<V>
[23:41] <lovesegfault> In the discussion for that post someone, I think pcwalton, managed to solve the core issue discussed in the post wrt. lifetimes <- I'd be very interested in seeing this
<V>
I assumed there would be no sane way of representing that in rust
<V>
other than, I guess, green threads/coroutines
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
<siraben>
etu: energizer: re: erase darlings, I see. I'm hoping to get the same effect without changing my filesystem from ext4 by using lustrate
andi- has quit [Ping timeout: 240 seconds]
<gchristensen>
ideally your erase and recreate step is extremely fast, and doing it at the FS layer is useful for that reason
<gchristensen>
having to find and unlink recursively can be slow, especially if you have a lot of files
<siraben>
can't the erase be `rm -rf ...`?
<siraben>
When using ZFS/btrfs, does it affect rescue? I usually use Tails as my rescue distro but don't know how compatible it is with those FS
<energizer>
siraben: rm -rf is slow
<energizer>
mv might be fast, depending
andi- has joined #nixos-chat
<siraben>
Right
<adisbladis>
lovesegfault: Do you know of amchoor?
<adisbladis>
Indian dried mango powder
<adisbladis>
Highly recommend in curries :)
<samueldr>
bless you
<adisbladis>
^_^
<samueldr>
[interpret this as if amchoor was a sneeze onomatopeia]
rajivr has joined #nixos-chat
<adisbladis>
gchristensen: etu gave me a fantastic chili the other day
<adisbladis>
Habanero dulce
<adisbladis>
It has the characteristic chili flavour
<gchristensen>
uh oh
kalbasit has quit [Ping timeout: 256 seconds]
<adisbladis>
But it's not spicy at all
<adisbladis>
Literally not at all
<gchristensen>
whoa what
<adisbladis>
It blew my mind
<adisbladis>
You bite down and the flavour hits you and you kind of brace for the heat
<samueldr>
I guess "not at all" is "comparatively", but yeah, given the right care hot peppers can be prepared in less spicy ways
<adisbladis>
But it never comes O.o
<adisbladis>
samueldr: This isn't less spicy
<adisbladis>
It's nothing
<samueldr>
nothing is a sort of less spicy :)
<adisbladis>
Technically correct :P
<samueldr>
I'd leave the "less spicy" judging to someone like my grandma who just about thinks bell peppers are hot
<adisbladis>
Normally I'm in to the opposite side of the spectrum
<adisbladis>
Like jolokia
<samueldr>
but I can totally get how it can be "not spicy", given the proper care
<hexa->
adisbladis++
<{^_^}>
adisbladis's karma got increased to 118
<hexa->
(for the amchoor tip)
cole-h has joined #nixos-chat
<lovesegfault>
adisbladis: now I do!
<lovesegfault>
That sounds really good
<lovesegfault>
I had been using fresh mango and cooking it down on low heat
<lovesegfault>
until it dissolved into the curry
<lovesegfault>
took forever
<lovesegfault>
V let me try to find it
<adisbladis>
lovesegfault: That sounds pretty darn delicious
<lovesegfault>
It is really good
<lovesegfault>
I also pull a trick where I almost caramelize my onions in the pot before anything else, and then when I deglaze I use some cheap cognac and I add the chipotles together with the cognac
<lovesegfault>
so I get chipotle onion cognac goodness before I drop my spice mix, garlic, ginger, and some sweet peppers (chopped)
<lovesegfault>
then I add unsweetened coconut milk and mango
<lovesegfault>
let it cook for like an hour on low heat
<lovesegfault>
and finally add in some tofu
<lovesegfault>
You could do it with shrimp too, I think it'd be good
<lovesegfault>
V: alright, so it wasn't pcwalton but regardless I found what I was talking about:
<V>
okay, so the solution does involve using contexts
<lovesegfault>
Yeah
<V>
Hm. should be possible to write a nice async API where each attempt to interact with the wayland protocol causes a yield, where the runtime can just cancel any threads working with a resource that's been remotely destroyed
<V>
I've still yet to properly get into Rust's async stuff though, mostly due to how young/incomplete it is
<lovesegfault>
We use it at work
<lovesegfault>
there are some rough edges, but it mostly seems to work
<V>
A lightweight rusty wayland executor/API would be super cool though
<lovesegfault>
Pin has complex semantics though, it's the hardest thing to grasp
<V>
what are your opinions on that vs tokio? I've seen a lot of people disliking tokio and have mostly tended to use Go for networking stuff instead
<lovesegfault>
My main issue with tokio is it's created a de-facto standard library for async
<V>
there's also async_std
<lovesegfault>
with its own little universe
<V>
and tml, and so on
<V>
and I have no idea which is the "correct" or better choice to make
<lovesegfault>
Right, but they're all in compatibility hell
<lovesegfault>
Well, yeah, that's a problem
<V>
as in, have to support tokio's APIs?
<lovesegfault>
the ecosystem is really fragmented
<V>
yeah
<V>
that's what's kept me away from it
<V>
with Go I just have http.Server, tls.Dial/Listen/etc, and so on
<lovesegfault>
The TL;DR for "I just want to get stuff done" is "just use tokio"
<gchristensen>
and avoid anything using futures 0.1 like the plague
<lovesegfault>
Yup
<lovesegfault>
That's another can of worms
<lovesegfault>
nothing is 1.0 yet
<lovesegfault>
so every 6 months or so a new 0.x release comes out and everything breaks
<lovesegfault>
well, they make compat shims but still
<lovesegfault>
PITA
<V>
just like JavaScript frameworks :)
<V>
I want to love Rust. I really do
<lovesegfault>
To be clear, I think the foundations are solid and the future is bright
<lovesegfault>
but we're in the early days of async still
<V>
certainly, but it's just not ready yet
<lovesegfault>
Well, maybe now it's like middle ages
<lovesegfault>
If you're not doing async the rest of the ecosystem is pretty polished
<lovesegfault>
Error handling is mostly in a good state, they're trying to iron out the last bits of standard backtraces
<pie_>
hey if we got to the middle ages that fast
<elvishjerricco>
I wonder how hard it'd be to use the compute module 4 to make a crappy cell phone
<lovesegfault>
samueldr probably has thoughts
<samueldr>
annoyingly, I'd say
<samueldr>
considering pine has the "SOPINE" module that has the same hardware as a pinephone, but they still went with an integrated motherboard rather than a "compute module" kind of board
<samueldr>
though, entirely possible I guess
<samueldr>
but yours is going to rival old "brick" cellphones
<samueldr>
and since the raspberry pi 4 is not a power sipper... well you need to account for that in your power needs
<samueldr>
your main problem will be modem and interfacing with it, I guess
<elvishjerricco>
samueldr: I didn't realize the pi used a comparatively large amount of power
<samueldr>
it's not _that_ bad, but it's not great
<elvishjerricco>
It's got pcie, so just need to hook one of those little cards up to it
<samueldr>
hah
<samueldr>
will it be usable for calls?
<samueldr>
might not be :)
<samueldr>
it might be only for data
<elvishjerricco>
Yea good question
<samueldr>
and then there's the whole blob on the PCI bus, which has access to DMA
<samueldr>
the pinephone uses a usb-based modem
<samueldr>
as long as the usb stack is as solid as one would expect, it should be safe from attacks from the modem
<samueldr>
it's usb based, but only for some features, e.g. data
<samueldr>
IIRC audio has to happen on other pins
<samueldr>
I'm not saying it's a good design, I don't know, but it's _a_ design you can look at
<elvishjerricco>
Interesting. I just have a really impractical love for the idea of the CM4 :P
<samueldr>
you probably can, but assume it's going to be chonkier than the librem 5 :)
<samueldr>
but DEFINITELY a good learning experience
<elvishjerricco>
I don't know the first thing about pcb design though, so that's a non-starter :P
<samueldr>
IIRC there was a pick and place tool for the CM3
<samueldr>
where you could make your own board with connectors for *things* and it'd make something
<samueldr>
I wonder if it's been updated for the CM4
<elvishjerricco>
samueldr: Any idea where I could find that/
waleee-cl has quit [Quit: Connection closed for inactivity]
sorki has quit [Remote host closed the connection]
sorki has joined #nixos-chat
red[evilred] has joined #nixos-chat
<red[evilred]>
is now a good time for me to propose ponylang instead of go and rust then :-)
<red[evilred]>
funny - I read back and was going to say something vaugly snnarky about wayland et al... then there was discussion about curries and cooking and all my snark dissapated.
<pie_>
i want some pie now
<samueldr>
no autophagia
<samueldr>
this is a family-friendly channel
<red[evilred]>
I just learned a new word
<red[evilred]>
and regret it
<samueldr>
you regret learning?
<red[evilred]>
When I read the definition I felt it
<pie_>
hahahah
<pie_>
stop hitting^w yourself
<red[evilred]>
ow
<red[evilred]>
quit it :-P
<red[evilred]>
on the subject of wayland tho - I have this crazy idea
<red[evilred]>
the fact that wayland (or X11) crashes isn't the problem
<red[evilred]>
it's that all your X11 clients terminate that;s the issue
<red[evilred]>
... that and teh author's point of view that concurrency is a bad thing
<red[evilred]>
they're in denial that conucenrecy is bad but ignore the fact that at some point in the future we'll have more CPUs than processes
<red[evilred]>
(Although I suspect this will be an opportunity that systemd won't be able to resist) ;-)
<pie_>
erlang here i come
<pie_>
io tho
<red[evilred]>
:-)
<pie_>
multicore is distsys
<pie_>
shared memory is a lie
<red[evilred]>
I use erlang / elixir in most of my work
* pie_
runs around screaming internally
<pie_>
oh cool
<red[evilred]>
but, for stuff that has to be performant - I use pony
<pie_>
i keep wanting to pick it up and then something else comes up
<red[evilred]>
the beam gives me reliability and a REALLY fast development cycle
<pie_>
i dont understand how some of these random languages just seem to take off
<red[evilred]>
Once you've used a language that allows you to do hot-loading of code - it's really hard to go back to an edit -> compile -> execute cycle again
<pie_>
theres a more recent derivative of E that i forgot the name of
<red[evilred]>
F? :-)
<pie_>
probably not
<red[evilred]>
E++?
<pie_>
yeah i think my current "equicalent" of hot reload is using interpreted languages?
<pie_>
oh irony
<pie_>
Influenced
<pie_>
Pony
<red[evilred]>
I'm a huge fan of the actor model
<red[evilred]>
the main selling point is OTP
<pie_>
yeah it sounds really great
<pie_>
@ actor model
<red[evilred]>
erlang or elixir without OTP wouldn't have 1% of the success
<pie_>
hell i think im accidentally designing stuff in the actor model now
<red[evilred]>
Good !
<pie_>
does interfacing components over the network kind of implicitly do that?
<red[evilred]>
I'll tell you though - once you've made that transition to the actor model it's really hard to transisition back
<pie_>
its basically message passing at that point
<red[evilred]>
exactly
<red[evilred]>
people who are scared of it I point out one thing - and it's a trade.
<pie_>
im bad at understanding second (or even primary :P) effects of design, until aybe ive tried it :P , its kind of random.
<pie_>
can you expand on the transition back
<red[evilred]>
"I'll let you into thhe secret that you've been doing functional reactive programming most of the time without realizing in exchange for recognizing that it was invented in the 1970s" :-)
<red[evilred]>
I never looked back
<red[evilred]>
I did a talk where I compared solving problems using both methods
<red[evilred]>
I had to ask someone else to design the "before" because I had literally forgotten how
<pie_>
oh cool<red[evilred]> "I'll let you into thhe secret that you've been doing functional reactive programming most of the time without realizing in exchange for recognizing that it was invented in the 1970s" :-)
<pie_>
i keep wanting to look into frp
<pie_>
i already "accepted" that all the cool stuff was already invented in the 70s
<pie_>
man i didnt know red[evilred] was so cool
<pie_>
:P
<red[evilred]>
yeah - if you've ever done any javascript, or any graphical programming
<pie_>
by the way your nick is actually red
<red[evilred]>
"on mouseover - execute this"
<red[evilred]>
"on button click - execute that"
<red[evilred]>
is basicaly FRP without the buzzwords
<red[evilred]>
but I'm sure someone got some sweet VC money in Calafornia when they "invented" it (again)
<pie_>
though really ijust know about it because i hang around #proglangdesign
<V>
There are no inventions, only discoveries :)
<V>
pie_: wait, is that not actually dead constantly
<pie_>
iunno h avent been looking around in a while there
<pie_>
i imagine its not recorded or youd have linked it <red[evilred]> I did a talk where I compared solving problems using both methods
<red[evilred]>
Oh - I'll take a look at that - thanks!
<pie_>
its just some rando lang to me tbh
endformationage has quit [Quit: WeeChat 2.9]
<red[evilred]>
So, I did an "intro to functional programming" 35 minute talk
<red[evilred]>
what was designed to just get people curious on the subjhect
<pie_>
oh wait i meant the
<red[evilred]>
not sure if it's been published yet
<pie_>
<pie_> can you expand on the transition back
<red[evilred]>
I will tell you that I wasn't happy with it at the time
<red[evilred]>
but I did a talk on the actor model that I was happy with if you're curious to see it
<pie_>
yeah
<pie_>
gib me mor material pls <3
<red[evilred]>
it's called: "Everything I learned about concurrency and reliability I learned at the waffle house"
<pie_>
the more really deep core stuff i dont have to grok by myself the better
<pie_>
i really do have to stop just polishing my tool polishers at some point...
<pie_>
mmmm
<red[evilred]>
it's great talk because there's enough non-technical examples there to keep people entertained even if they don't fully grasp the concepts
<pie_>
i really need to go to sleep now but my ears are always open for this stuff (as long as i notice)
<red[evilred]>
the talk is worth it just to learn about the waffle house
<red[evilred]>
:-)
<red[evilred]>
I'll check up on iof the functional programming one was published
<pie_>
i heard bad things about elixir tho so i think ill just stick to erlang
<pie_>
also i think VMs are neat (so BEAM too)
<red[evilred]>
I'm curious - what did you hear about elixir?
<red[evilred]>
and btw... elixir compiles doiwn to the exact same bytecode,,,
<V>
isn't elixir just Ruby for BEAM
<red[evilred]>
(fwiw - there are parts of elixir I won't touch for religious reasons)
<red[evilred]>
there's certainly a level of syntax that migrated over
<pie_>
stuff like drawing the line between where you can interface with with erland in elixir is arbitrary, types (iirc?) are kind of arbitrary
<red[evilred]>
but then you could call erlang "prolog for the beam" and it would be just as accurate)
<pie_>
zoxic community management?
<pie_>
toxic
<red[evilred]>
oh - that I've not seen at all
<pie_>
h,
<red[evilred]>
I would love a reference to that
<pie_>
*hm
<V>
are you perhaps thinking of Elm
<pie_>
V: probably, i keep mixing them up
<V>
Too many languages starting with El
<pie_>
but i deliberately looked at both when i was looking and for some reason they ended up in my brain as having similar pathologies???
<red[evilred]>
I really like elm too - but I'm not really involved with that community at all
<red[evilred]>
btw V - do you use the V language?
<V>
I do not
<V>
I have negative interest in it given its history
<red[evilred]>
nah, elm-lang transpiles to javascript and is front-end. Elixir compiles to beam bytecode so runs on the backend
<pie_>
ive seen the prolog stuff but are there any similarities besides the syntax and unification
<red[evilred]>
So, I don't know V's history... but I will say that my first interaction with the author left me walking away with zero interest in persuing it any further
<pie_>
then again i probably just described like half of prolo
<V>
ah, yes. the yanderedev of programming languages
<red[evilred]>
I don't know what that means...
<pie_>
i mean does erlang do the uh...search stuff, i guess? the logic programming stuff
<red[evilred]>
but I asked a question about why something was invalid
<red[evilred]>
and he said: "because there's only one way"
<red[evilred]>
so I asked about why it was valid elsewhere in the language
<red[evilred]>
so I say that I find the parens make readibilty easier for me and he replies: "It's wrong to use perens - learn about orders of operations" or some such bs
<red[evilred]>
so I point out that:
<red[evilred]>
z = ( a < b );
<red[evilred]>
if z {true}
<red[evilred]>
IS legal
<red[evilred]>
and he's all like: That should be illegal. It should be:
<red[evilred]>
z = a < b
<red[evilred]>
...
<red[evilred]>
at that point I checked-out
<red[evilred]>
it's a shame - because I liked the premise of the language
<pie_>
i... itts weird but at least its consistent
<red[evilred]>
Whhat he fails to understand is that language syntax is for humans, not computers.
<red[evilred]>
and as a human, if I find precidence operators hard to remember (or DIFFERENT in different languages too - they're NOT standard no matter what anyone tells you) then I should be able to be explicit with ()
<pie_>
yeah i never gave a crap about orecedence... parens of fear everywhere
<red[evilred]>
I'd rather be explicit than implicit
<red[evilred]>
whenever I see stuff like: 24 * 7 + 14 / 24
<red[evilred]>
I just wanna be like smack
<nicolas[m]>
we should just use reverse polish notation, that will solve all problems!
<red[evilred]>
Postscript baby!
<pie_>
or lisp
<samueldr>
red[evilred]: obviously, you multiply by 24, then divide by 24, so you can just remove the terms, 7+14 = 21
<pie_>
xD
* red[evilred]
chokes
<samueldr>
I'll take my pullitzer in shitposting now, thank you very much
<pie_>
samuepdr has so much humour these days :'D
<samueldr>
don't tell the middle ages, they'll try to bleed it out of me
<pie_>
youll see at next nixcon the ladies will be all over him
<red[evilred]>
I'm curious if nixpkgs has an apl compiler...
<red[evilred]>
ssec
<samueldr>
(this is how you know I'm trying to avoid actually writing what I should be writing)
<pie_>
how do tou do it
<red[evilred]>
btw
<red[evilred]>
if you want to confuse someone
<samueldr>
pie_: big thonk in my thonker
<red[evilred]>
nix-shell -p gnuapl
<red[evilred]>
apl
<pie_>
iant there a gcc apl? so at least thats probably in there
<red[evilred]>
then enter that expression:
<red[evilred]>
24 * 7 + 14 / 24
<red[evilred]>
and then explain how it got that answer :-)
<pie_>
in b4 that operator does not mean what you think it means
<samueldr>
something about 14
<samueldr>
probably the + or postfix /
<samueldr>
(I have no idea what APL should be, other than sounding oddly academic and old)
<pie_>
apl was very enterprise at one point
<pie_>
its excel before excel
<pie_>
or somethibg
<red[evilred]>
you actuially have to use:
<red[evilred]>
24×7+14÷24
<pie_>
hey red do you regularly combine erlang with anythong
<samueldr>
(erlang with strong drinks)
<red[evilred]>
APL is STILL a huge language in its field
<red[evilred]>
in use in business
<red[evilred]>
mainly elixir
<pie_>
also sorry for all the typos, im typing over vnc on my phone yes i know its dumb :p
<red[evilred]>
I use the two interchangaably
<red[evilred]>
in fact, I think the most modern APL implementation is maintained by merryl lynch iirc
<pie_>
who?
<red[evilred]>
they're a huge financial house
<red[evilred]>
APL's niche is definitely in finance
<red[evilred]>
there are some problems where APL is the only language used.
<red[evilred]>
it;'s oevr 50 years old
<red[evilred]>
won the author a turing award
<red[evilred]>
and still does billions of dollars of work in finance
<red[evilred]>
it's exceptionally good at what it does
<red[evilred]>
(apart from sell custom keyboards) ;-)
<red[evilred]>
since most people don't have × and ÷ on their keyboards
<samueldr>
× and ÷ are just there nestled in the compose key
<red[evilred]>
99.9% of people don't know what a compose key is
<red[evilred]>
or alt-gr
<pie_>
having a keybosrd might be nice
<samueldr>
I wouldn't say 99.9% for Alt Gr
<samueldr>
considering that with many non-US layouts it's required to type many characters
<samueldr>
unless you're one of those that think 100% is the US :)
<samueldr>
;) I should have emoted
<red[evilred]>
I would have thought that that would be restricted to people who type in multiple languages
<red[evilred]>
i mean - if your native language is non-english and has otehr characters - most people typically have a national layout right?>
<samueldr>
yes
<samueldr>
and many of them have Alt+Gr
<samueldr>
I know the former and the current official layouts here do
<samueldr>
I'm not familiar with the newer official layout
<red[evilred]>
... and on windowes keyboards?
<samueldr>
??
<samueldr>
but at least with the former, it's crucial to typing many commonly in use symbols
<samueldr>
like @ is Alt Gr + 2
<red[evilred]>
there is a simple solution for that
<samueldr>
wow, the wikipedia section of that page for the layouts is... pretty bad
<siraben>
The HM module for wlsunset hasn't been merged yet
<colemickens>
I use my own hm fork
<colemickens>
also, I don't think it works yet, I don't have my sway set to activate the graphical.target, so wlsunset doesn't start. (at least I think that's what's up)
<siraben>
I wish the modules were more modular so I could mix unmerged modules together
<siraben>
Well, if that is even possible
<colemickens>
for better or worse I've embraced carrying my own fork, probably forever, so that I can do just that
<colemickens>
I have a couple pending PRs mixed, along with commits from one of my own open PRs on the branch that I use.
<colemickens>
if nixcfg -> flake.nix, you'll see I'm using the 'cmhm' branch of my home-manager repo.
<siraben>
Is there another way to set up wlsunset without running it manually?
<siraben>
I could put it in my sway config
<colemickens>
yeah, that should be fine
<siraben>
I recently learned HM was passing in pkgs to home.nix, so it wasn't using my pinned packages all along, heh
<siraben>
Removing that input and let-binding niv's sources fixed it
<siraben>
Hm, need to tie my configuration.nix and home.nix together somehow so that switching to a named profile at boot changes both
<eyJhb>
I just need to find some time, to look at it thibm . THat is always the struggle
<eyJhb>
(also need to get the phones, as they were so cheap I might have gotten scammed)
<thibm>
eyJhb: like someone sold 4 units at once?
<eyJhb>
4 units at once, for under 40 EUR + shipping
<eyJhb>
> EUR 40
<{^_^}>
"40 EUR = 300.000000 DKK"
<eyJhb>
Okay, exactly 40 EUR
<thibm>
wow
Peetz0r has quit [Ping timeout: 240 seconds]
<eyJhb>
200 DKK without shipping. But I offered to pay more, and even more than 40 USD. So I hope it is not a scam. I really need the backcover, as mine is broken.. And the antenna is in there, so the reception is NOT that good anymore
<sphalerite>
eyJhb: if that doesn't work out I can send you my old one maybe
<thibm>
Hm. Mine is working perfectly fine expect for the antenna stuff (radio) (too bad for a phone). I could try to fix it
<thibm>
except*
<sphalerite>
though the back cover is far from flawless
Peetz0r has joined #nixos-chat
<eyJhb>
sphalerite: Would be cool if it doesn't go as expected! Luckily the only thing that makes me think that it won't is the price, But he seems cool!
<eyJhb>
The sides in mine are semi broke, so the top right antenna is half broken off. It doesn't connect all around
<eyJhb>
thibm: What antenna/radio stuff are you experiencing then?
<sphalerite>
eyJhb: huh, the phones were used for smarthome stuff?
<etu>
siraben: The emacs overlay is great, it has all of melpa, elpa, etc, plus some neat tooling
<siraben>
Ooh, there's emacs-pgtk + gccemacs!
<thibm>
eyJhb: things like no signal from operator 50% of the time. I didn't really investigate
<eyJhb>
sphalerite: Yeah, the ones he is selling was used for smarthome stuff. I am guessing like remotes for TV/chromecast, lights, etc. etc.
<etu>
siraben: Like it can parse your emacs config to find use-package declarations to extract which packages your config actually needs. This is done with an elisp parser written in nix.
<eyJhb>
Ohh weird, even with my borked antenna it works OK :D Are you in a urban area?
<sphalerite>
huh ok. That's one way of doing it I guess x)
<siraben>
etu: wow. nice
<eyJhb>
It is actually somewhat commen here in DK (I know more than one, that have a smartphone they use for the TV, and then only pay for streaming the TV Channels)
<siraben>
Just installed emacsPgtk, looks sharp again
<thibm>
eyJhb: no currently but it did not work well in urban areas neither
<eyJhb>
I am hoping that mine works better with a new back :p I have some crashes when it switches from WiFi to not WiFi sometimes...
<eyJhb>
Need to be better friends with logcat
<__monty__>
Just join the feature phone fold again. Your phone doesn't need to be a smartass.
<eyJhb>
well, that costs about the same as what I am paying now! But I have considered it
<siraben>
Does Brave/Chromium have wayland support?
<__monty__>
Just fyi I've heard Brave is pretty shady.
<siraben>
Ah. I only use it to run Element and other websites without any bars
<siraben>
If there's a way to do that with Firefox that'd be good
<sphalerite>
siraben: firefox --kiosk
<thibm>
__monty__: yeah, each time I "need" a new phone (meaning old unit is dead or malfunctioning), which happened only twice, I consider the feature phone option. We'll see next time
FRidh has quit [Ping timeout: 272 seconds]
<sphalerite>
siraben: I also run chromium on wayland, not sure if it's "wayland-native" though
FRidh has joined #nixos-chat
<__monty__>
Once my T303 dies I will get a smartphone tbh. GPS and being able to check train schedules and the like is really useful and I often happen to end up traveling solo.
<sphalerite>
Now I kind of want a feature phone that runs linux.
<__monty__>
Yes, that'd be awesome.
<eyJhb>
nixos-mobilev2 ^
<__monty__>
Though I suspect that would have all the features that make a smartphone more of a privacy burden than a feature phone...
<sphalerite>
like, the chrome of a nokia 3310, with a bit more beef (CPU/RAM/network) and USB charging.
<__monty__>
I never got around to programming my T303, it runs Java ME or something so I think it should be fairly easy?
<siraben>
sphalerite: firefox --kiosk seems to still have my bookmarks bar
<siraben>
sphalerite: chromium appears somewhat blurry on my HiDPI display
<__monty__>
That can be hidden I'm pretty sure?
<siraben>
I'll look into it
<__monty__>
Hmm, --kiosk forces full screen? Not exactly what I hoped.
KREYREEN has quit [Remote host closed the connection]
<__monty__>
siraben: It might be possible to hide almost all the chrome with via userChrome.css if you really want.
<siraben>
__monty__: where did you hear Brave was pretty shady? I didn't research it properly
<__monty__>
Here I think. Maybe from joepie91?
<siraben>
Maybe there are less bloated Matrix clients than Element, heh
<siraben>
I'm basically running an electron app by opening ungoogled-chromium
<__monty__>
Hmm, I cannot corroborate that claim with logs though.
<siraben>
sphalerite: are you doing anything special with chromium on wayland?
<sphalerite>
siraben: nope. Though my display isn't ultra-high-res so maybe that's the difference
<siraben>
Oh my non-retina external monitors I don't notice any difference between apps running under wayland natively and those running under xwayland
FRidh has quit [Ping timeout: 264 seconds]
FRidh has joined #nixos-chat
<siraben>
s/Oh/On
<gchristensen>
do you used mixed DPI?
<siraben>
gchristensen: my laptop screen is an Apple Retina display, my external monitor has a regular DPI
FRidh has quit [Ping timeout: 256 seconds]
FRidh has joined #nixos-chat
z1g34n3r has quit [Remote host closed the connection]
FRidh has quit [Ping timeout: 256 seconds]
<lukegb>
siraben: you want the firefox site-specific-browser feature, I think
<siraben>
Does xdg-desktop-portal-wlr resolve problems with screensharing in programs such as Discord?
<siraben>
lukegb: what's that for?
<lukegb>
Renders a webpage with literally no browser chrome in Firefox
<lukegb>
about:config -> flip browser.ssb.enabled=true, then you can launch e.g. "firefox -ssb https://www.google.com"
<siraben>
lukegb: Thanks, I'll try it
<lukegb>
Or open it in a normal tab and click the three-dot menu inside the URL bar and select "use this site in app mode"
<siraben>
Ah, but now it needs some mutable state somewhere in my filesystem
<lukegb>
You should be able to set browser.ssb.enabled without mutable state using the admin profile functionality stuff
<bbigras>
oh shit, that explore page is awesome and pretty. I saw and liked the new https://nixos.org/ a while ago but I didn't know about the explore one.
<pie_>
that might be new?
<bbigras>
well I didn't explore the new site since I thought I already knew what I needed to know but maybe it was part of the same update.
<bbigras>
I'm sure it will help new users give NixOS a try.
<pie_>
what might be helpful is a page with pain points
<pie_>
"here are things that we have some issues with"
<andi->
Despite all my issues with the initial redesign this page is great! Who made those cute images?
<pie_>
andi-: +1
<gchristensen>
this is really stunning
<andi->
It is missing NixOS but I believe we are getting there :)
kalbasit has joined #nixos-chat
FRidh has quit [Ping timeout: 240 seconds]
FRidh has joined #nixos-chat
<andi->
Is there some way to subscribe (webhooks, machine readable, …) to all events of Nixpkgs without being an owner? I really want to test some stuff.
<gchristensen>
I could add one for you
<andi->
Ok, that would be great. I'll have to punch out a minimal implementation then
<__monty__>
Hmm, I wonder whether the explore page is a little too optimistic? "Stop worrying how to install dependencies for your project," may be true from some POV but it's really not UX-wise. Each language ecosystem (within nixpkgs) has its quirks.
<__monty__>
: / The nixos-homepage nix-shell requires a nix with flake support?
<__monty__>
,locate bin tpage
<{^_^}>
Found in packages: perl528Packages.TemplateToolkit, perl530Packages.TemplateToolkit
<abathur>
always hard to find the goldilocks zone between marketing and managing expectations :)
<thibm>
abathur: --yaml=file.yml taking precedence over the environment variable is weird
<abathur>
it is
<abathur>
maybe "interesting idea" doesn't convey my meaning quite right; I mean something less like "I want to use this" and more like, "I'd never thought of using a shebang like this; I should probably reflect on it long enough for it to get a page in my mental book of weird arcana"
KREYREEN has joined #nixos-chat
<abathur>
does anyone know if github notifies you of an @mention in a repo you haven't contributed to?
<FRidh>
by default, yes
<FRidh>
I occasionally manage to ping someone who has a handle close to that of a nixpkgs contributor :)
rajivr has quit [Quit: Connection closed for inactivity]
<__monty__>
Heh, recently realized it doesn't provide completion for such names though. I think it only provides completion for contributors and conversation participants maybe?
KREYREEN has quit [Remote host closed the connection]
hodapp has joined #nixos-chat
KREYREEN has joined #nixos-chat
cole-h has joined #nixos-chat
aleph- has quit [Read error: Connection reset by peer]
Church- has joined #nixos-chat
<abathur>
__monty__: its completion rules elude me; similar with issues--sometimes I type the full number of an issue and it never shows up in the dropdown.
<gchristensen>
compared to some real datacenter racks I've seen, A+. compared to my home network, A
<sphalerite>
(I _totally_ didn't injure my fingers on the running fan)
<gchristensen>
my home storage server has its top open and a statue named "the maiden" sitting inside it... plus a massive firehazard of cables around it
<hexa->
looks roughly like my rpi setup
<hexa->
loving the backplane
<hexa->
hating the MAX! thermostats
<sphalerite>
hahahaha
<sphalerite>
you're familiar with them?
<__monty__>
What are we looking at, the networking setup from Mad Max?
<sphalerite>
my hacky NAS
<sphalerite>
it used to be in an ATX case with an ATX power supply, but I don't have a good spot for that in my new flat
<sphalerite>
srk: clearly the heatsinks you used weren't big enough :p
<sphalerite>
srk: I imagine a heatsink like the nanopi m4's (shown in my picture before) would probably do the trick
<vika_nezrimaya>
I use a PoE hat for cooling my RPi4 with a modified fan curve so it's not too noisy
<srk>
ty, cool :)
<srk>
sphalerite: still the contact surface is not that large
<__monty__>
srk: They were built to play back media. But not as NASes afaik? I assume everything's limited by the USB buss when using an RPi as a NAS?
<srk>
__monty__: yup, for set-top boxes, which can be sort of NAS as well
<samueldr>
for previous models yes
<srk>
sphalerite: one more reason to play with flir lepton that's collecting dust on my table .. :D
<samueldr>
for newer, PCIe could have changed things
<samueldr>
though I don't recall if the pi 4's ethernet is limited by USB or not
<srk>
it is, but USB3 is not that limiting as before
<__monty__>
I haven't found a satisfactory way of using an old RPi as an HTPC tbh. I've tried NFS and Samba on my server but playback's just always very choppy and basically unwatchable.
<samueldr>
__monty__: is the same playback local bad?
<samueldr>
(I would assume yes)
<srk>
old RPis have issues with too-many-interrupts coming from USB network chip iirc
<samueldr>
I haven't tried hard, but hardly trying and it was just on the border of not great with the pi 4
<samueldr>
using the foundation's kernel even, but not their distro
neeasade has joined #nixos-chat
<__monty__>
From a usb (2) thumbdrive the playback's fine actually.
<sphalerite>
srk: I don't see how that's supposed to be any better in a tablet…
<srk>
yeah, doesn't make much sense. maybe if it's cooled from both sides or with heat pipes
pinpox is now known as |^_^|
<srk>
my imx6q in novena is passively cooled and tops at like 95 degrees when nix-building. still stable at that temp .. :D
<samueldr>
because that's sending the message that we're okay with entities giving 0 documentation and just scrounging up money to somehow hack it ourselves
<andi->
Yeah
<samueldr>
marcan _is_ an individual that can make that happen though
<samueldr>
but that sets a terrible precedent
<andi->
That is why I am linking it. I kind have some sort of trust.
<andi->
I am not happy about the state of hardware :/
<samueldr>
yeah
<samueldr>
we sorely need a wintel alliance for ARM
<samueldr>
because as bad as it was for alternative operating systems
<andi->
Like I can buy a RISC-V board for >1k and get something that is like 10x slower (or so?) and more expensive than the entry level device?
<samueldr>
it at least made things standardized enough for decades
<samueldr>
RISC-V, AFAIUI, is far from being a thing for anything else than embedded uses
<samueldr>
like CPU in your hard drive
<samueldr>
but I don't know enough about it to actually say
<samueldr>
hopefully they can climb the rest of the way and at least get to perfs parity with designs with similar deployment
<andi->
Apple didn't add crazy levels of protection to these device and you are able to boot something else on them.. I wonder what the motivation is? They surely could just have done what they do on iPad/iPhone/…
<samueldr>
I do wonder
<samueldr>
cold feet from lawyers?
<samueldr>
while in tablets and phones it's normal, it's not normal yet in computers?
<andi->
maybe purely regulation based? Are they afraid of the US/Eu/…?
<samueldr>
btw, marcan has been in talk with "never released" (longhorn) on twitter
<samueldr>
the pongo OS team was currently designin an UEFI implementation for the M1
<samueldr>
their approach, while marcan seems hesitant, is better imo as it uses a standard boot chain
<samueldr>
if there are reasons to think perfs is going to be harmed, they can always implement an escape hatch down the line
<samueldr>
but having a standard boot environment will help other non-Linux OS too
<samueldr>
given who's involved, they probably are thinking about Windows too lol
<andi->
yeah, the twitter thread this morning had surprisingly many *BSD users voicing their opinions
<samueldr>
I would really hate to see another special fancy pants boot flow for Linux on ARM
<samueldr>
as if there wasn't too many already
<andi->
Everyone needs their own!
<andi->
That is the common theme
<samueldr>
NO
<samueldr>
:)
<samueldr>
please, either port u-boot with support for everything your platform does, OR provide a full UEFI environment
<samueldr>
(I don't personally care about ACPI vs. DT for UEFI, but if you're doing it, do ACPI for full server ready)
<samueldr>
don't forget to provide the sources!
KREYREEN has quit [Remote host closed the connection]
FRidh has quit [Quit: Konversation terminated!]
KREYREEN has joined #nixos-chat
endformationage has joined #nixos-chat
<colemickens>
siraben: hey, I'm around, not sure if you got all your questions about emacs/chromium answered
<colemickens>
Fwiw, chromium ozone sort of works for me, but is very prone to crashing (at least under Sway) right now.
<colemickens>
Unrelated, I probably won't use this for long since I'm going to try to migrate off gmail, BUT, if you use gmail with an imap client, this might be useful? https://github.com/colemickens/get-xoauth2-token
<colemickens>
I can use `meli` now with my gmail account without having to enable an insecure "app password".
<__monty__>
How does an oauth2 token differ from an "insecure" app password?
<bbigras>
just in case one of you is planning to use gmail with notmuch, https://github.com/gauteh/lieer is great. I think it uses the gmail api instead of imap.
<colemickens>
__monty__: that feels like a pretty foundational question to address
<colemickens>
for one, the oauth token can only access the gmail api
<colemickens>
for another, it's limited duration, so if its stolen, blast radius is reduced
<colemickens>
bbigras: nice
<cole-h>
colemickens: Where are you planning to migrate your mail?
<colemickens>
cole-h: I'll let you know when I decide. Hosting it myself (at least the SMTP side) seems ... like something I will regret.
<colemickens>
cole-h: I really don't know. There's Migadu that people seem to kind of like?
<cole-h>
colemickens: I've heard the same, and I was leaning that way as well.
<bbigras>
are providers still blocking the inbound smtp port? I guess even if they didn't you would want more than 1 mx server.
<colemickens>
bbigras: more worried about orchestrating all of the pieces to not have outgoing mail flagged as spam
<cole-h>
^
<colemickens>
I've read anecdotally that even if you do everything right, you can wind up blocked by lazy/bad sysadmins that blanket ban ranges, etc.
<bbigras>
yeah probably.
<cole-h>
I also need to pick up a domain, or co-opt a domain my dad already has :P
<LinuxHackerman>
I self-host my mail and only rarely have deliverability problems
<LinuxHackerman>
(using simple-nixos-mailserver)
<bbigras>
I had my own email server back in the days. I had to pay no-ip or something to get my emails sent on another port than 25. that was before gmail was a thing.
<__monty__>
colemickens: How do you refresh the oauth token. And wouldn't the gmail API be more sensitive than IMAP access?
<bbigras>
I wouldn't be surprised if people block dynamic ips
<colemickens>
__monty__: I don't understand the second question. The tool stores a refresh token on disk. Obviously if someone compromises my laptop its game off. More worried about a third party IMAP client stealing creds, or them being sniffed over the wire, or a hosted email client being compromised, etc.
<LinuxHackerman>
oh yeah I defninitely wouldn't recommend doing it on a home connection.
<bbigras>
which provider has this imap's successor protocol or whatever.
<colemickens>
Linux Hackerman: when you _do_ have issues, how long does it take to resolve? Are there actions you can take, or do you just workaround?
<bbigras>
is it jmap?
<LinuxHackerman>
what's wrong with IMAP?
<samueldr>
not j enough
<gchristensen>
it is a terrible protocol completely unsuitable for low power devices when low latency is wanted
<LinuxHackerman>
colemickens: it varies a _lot_ from case to case. Sometimes the destination says "you're on this here blacklist" and I can tell the operators of that blacklist that I'm fine actually
<LinuxHackerman>
colemickens: other times the destination is a lot more obtuse about it
<__monty__>
colemickens: The second question is because you said the oauth token can only access the gmail api. While I assume app passwords are just for imap access?
<samueldr>
what about those that just eat the message as if received and actually don't deliver?
<colemickens>
__monty__: my understanding is that Google's app passwords are ... just an extra password to your account for nearly unrestricted access
<bbigras>
that hey.com one seemed refreshing in their new approach but it was a bit expensive.
<colemickens>
__monty__: afaik the oauth token is actually only good for IMAP
<gchristensen>
I'd pay for hey as soon as I can use my own domain
<bbigras>
they still don't support that? it's been a while
<gchristensen>
GA is early 2021
<colemickens>
what has you intriged about hey? they have a nice client?
<colemickens>
I skipped the buzz -_-
<gchristensen>
I like that I, by default, don't receive your mail
<__monty__>
colemickens: How does the refreshing work? You have a daemon which refreshes the token and then your email client uses that token?
<bbigras>
yeah hey has just a different nice UI.
<bbigras>
but 100$ per year is a bit much when I almost never receive emails from a human anymore
<colemickens>
__monty__: meli invokes the comamnd whenever it needs a token. The refresh token never expires (with Google). So, on first boot, it tells you to go do an interactive auth session to establsih the refresh token. Then on subsequent runs, the tool either returns a cached active token, or uses the refresh_token to get a new access token, caches it and returns it
<abathur>
I did find them for the same reason, and I guess I probably find them interesting for the same reason, but they aren't particularly OT for me at the moment
<abathur>
so I've been searching around to see if there are any existing projects that would bolt on...
* samueldr
frowns
<abathur>
and found some random things that just happen to mention shebangs
<samueldr>
another AST that treats comments as garbage
<abathur>
well
<gchristensen>
well
<bbigras>
this is a bit like doctests. I also tried once a tool to generate the README.md by running the code and testing it.
<samueldr>
well
<abathur>
hehe
<gchristensen>
bash doesn't do anything with the shebang
<samueldr>
I **know** that sh as per POSIX defines comments as whitespace
<samueldr>
look: I said *comments* :)
<samueldr>
I know the interpreter doesn't need to care about the shebang
<gchristensen>
;)
<gchristensen>
fair
<samueldr>
but still, I'm always a bit frowny when a parser doesn't keep comments or whitespace in a manner that allows it to reproduce the same input to an output
<gchristensen>
agreed
<samueldr>
to me, it means it's an artificially limited tool
<samueldr>
you won't be able to show the location of an error properly
<samueldr>
you won't be able to use the same parser to provide linting or formatting facilities
<abathur>
it might not be that hard to modify oil to keep them in the AST; it doesn't seem to completely discard them; but it's not a tree I feel terribly inspired to bark up
<samueldr>
haha
<abathur>
it basically has them up to some point, and sets them to some "ignore" type at some point
<abathur>
so it might be as simple as changing that
<abathur>
or it might be a two week yakshave
<abathur>
samueldr oil (and thus resholve) don't eat the comments--I just don't think they're exposed to my visitor
<samueldr>
ah
<samueldr>
less terrible :)
<gchristensen>
resholved is actually quite respectful of the input
<samueldr>
I know that the sh implementation as per POSIX states that comments are whitespace
<samueldr>
well, I seem to remember
<abathur>
I won't swear that; I've noticed it, but the behavior I get now is what I want, so I haven't done a deep dive to ensure I'm not just being dense about how to visit them
<samueldr>
been about 5 years since I tried implementing an sh parser
<gchristensen>
the thing that trips me up about oil is the py2 business, even if it is sensible from a lot of angles
<samueldr>
(I wanted to make a superset language on top of shell, that transpiled to shell... how foolish)
<__monty__>
Py2 business?
<cole-h>
oil has a (fork of) py2 in-tree that it uses
<infinisil>
"Persistent L2ARC - This feature makes the L2ARC cache device persistent across reboots thereby eliminating the usual cache warmup time normally needed after importing your pool."
<infinisil>
Ohh nice
<infinisil>
Well I don't need it anymore, but I could've used that a year ago or so!
<infinisil>
I had a setup where I used the SSD only as a cache device, everything was stored mainly on an HDD
<infinisil>
But every restart would be super slow as everything needed to be reloaded into the SSD
<infinisil>
But with persistent caches I might reconsider doing this..
<infinisil>
ZStandard compression and redacted zfs send/recv are also hella cool!
<infinisil>
"Added fallocate(mode-0/2) compatibility to preallocate space"
<infinisil>
"Enabled the systemd zfs-mount-generator by default on Linux"
<infinisil>
This is a nice release
__monty__ has quit [Quit: leaving]
aminechikhaoui has quit [Read error: Connection reset by peer]
aminechikhaoui has joined #nixos-chat
srk has quit [Ping timeout: 240 seconds]
KREYREEEN has quit [Read error: Connection reset by peer]
KREYREEN has joined #nixos-chat
srk has joined #nixos-chat
vika_nezrimaya has quit [Ping timeout: 272 seconds]