<elvishjerricco>
How do platforms like iOS and android wake the device for internet-delivered notifications? And how do they wake instantly compared to linux taking a few seconds to wake from suspend?
<samueldr>
it's a bit more involved than "wake from suspend"
<samueldr>
in addition, the phones don't really suspend, I don't really get the details, but most of the hardware is put in suspend, and the cpu left in the lower power modes, with obviously modems working still, as needed
<samueldr>
(unless things ask, through wakelocks or whatever replaces them now, to continue running)
<samueldr>
all of this is why apps are strongly suggested to use "GMS", for push notifications (or forced to use the apple equivalent on iOS)
<samueldr>
this allows the system to have a single app waiting on push notifications, which in turn can alert the proper apps
<samueldr>
rather than have each apps build their own
<elvishjerricco>
Huh. So in effect, that process that listens for notifications is one of a very select few processes running while the phone is put to sleep, and the system isn't actually suspended?
<samueldr>
AFAIUI yes
<samueldr>
and anyway, the biggest power consumer on a phone is the display, then the GPU
<samueldr>
(assuming you're not doing CPU intensive tasks)
<samueldr>
so having the display off is already helping a lot
<samueldr>
pushing photons to your retinas and everywhere else around is intense!
<elvishjerricco>
Interesting... I'm curious if operating systems like mobile NixOS are ever going to get that same sort of behavior. Seems rather necessary for an acceptable smartphone experience.
<samueldr>
I guess so, the nice thing is that it's less of a feature of the "distro", but more of the "desktop environment"
<samueldr>
but will need integration and knowledge at the distro level
<samueldr>
I still don't *know* enough to talk about it, but yeah, I assume it'll need to be tackled
<samueldr>
one "nice" thing is that AFAIUI the bad suspend/resume cycle on x86_64 are x86_64 things
<samueldr>
already going into "full" suspend on the pinephone (which still gets woken by modem events!) is not bad
<lovesegfault>
gchristensen: you use prometheus with NixOS, right?
<supersandro2000>
GSM is basically the one proxy which takes notifications for all apps
<samueldr>
GMS* :)
<samueldr>
since GSM is also a (n obsolete) celluar thing it could get confusing
<supersandro2000>
and since google can optimize the delivery and combine it, is is efficient
<samueldr>
on android it's not mandatory
<samueldr>
on apple their equivalent is
<samueldr>
[20:15:48] <samueldr> all of this is why apps are strongly suggested to use "GMS", for push notifications (or forced to use the apple equivalent on iOS)
<samueldr>
APNS
<supersandro2000>
isn't it google cloud messaging GCM?
<{^_^}>
Ping for space stuff (edit this command to add yourself, see ",help"): infinisil Taneb ldlework etu philipp[m] eyJhb gchristensen __red__ red red[evilred] risson
<eyJhb>
Qo
<eyJhb>
I know I will miss it as always.
<eyJhb>
Need to develop a website with a big red button which will make my phone go into alarm mode. And no, setting an alarm just isn't the same.
<srhb>
eyJhb: Just keep the tab open and set the notification button :P
<eyJhb>
I will try to do that. But my tabs just seem to go missing in the mountain of other tabs, which sometimes are cleaned up
<eyJhb>
Placing it somewhat safely as the first tab!
<srhb>
eyJhb: I pin tabs so I don't accidentally kill them :P
<srhb>
Modern browser features, they're useful!
<eyJhb>
:o srhb teaching me stuff and things
<eyJhb>
Didn't even know I had pinning
<eyJhb>
(and don't even know what it does *presses it*)
<talyz>
eyJhb: [impermanence] do you still have issues with the bind mounts on your laptop?
<srhb>
eyJhb: Most browsers will ignore whatever hotkey you have to kill tabs and go to the next tab instead
<eyJhb>
eyJhb: I pinned the old version, I haven't tried any newer since we last spoke about it. Does the now inherit permissions?
<eyJhb>
Seems like ctrl+w will still kill it in chromium srhb :p
<srhb>
eyJhb: Oh, meh, I thought it was universal. firefox does it at least.
<eyJhb>
Yup. Chromium will kill the **** out of it, ctrl+w
<eyJhb>
Surfingkeys as well :p
neeasade has joined #nixos-chat
<supersandro2000>
I am starting to understand nspawn and networkd
<eyJhb>
What are you doing with it supersandro2000 ?
<supersandro2000>
I build a nixos containers container on a Ubunut host with working private networking
<supersandro2000>
I noticed that it got random subnets which makes networking a bit hard
<supersandro2000>
and this is the nspawn default behavior for whatever reason
<talyz>
eyJhb: I added an option to turn on the fuse "allow other" flag. You need to allow it globally in fuse through the NixOS programs.fuse.userAllowOther option first, though.
<supersandro2000>
and I thought all the time unbound is as fast as using a public DNS because I didn't use... derp.
linear_cannon has joined #nixos-chat
<eyJhb>
talyz: But that would still screw up my NOEXEC, etc. as far as I can tell :D
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos-chat
<talyz>
eyJhb: Ah, you have noexec set on your persistent storage?
<eyJhb>
talyz: Yuuup :D
<eyJhb>
And nosuid stuff :D
Synthetica has joined #nixos-chat
<eyJhb>
It is actually working quite well, don't really miss it
<eyJhb>
18 minutes! :o
<aaronjanse>
Thank you eyJhb !
<MichaelRaskin>
Ingenuity: #WhetherHop
<eyJhb>
aaronjanse: Wait, what did I do? :p
<aaronjanse>
You reminded me. I almost forgot
<aaronjanse>
^ eyJhb
<eyJhb>
Oh :D
<eyJhb>
I almost forgot myself...
<eyJhb>
I am like 10/10 in doing that :p
<aaronjanse>
I've regretted every time I've missed a space thing
<aaronjanse>
So I'm glad I'm watching this one
<eyJhb>
It should be about now, right?
<aaronjanse>
I think so
<eyJhb>
Also, I don't suppose they had a cam on it :(
<aaronjanse>
They say they have a camera on it
<eyJhb>
:o
<eyJhb>
Coooooool
<srhb>
It should have been 15m ago if you ask me, but apparently they need to explain to us, in advance, what's going to possibly happen, multiple times, and explain to us that they are, in fact, a team.
<eyJhb>
srhb: What, they are a TEAM?!
<srhb>
I was shocked too. Glad that they're explaining things so clearly.
<eyJhb>
Also, they are getting data.
<eyJhb>
just.. Just so you know srhb .
<philipp[m]>
I always have to think that those webcasts are designed to cater to congresspeople and senators that might watch.
<eyJhb>
People who might have money and can do funding :D
<srhb>
Ideally the helicopter can move around :o
<srhb>
I'll stop myself now.
<eyJhb>
It is a little pros and cons
<eyJhb>
I would have like to have "seen" it fly at the start, and then they could explain what should/what did happen
<aaronjanse>
Lmaooo it's an altitude graph
<eyJhb>
Sounds like no video :(
<eyJhb>
inb4 they are faking it! \s :D
<aaronjanse>
It'll be like watching a stock market ticker
<eyJhb>
GME to Mars!
<aaronjanse>
Ahaha
<aaronjanse>
Hopefully there's some background music with drums
<aaronjanse>
Like some disney amusement park ride
<eyJhb>
They are really paddling atm.
<aaronjanse>
Wow it doesn't look like 4 ft in the photos
<aaronjanse>
The props
<aaronjanse>
*span I think?
<aaronjanse>
YAY!
<eyJhb>
I have no reference for inches
<eyJhb>
Something is happening!
<aaronjanse>
2.2 meters
<eyJhb>
1.2 meters*?
<eyJhb>
4 feet in meters, right?
<philipp[m]>
I hope they are running nixos.
<aaronjanse>
The suspense
<aaronjanse>
NO SOCIAL MEDIA QUESTIONS
<srhb>
They're not, but they _are_ in fact running a more or less generic Linux, which I think is for the first time ever?
<{^_^}>
nasa/fprime#478 (by RichiH, 7 minutes ago, closed): Make a helicopter fly on Mars
<eyJhb>
Lol :D
<Synthetica>
If we don't even have F' in nixpkgs, how do we expect space helicopters to run NixOS D:
<MichaelRaskin>
Tearing up contigency plans on-air looks like a Perseverance meme now
<srk>
eyJhb: I don't think it itself costs 80$ mil :)
<eyJhb>
`NASA has invested about US$80 million to build Ingenuity and about US$5 million to operate the helicopter.`
<Synthetica>
I assume most of that is r&d not materials
<eyJhb>
MichaelRaskin: Tearing up access codes
<srk>
well with all the research
<eyJhb>
Fuck.
<ashkitten>
the video is very low framerate which is why it looks so jerky and weird i guess. was same with the video of the spin-up test
<ashkitten>
this is so cool though. first powered aircraft on another planet
<ashkitten>
im glad i stayed up for it
<MichaelRaskin>
eyJhb: nah, access code they actually still need!
<MichaelRaskin>
Synthetica: well, I guess there is also that minor detail of radiation protected components…
<srk>
not many of these on the copter
<srk>
it's mostly off the shelf hardware
<ashkitten>
a qualcomm soc right?
<srk>
yeah, some sparkfun sensors .. :)
<MichaelRaskin>
Oh, even the SoC is standard
<ashkitten>
yep
<MichaelRaskin>
I wonder whether the test room (controllable low pressure one) needed to be built/modified…
<MichaelRaskin>
(or could they reuse some existing facility)
<__monty__>
Because of horizontal translation requirements?
<MichaelRaskin>
Vertical might be a larger issue
<__monty__>
Aren't vacuum chambers usually higher than wide?
<__monty__>
Like, for drop tests and stuff?
<pie_>
MichaelRaskin: theres this one ridiculously big vacuum chamber building
<pie_>
no idea if they used it i just know it exists
<eyJhb>
I think they generally have access/has these champers. And they are usually very giant, and can fit most things.
<eyJhb>
Also
<eyJhb>
Anyone using System76?
<LinuxHackerman>
I mean, they do work with rather large vehicles that need to work in vacuum
genevino has quit [Ping timeout: 248 seconds]
genevino has joined #nixos-chat
neeasade has quit []
<hodapp>
I have a System76 laptop but not on NixOS
<eyJhb>
hodapp: POP OS?
<hodapp>
I think just Ubuntu at the moment
<hodapp>
I didn't set it up myself - it's a work laptop
<eyJhb>
Which one do you have, and do you enjoy it?
<hodapp>
Oryx Pro, and yeah, I've liked it pretty well.
<eyJhb>
Considering instead of doing a Lenovo again. But I really like the on-site-repair of the Lenovo.. And actually seeming like a company that cares about, anything outside of US.
waleee-cl has joined #nixos-chat
<talyz>
eyJhb: no idea if any of that is inherited :)
<supersandro2000>
samueldr: that gitlab bug is intended beauvoir
<supersandro2000>
At least they don't consider it an issue
<eyJhb>
Ohh, so leaking data is intended?
<eyJhb>
Might even be a feature
<supersandro2000>
Well they don't really consider it an issue worthy their big bounty program
<eyJhb>
Well, that might be another thing. But sure as hell is not intended
<supersandro2000>
They also excempt text injection in none critical pages
<MichaelRaskin>
hexa-: re: Linux kernel — sounds like «Github data quality» more than Linux actually not being used on Perseverance and Ingenuity
rajivr has quit [Quit: Connection closed for inactivity]
<drakonis>
supersandro2000: it had those for a while now
<drakonis>
some profiles have badges for having code in the arctic time capsule
<supersandro2000>
they where called badges before
<drakonis>
gamification...
<drakonis>
amazing
endformationage has quit [Ping timeout: 260 seconds]
endformationage has joined #nixos-chat
rj has quit [Ping timeout: 240 seconds]
<infinisil>
Scary: Deleting a folder with name "~", `rm -rf \~`
<MichaelRaskin>
I would use ./~ instead, actually
rj has joined #nixos-chat
<MichaelRaskin>
Also, with echo first!
<gchristensen>
+1
<MichaelRaskin>
(I guess you could also move it to a safe name first, which is easy to revert if needed, then remove a safely-named directory)
<gchristensen>
I used to `touch ~/-i` so I couldn't accidentally rm -rf * in there
<gchristensen>
turns out backups are better anyway
<MichaelRaskin>
I think in _some_ place I had a directory named aaaa which had a lot of nested subfolders
<samueldr>
just use a gui file manager
<gchristensen>
there you go
<MichaelRaskin>
To be fair, mud folder was also for accidental ancestor folder deletion, which is the same risk for GUI and for CLI
<MichaelRaskin>
As for weird names, well, I do not use them, and also it is still faster to write a safe command than to do anything in GUI filemanager
<__monty__>
Can I recommend a TUI file manager instead? Ranger's pretty nice : >
<MichaelRaskin>
If I want a TUI browsable file list, I can just as well ask Vim for it
<__monty__>
Re the gamification, a new badge that can't be earned doesn't sound like a very succesfull gamification strategy.
<__monty__>
Netrw isn't exactly a pleasure to use. I hear good things about dired though.
<MichaelRaskin>
Well, if you announce new badges from time to time with only postfactum announcement of criteria, this might create some weird incentives
<__monty__>
Like shitposting but for contributions?
<__monty__>
Wait, is that "Yay TUIs! \o/" or "Boo GUIs and by extension TUIs?"
<MichaelRaskin>
Former from samueldr and latter from me?
<samueldr>
__monty__: needs to be collapsed by observation of the individual
<samueldr>
and yeah, TUIs are juste as GUIs, many are terrible, many are amazing, most are in-between, hopefully
<samueldr>
(CLIs too)
<samueldr>
UX is not really better in any "realm"
<samueldr>
though CLI has the advantage of having easier cross-application interaction, which is not exactly a per-application UX thing
<hodapp>
bring Win95 GUI back!
<MichaelRaskin>
Leaving aside that even for a given UI structure and a given problem, there is no single better for UX
<hodapp>
it was ugly, functional, and ugly, consistent!
<samueldr>
hodapp: you say ugly, I say baseline appreciable looks, even customizable in a trivial manner
<__monty__>
TUIs tend to fit my workflow better tbh. The universal focus on keyboard input makes life so much better.
<MichaelRaskin>
GUIs can and should provide keyboard flows
<MichaelRaskin>
And some even do!
<hodapp>
samueldr: whatever the case, they put a lot of work into its design and usability
<__monty__>
Wholly agree. There's nothing stopping GUIs from being amazing. Just seems people rarely care about anything past clickable buttons and basic shortcuts, think copy/paste.
<hodapp>
even if Win95 was otherwise a royal cesspool of fail
tomberek has joined #nixos-chat
<MichaelRaskin>
For all my preference for CLI-only and keyboard-only, even I have to admit that this is an impressively deep hole we find ourselves in, when the only way to get a good keyboard flow is to literally make mouse use prohibitively annoying
<hodapp>
__monty__: the thing is, they also devote a whole lot of time to aesthetics, it's just that they rarely seem to care much on actual usability and design
<hodapp>
in the same way that e.g. Apple did with HCI rules in the '80s
<pie_>
__monty__: idk programming guis also seems harder
<MichaelRaskin>
I mean, there _are_ things that are actually _easier_ in CLI than in TUI/GUI without an integrated CLI-lite.
<pie_>
yeah you reallt want all three :C
<MichaelRaskin>
pie_: which it would not have to be, if not for active selection
<__monty__>
Apple's actually pretty good at having consistent behavior throughout the ecosystem though. Shortcuts are very consistent and nearly every text field has readline-ish bindings.
<MichaelRaskin>
Delphi made making GUIs the easy part, and Lazarus is still working (but behind times re: DPI range, that's true)
<hodapp>
__monty__: sadly, a lot of their own design rules went out the window the past 1-2 decades and they focused a lot more on aesthetics
rj has quit [Ping timeout: 240 seconds]
<MichaelRaskin>
Web frontends used to be almost trivial to make
<hodapp>
my pet peeve for the past few years has been how websites and mobile appications have both converged on this faddish 'app' nonsense where discoverability is usually garbage and where it's almost never clear how to effectively do things like undo or back out of an operation
<MichaelRaskin>
Then aesthetics comes, yeah
<pie_>
MichaelRaskin: keeps making me want to tr ydelphi
<__monty__>
hodapp: Maybe? There's definite focus on aesthetics but iOS UX has consistently improved at a slow pace imo.
<pie_>
never did get around to digging into it because the opportunity evaporated
<hodapp>
and you never quite know if pressing the Back button is going to do the right thing or not
<MichaelRaskin>
Has weird bugs around the packaging that I cannot get around fixing
<__monty__>
Duolingo is one of the few commercial web apps that actually does good keyboard UX ime. And yet today I ran into a situation where it was impossible to solve an exercise without a mouse.
<pie_>
srk: yeah I fixed _an_ issue with it with MichaelRaskin at some point
<hodapp>
duolingo has a web app?
<srk>
hah, cool
<srk>
nostalgia much
<__monty__>
The site is a web app imo.
<pie_>
hodapp: surprise. and when i poked it it didnt even have the limitaitons of themobile version
<hodapp>
did they ever figure that in German it is "es tut mir leid" and not "es tut mer lied"?
<MichaelRaskin>
Duolingo definitely deserves being called a web app
<MichaelRaskin>
hodapp: well, if it got reported enough times…
<MichaelRaskin>
It's actually a good web app, too
<__monty__>
hodapp: Probably, course material can be reported by users and does seem to be actively updated.
<{^_^}>
karelzak/util-linux#1286 (by samueldr, 1 day ago, merged): libfdisk: Include table-length in first-lba checks
<samueldr>
they merged it even though CI was entirely red
<samueldr>
it looked like it wasn't issues with my changes
<samueldr>
and their test suite is totally not good, since it **requires** running with elevated privileges
<samueldr>
(and runs things like fdisk)
Jackneill has quit [Ping timeout: 240 seconds]
<samueldr>
and there is no facilities provided for easily testing it in a VM or anything
<samueldr>
but still, no questioning the redness of the CI
<__monty__>
They recently introduced exercises with a sentence with gaps and then buttons with words to put in those gaps. At first I hated it, at least with the numbered buttons you could type a number for selection. Then they made it possible to type part of the words and the first unique prefix causes selection. If a word is a prefix of another one you can hit enter. Today I ran into `a la` and `al` though,
<__monty__>
and "a la" was for the first gap. You'd think that's no problem but spaces seem to be ignored and at the same time the prefix "al" doesn't match "a la" >.<
<MichaelRaskin>
Ouch
lunc has joined #nixos-chat
<MichaelRaskin>
But sounds like a bug they might fix eventually
Jackneill has joined #nixos-chat
<__monty__>
Yeah, I need to make sure it's not vimium interfering and then figure out how to report such a bug.
tomberek has quit [Ping timeout: 240 seconds]
rj has quit [Ping timeout: 240 seconds]
addcninblue has joined #nixos-chat
rj has joined #nixos-chat
tomberek has joined #nixos-chat
<sphalerite>
samueldr: sounds like fertile ground for an upstream nixos VM test :p
<__monty__>
TIL, on a mac a window title can simply write over the part of the status bar that has the clock and wifi and such.
<lovesegfault>
hexa-++
<{^_^}>
hexa-'s karma got increased to 31
rj has quit [Ping timeout: 240 seconds]
<lovesegfault>
do any of y'all use users.users.?.passwordFile?
rj has joined #nixos-chat
<cole-h>
nah
<lovesegfault>
cole-h: Do you understand the docs for it? is the passwordFile supposed to hold the hashed pw or the plaintext pw?
<cole-h>
The password file is read
<cole-h>
on each system activation. The file should contain exactly one line, which should be
<cole-h>
the password in an encrypted form that is suitable for the chpasswd -e command.
<cole-h>
assuming hashed :P
<lovesegfault>
hashing != encrypting
<lovesegfault>
That's what's confusing me
<cole-h>
maybe wrong word used on accident?
<lovesegfault>
maybe, I should make a PR
<cole-h>
oh hm
<cole-h>
maybe it's actually supposed to be encrypted
<cole-h>
chpasswd -e is literally chpasswd --encrypted
<lovesegfault>
huh
<lovesegfault>
encrypted _now_
<cole-h>
but, --crypt-method has md5, sha256, sha512 as options
<cole-h>
🤯
<cole-h>
lovesegfault: now I'm confused, why would you do this
<cole-h>
;P
<lovesegfault>
lol
<sterni>
this is just an old terminology issue as far as I know
<cole-h>
yeah, seems like it
<sterni>
since on unix crypt(3p) has been used for password storage usually
<sterni>
which is in libc
<cole-h>
lovesegfault: so if I were to try it, I'd use the hashedPassword
<cole-h>
:P
<cole-h>
despite any clashes in terminology
<sterni>
I'm guessing it used to be an actual encryption function originally maybe, instead of a hashing function that accepts a salt
<lovesegfault>
gchristensen: are you responsible for the new look of Hydra?
<gchristensen>
in the sense that I tested and pressed the merged button, yeah. in the sense that I did the work, no -- twhitehead did most of the work, and samueldr did the second most work :)
<cole-h>
in part, by pushing the PR forward, I think :P
<samueldr>
so, we could, we can, it mainly needs to be _tested_ and documented
<samueldr>
right now I have had three boards, with two different SoC families tested as working
<samueldr>
currently validating on a fourth (nixos-install is running) while eating
<hexa->
neat
<hexa->
samueldr++
<{^_^}>
samueldr's karma got increased to 336
<hexa->
good read
dingenskirchen has quit [Ping timeout: 240 seconds]
<MichaelRaskin>
impressive
<samueldr>
right now I had to use a modified iso for the pinebook (allwinner a64), but only for availableKernelModules, and a driver that should have been turned on, but nothing actually specific to the pinebook a64
<samueldr>
either allwinner specific, or relatively generic, but required
Synthetica has quit [Quit: Connection closed for inactivity]
dingenskirchen has joined #nixos-chat
supersandro2000 is now known as Guest61649
Guest61649 has quit [Killed (hitchcock.freenode.net (Nickname regained by services))]