework>
that you have to manually run
<ldlework>
pactl load-module module-jack-sink
<ldlework>
not sure how to best automate that
<cole-h>
No, but I didn't have to on my last install. Cadence does everything for me (or did)
<ornxka>
eyJhb: i read "took 6 hours to..., because of rust" and immediately thought "wow fighting the borrow checker isnt THAT bad, is it?"
<sphalerite>
YAY I HAVE A U-BOOT PROMPT
<ldlework>
cole-h: did you try it?
<cole-h>
ldlework: How did you enable jack? Just `services.jack.jackd.enable`?
<sphalerite>
samueldr: well, you can always flash back the original firmware
cole-h has joined #nixos-chat
<hexa->
lots of 500 tonight
<samueldr>
sphalerite: yes, though makes it painful to test :)
<hexa->
http 500 that is
<samueldr>
though, from my experience, it looks like most of the bootloader-specifics are well abstracted, so it might not be such a bad thing to get a u-boot instead
<samueldr>
and cross fingers for depthcharge
<samueldr>
or dig out the chromebit and use it for a depthcharge target
<samueldr>
(armv7l though)
<cole-h>
ldlework: One thing I dislike about this approach is that you can't use `pactl set-sink-volume` to change the volume...
<sphalerite>
samueldr: was your gru without CCD?
<samueldr>
with
<samueldr>
really just for the fact that it would be annoying to go back and forth
<sphalerite>
then I don't see what's so painful? :)
<sphalerite>
ah right
<samueldr>
I have to *do something*
<sphalerite>
hm, it should be flashable on-device via userspace as well…
<samueldr>
why do you think I became a developer? do I don't have to do *anything*
<samueldr>
so I*
<sphalerite>
lol
<ldlework>
cole-h: i literally don't know shit about this, and if you come up with better configuration, please send it over :)
<cole-h>
Roger. Let's see if I can't figure out why cadence used to work but doesn't anymore...
<sphalerite>
samueldr: but you could set the eMMC partitioning up such that both u-boot and depthcharge will work, no?
<sphalerite>
then it doesn't matter which is currently active, unless you want to work with a specific one
<samueldr>
I guess so, but both would need a different stage-0 (stage-1), no?
<samueldr>
well, yes with the current expectations in how mobile nixos is setup
<samueldr>
but maybe u-boot can boot that depthcharge partition?
<sphalerite>
hm, tricky I think
<sphalerite>
maybe chainloading depthcharge would be easier
<sphalerite>
but at that point, why u-boot at all? :p
<samueldr>
though at that point I'd be flashing depthcharge, and using sd card to test lol
<samueldr>
sphalerite: to remove that annoying menu at boot!
<samueldr>
and possibly use BMP support to show a full screen logo
<samueldr>
because pretties
<sphalerite>
\o/
<sphalerite>
though I haven't got any output to the monitor with u-boot currently.
<samueldr>
ah, so BMP support later :)
<emily>
hm, so does mobile-nixos cover chromebooks too?
<samueldr>
but no menu, and it would basically boot just like the pbp does
<samueldr>
emily: yes, because of the two "gru" tablets
<samueldr>
well, three?
<sphalerite>
samueldr: now checking if that's just a question of config :)
<samueldr>
anyway, there are Rockchip OP-1 (RK3399) tablets with chromeos, actual tablets, not 2-in-1
<samueldr>
and I want to make it a first class citizen where there is no keyboards
<emily>
ah
<samueldr>
I also have a cheap atom-based tablet (which can dock in a keyboard dock) which I intend to look into supporting "well"
<samueldr>
and I don't know if it'll only be UEFI, or through whatever weird bizarro android thingy intel was using
<emily>
I was basically wondering "is there value in basing a Pixelbook coreboot/NixOS/... setup on mobile-nixos" and I guess the answer is "only if I care about tablet-y stuff in the graphical environment", or?
<emily>
(I hvaen't actually played with mobile-nixos so I'm not sure exactly what it gives you)
<samueldr>
ah, stage-2 of mobile nixos is basically mobile nixos, so "no"
<samueldr>
it's more about the boot chain
<samueldr>
and abstraction over quirky things
<samueldr>
on an x86_64 system, I would rather use mrchrombox's UEFI and boot normal things
<samueldr>
BUT, I intend to add uefi support in mobile nixos for dogfooding reasons
<samueldr>
(and those atom tablets)
<samueldr>
so, it's kinda all over the place :)
<emily>
ah, ok
<emily>
I think I'd rather avoid UEFI.
<samueldr>
but if you care about tablet-y stuff in the graphical environment, mobile nixos is not (yet) better than nixos, and all improvements should trickle back up
<emily>
I value the Pixelbook boot speed a lot and also Secure Boot is its own huge hassle already.
<emily>
right
<emily>
so coreboot/depthcharge type stuff is in-scope then?
<cole-h>
ldlework: I think I got it! Basically everything you have except place this at `~/.pulse/default.pa`: https://paste.rs/sXY
<samueldr>
yeah, I want to support booting its stage-1 through depthcharge, or any coreboot target that I can get it to
<samueldr>
I already abstract away a chunk of different bootloader configs
<ldlework>
cole-h: what's that do?
<ldlework>
cole-h: also if you edited my gist and sent it back, that'd be pretty cool :)
<cole-h>
ldlework: Automates the `pactl load-module` for you ;)
<ldlework>
ah nice
<cole-h>
ldlework: Not logged into GH right now
<ldlework>
cole-h: should this be part of nixos?
<ldlework>
no doubt there's an /etc/ place where it could read it too?
<ldlework>
or whatever
<cole-h>
Yeah, there is.
<ldlework>
cole-h: i bet you didn't find the answer in the nixos documentation :)
<cole-h>
Definitely not lmao
<ldlework>
:P
<ldlework>
so this is probably a source of confusion for new users
<ldlework>
"audio doesn't even work"
<cole-h>
I used cadence on my Arch system (and my previous NixOS install; really weird it stopped working...)
<ldlework>
reminding them of slackware in the 90's
<cole-h>
And I noticed that it starts pulseaudio with a `--file=/nix/store/....-cadence/share/cadence/pulse2jack/play+rec.pa`
<ldlework>
there should be a "desktop experience" team
<cole-h>
So I just copied that file into my default.pa and it worked!
<emily>
samueldr: ok, maybe i'll take a look at mobile-nixos then ^^
<cole-h>
Ahhhhh, that feels so good.
<ldlework>
cole-h: jack is so fun
<ldlework>
linux has so many quirky audio apps
<samueldr>
emily: if you do and are confused by anything, not everything's obvious or well documented, don't hesitate and ask
<ldlework>
so if you manage to get jack to actually work, downloading random ones and making them control each other is loads of fun
<cole-h>
tbh I like it because it makes everything quieter. With just pulse, I'd have to turn volume down to like 10 or 15 %. Now I can jam at 50+% (depending on the source volume)
<emily>
samueldr: does the boot/stage1 inherently have the fancy graphical things or can you strip it down to be more like a "traditional" nixos initrd? not that I mind the former, just curious since I'll probably want to do some fairly custom things
<samueldr>
there is no strong ties to graphics, but as I support targets without virtual consoles, they are the main focus for input/output
<samueldr>
but everything is built in the mindset that it should be switchable with anything, so e.g. you would run openssh or dropbear and ssh into the (stage-0) initrd to select a generation to kexec into
<samueldr>
or, I say switchable, but here it would likely run in parallel with a graphical one
<aleph->
I should go set up mobile-nixOS on my pinephone finally
<samueldr>
or, more useful example, serial
<samueldr>
not implemented, but things are written in a way I'm not in a dead-end
<sphalerite>
SO EXCITED
<samueldr>
hm?
<sphalerite>
u-boot loaded the kernel and initramfs etc successfully
<sphalerite>
missing piece: mounting the right root filesystem.
<samueldr>
oh, that's, like, your opinion, man
<sphalerite>
so now I'm waiting impatiently for my next initramfs to build
<samueldr>
:)
<samueldr>
so in theory you can use the extlinux thing as sd_image does, for your chromebook?
<sphalerite>
yes that's exactly what I'm doing
<samueldr>
nice, though without gfx it's as useful as it is on the pbp
<samueldr>
not very
<sphalerite>
I `nixos-install`ed to a USB stick and am plugging that into the chromebook
<sphalerite>
graphics should probably work as well, but the current u-boot image doens't have graphics support
<sphalerite>
will need to try that, but my first goal is to boot nixos :)
<samueldr>
yeah, and that is great
<sphalerite>
the nixos-install was a bit roundabout though because I was too lazy to go upstairs and plug the USB stick into my nanopi :D
* sphalerite
waits impatiently for USB stick to unmount
<sphalerite>
(could you tell I'm excited?)
<samueldr>
it's pouring from you
KeiraT has quit [Remote host closed the connection]
<sphalerite>
it's booooteeeeeeed
parsley936 has quit [Remote host closed the connection]
KeiraT has joined #nixos-chat
<infinisil>
What if {^_^} could kick people if some number of trusted nicks voted for it