<samueldr>
The best errors are those caused by a user’s browser extension and that have nothing to do with your app at all
<samueldr>
or even worse, errors coming from poor cellphone reception that end up being logged, where you cannot really act on the error since there is no error typing, you'd have to grep on the message... which is in the user's locale, and differs between browsers
<gchristensen>
heh
<drakonis>
lol
<drakonis>
real talk tho
<drakonis>
that guix blog gonna be bad for them because it'll lead to a lot of people thinking they actually axed linux support
<drakonis>
despite it being april fools, because we're in the middle of a pandemic
<pie_[bnc]>
i desperately want someone to merge it into mainline
drakonis has quit [Quit: WeeChat 2.7.1]
<lovesegfault>
I wish there was a way to nest nix-shells in shebangs
<lovesegfault>
like #!/usr/bin/env nix-shell \n #!nix-shell foo/bar.nix -A foobar -i bash \n #!nix-shell -p hello -i bash \n ....
<lovesegfault>
Maybe that works? I haven't tried :P
<lovesegfault>
but I kind of doubt it
drakonis has joined #nixos-chat
<ashkitten>
gchristensen: we may have acquired 3 xeon phi cards
<lovesegfault>
They still make those?
<lovesegfault>
wait
<lovesegfault>
it's April 1st
<ashkitten>
i don't do april fools lol
<ashkitten>
a friend has 3 cards and is giving them to us
<lovesegfault>
who is us? Nix?
<ashkitten>
me and ky0ko
<lovesegfault>
Oh, sweet :)
<ashkitten>
244 threads babeyyyy
<drakonis>
amazing
<lovesegfault>
is that x86?
<lovesegfault>
How do you even use those things
<ashkitten>
uh
<ashkitten>
apparently with opencl or as separate linux systems
<ky0ko>
yes, it's x86_64. they run linux.
<lovesegfault>
what the florb
<ky0ko>
they're basically a pentium pro, but overclocked to 1.2ghz, with added 64 bit support, avx1/2/512. 61 cores, 4-way hyperthreading (244 threads), 16gb of GDDR5 in a 16-channel config.
<samueldr>
ashkitten: do you know which ones? atom or pentium based?
<ky0ko>
pentium based
<samueldr>
and there is the answer, if I wasn't so impatient :)
<drakonis>
xeon phi is x86 yes
<ashkitten>
oh wait, each card is 244 threads
<ashkitten>
732 threads.....
<drakonis>
well
<samueldr>
ugh, I'd like to have some kind of recursive bisect, where I can "bisect" a swath of commits that are known-ungood, but not "bad"
<samueldr>
annotate them as "off-topic" without having to `skip` through them all
<samueldr>
and another relevant bit, **annotating** them, so I can know which ones built, and failed in an ungood way, but not bad way, so I can go back to investigating them afterwards
<samueldr>
like right now, I took the bull by the horns, and stopped blindly skipping, and took to the history to elect likely candidates and almost directly found the commit where my bad behaviour starts, and where that ungood stops, instead of probably skipping over 27 commits
<samueldr>
and I need a way to differentiate two issues in the `bad` bisection :(
<samueldr>
(it's not nixpkgs that I'm bisecting)
<infinisil>
But I would implement this by looking at which files a Nix evaluates, then skipping the commits that don't touch any of those files
<samueldr>
nix evaluation is not the problem... or rather, it *is* the problem
<samueldr>
something "broke" (or was fixed and I need to fix something?) in Nix itself
<infinisil>
Seems like a case for
<infinisil>
> yesn't
<nix-build>
false
<samueldr>
hm?
<infinisil>
(yesn't = it both is and isn't, and here it's defined to randomly choose between either)
<samueldr>
ah
<samueldr>
now I'm thinking that my first ungood revision may be the first bad revision, but that the error message changed due to other conditions
<samueldr>
since there's no good revision between the ungood revisions and the first bad
waleee-cl has quit [Quit: Connection closed for inactivity]
<samueldr>
aw yiss, that "first bad" was a fix for that "last ungood", and it's trivial to apply all along the chain so I should be able to get the actually breaking commit
<infinisil>
"During the run it notices several intermediate failures which prevent it from deciding whether the commit is good or bad. It determines which commits fix those intermediate failures and automatically cherry-picks those commits to continue the bisection"
<infinisil>
samueldr: Seems like nix-bisect does pretty much that :o
<samueldr>
but on nixpkgs, no?
* infinisil
doesn't know
<samueldr>
hm, looks like it might be able to work with something else
drakonis has quit [Quit: WeeChat 2.7.1]
evanjs has quit [Ping timeout: 240 seconds]
evanjs has joined #nixos-chat
lovesegfault has quit [Ping timeout: 272 seconds]
lovesegfault has joined #nixos-chat
lovesegfault has quit [Quit: WeeChat 2.7.1]
lovesegfault has joined #nixos-chat
<sphalerite>
is it just me, or… is lisp syntax pretty hard to work with, coming from haskell?
<sphalerite>
I keep getting lost in all the parentheses
cole-h has quit [Ping timeout: 256 seconds]
<sphalerite>
(haskell or nix, hehe)
<ldlework>
sphalerite: lispy-mode in emacs pretty much solves that problem for me
<sphalerite>
ldlework: ooh, will take a look
<sphalerite>
I mean, the only reason I'm writing lisp is to try and make my emacs better, so… :)
<ldlework>
sphalerite: cool, what kinds of things are you adding?
<sphalerite>
ldlework: trying to make it possible to "reload" my config (managed by nix) after updating it
<sphalerite>
oh, lispy seems to be not very compatible with evil
<sphalerite>
but quite vimmy
<danderson>
fwiw, emacs-lisp is not proper lisp
<danderson>
it's 30-years-ago lisp without all the nice things :)
<ldlework>
danderson: not sure what it's worth since we're discussing customizing emacs
<ldlework>
also it's not like elisp hasn't seen any work in 30 years, they work on it all the time
<danderson>
it gave me a bad opinion of lisps for a long time until I got my hands on a nice one. But yeah, you're stuck with it for emacs improvements.
<ldlework>
it has some quirks but is largely usable
<sphalerite>
danderson: is the _syntax_ of elisp really that different from other lisps?
worldofpeace has quit [Quit: killed]
dtz has quit [Quit: killed]
colemickens has quit [Quit: killed]
leons has quit [Quit: killed]
Irenes[m] has quit [Quit: killed]
rycee has quit [Quit: killed]
aanderse has quit [Quit: killed]
emily has quit [Quit: killed]
tokudan[m] has quit [Quit: killed]
jtojnar has quit [Quit: killed]
arcnmx has quit [Quit: killed]
thefloweringash has quit [Quit: killed]
vaibhavsagar has quit [Quit: killed]
yegortimoshenko has quit [Quit: killed]
lovesegfault has quit [Ping timeout: 272 seconds]
jtojnar has joined #nixos-chat
yegortimoshenko has joined #nixos-chat
vaibhavsagar has joined #nixos-chat
Irenes[m] has joined #nixos-chat
dtz has joined #nixos-chat
worldofpeace has joined #nixos-chat
arcnmx has joined #nixos-chat
aanderse has joined #nixos-chat
rycee has joined #nixos-chat
thefloweringash has joined #nixos-chat
colemickens has joined #nixos-chat
tokudan[m] has joined #nixos-chat
leons has joined #nixos-chat
enick_900 has joined #nixos-chat
__monty__ has joined #nixos-chat
Emantor has joined #nixos-chat
CRTified has quit [Ping timeout: 240 seconds]
CRTified has joined #nixos-chat
__monty__ has quit [Quit: leaving]
CRTified has quit [Quit: Gateway shutdown]
__monty__ has joined #nixos-chat
waleee-cl has joined #nixos-chat
<gchristensen>
my favorite anti-feature of wayland is how if I copy something from a program and exit the program, the copied text is no longer in my clipboard
<viric>
hoho
<viric>
Is it like Macintosh, where all this is named "user fault"?
<MichaelRaskin>
gchristensen: I think this is X11 compatiblity
<qyliss>
yeah x11 is the same
<gchristensen>
that can't be true ...
<MichaelRaskin>
Of course they picled picked the bad parts of X11 to inherit
<gchristensen>
misery
<MichaelRaskin>
gchristensen: well, in X11 your _DE_ might have fixed the problem
<MichaelRaskin>
And of course in Wayland everything is in one process so the chance that small things get fixed in a separate configurable layer (which is cheap to change) is… not there
<qyliss>
The X11 solution is to use a "clipboard manager" program
<gchristensen>
well anyway, it is still my favorite anti-feature of wayland. where it comes from I don't really care :P
<MichaelRaskin>
It is meaningfully complicated to do generically, though!
<qyliss>
yeah
<qyliss>
GNOME has one also
<qyliss>
as part of mutter
<qyliss>
that does both x11 and wayland
<MichaelRaskin>
wl-paste, they say… can we also have wl-wm?
ekleog has quit [Remote host closed the connection]
drakonis has joined #nixos-chat
<MichaelRaskin>
Ah, boring. Wayland allows any client to watch the clipboard and grab it on every change.
<gchristensen>
didn't wayland make some claims about being more secure than x11?
<qyliss>
Both
<gchristensen>
both?
<qyliss>
It _can_ be secure, unlike X11
<MichaelRaskin>
It is sometimes more secure in some respects.
<qyliss>
Every compositor _chooses_ to let every application access the clipboard
<gchristensen>
ah
<qyliss>
But the functionality is all there in the protocols and stuff to allow the compositor to decide whether or not to share clipboard however it likes
<qyliss>
unlike with X11
<qyliss>
Similarly, most Wayland compositors allow any application to screenshot at any time, with no user interaction
<gchristensen>
ah
<qyliss>
But they _could_ prompt. On X11, any application can access the whole screen, always, at any time.
<MichaelRaskin>
Well, an X server also could deny some requests
<qyliss>
My understanding is "sort of"
<qyliss>
I'm not an expert, but the experts tell me this is how it is.
<MichaelRaskin>
From the basic protocol structure level, there is nothing that impedes, say, an X-based AIGLX compositor to just lie to each client it is alone, then draw the combination however.
<MichaelRaskin>
And, after all, Wayland allows XWayland and probably gives the same guarantees as for normal clients.
cjpbirkbeck has joined #nixos-chat
worldofpeace_ has joined #nixos-chat
worldofpeace_ has quit [Quit: worldofpeace_]
rardiol has quit [Ping timeout: 265 seconds]
rardiol has joined #nixos-chat
cole-h has joined #nixos-chat
wildtrees has joined #nixos-chat
wildtrees has quit [Remote host closed the connection]
CRTified has joined #nixos-chat
rardiol_ has joined #nixos-chat
rardiol has quit [Ping timeout: 264 seconds]
mrCyborg has joined #nixos-chat
edef has quit [Read error: Connection reset by peer]
edef has joined #nixos-chat
rardiol has joined #nixos-chat
rardiol_ has quit [Ping timeout: 240 seconds]
<srk>
hehe <nixos/profiles/ubuntu>
<srk>
next step would be ubuntu built from nixpkgs
<drakonis>
fun
KeiraT has quit [Remote host closed the connection]
KeiraT has joined #nixos-chat
lovesegfault has joined #nixos-chat
<pie_[bnc]>
srk: >:)
<ashkitten>
MichaelRaskin: xwayland clients can do whatever they want to other xwayland clients
<MichaelRaskin>
Not even a togglable option?
<ashkitten>
xwayland does not protect x11 clients from each other, but it does provide the same security guarantees between xwayland clients and sway clients
<MichaelRaskin>
Ouch
<ashkitten>
i mean, it's x11, and they're connected to the same display. what can they even do? x11 has no security model whatsoever
* srk
had high hopes for 2020 to be THE year
<MichaelRaskin>
What same display means, is also kind of underspecified
<MichaelRaskin>
They can make each connection see a display only with it, for example
<ashkitten>
i don't know how well that would work, but i'm not an expert
<ashkitten>
idk if xwayland is a primary concern for wayland devs anyways
<ashkitten>
more things should support wayland
<MichaelRaskin>
I always had an impression that it is just too much work to support the mess of extensions, so instead of doing X12 they decided to scratch everything.
<MichaelRaskin>
Wayland should support more non-single-PID setups.
<ashkitten>
what does that mean exactly?
<MichaelRaskin>
WM logic and OpenGL logic should not be in the same process.
<ashkitten>
oh, i don't know anything about that
<MichaelRaskin>
Major X11 DEs are annoying; niche WMs fit the niche preferences well.
<MichaelRaskin>
Of course some desktop automation is needed, and this is a separate set of niche tools.
<MichaelRaskin>
In Wayland… well.
<ashkitten>
btw it looks like xwayland is literally just the xorg server but it uses wayland devices. also it's apparently merged into xorg so i don't think they're gonna be adding isolation features or anything like that
<MichaelRaskin>
Ah.
<ashkitten>
like i said, i would be surprised if applying security concerns to x11 where clients expect to be able to do whatever they want is actually worth anyone's time
<MichaelRaskin>
Well, most of the clients work fine being the only clients. And one could just socket-activate a new copy of Xorg per client if desired.
endformationage has joined #nixos-chat
<ashkitten>
still have no idea what it'd break
<MichaelRaskin>
I do (I actually stopped using mainline NixOS because quickly launching a second Xorg instance became too annoying)
<MichaelRaskin>
Not much for applications
<ashkitten>
well, maybe you can send a patch somewhere
<MichaelRaskin>
Some tools would stop working, sure
<MichaelRaskin>
Well, if XWayland is is just Xorg with Wayland as backend, and Wayland has protocol notions for security that are not filtered in practice…
<ashkitten>
eh?
<MichaelRaskin>
The compositors could filter screenshot access, clipboard access etc., but choose to let any client access whenever
<ashkitten>
i mean, it's the clients choosing that themselves when they connect to the x11 server
<MichaelRaskin>
No, the same holds between wayland applications
<ashkitten>
oh, that a client can use wayland protocols to capture a different client?
<MichaelRaskin>
Look in today's logs, if someone here has checked this it is q*liss
<MichaelRaskin>
Well, at the very least a full screenshot
<MichaelRaskin>
Looks like by now the only actual difference in Wayland is better integration of shared-memory buffers
<ashkitten>
i wonder if there are things a client could do to mark itself as uncaptureable
<MichaelRaskin>
Probably not
<MichaelRaskin>
Also, are we going back to SecureKeys of old X11 that many people have forgotten by now?
<ashkitten>
a compositor could enforce these security things in practice if that was useful to the compositor
<MichaelRaskin>
(X11 protocol allows an application to require its keyboard input uncapturable)
<ashkitten>
wayland clients can't mess with other wayland clients, which is better than x11 provides
<MichaelRaskin>
The idea of things being useful to compositor is … interesting
<ashkitten>
in x11 any program can draw on another window if they want
<MichaelRaskin>
In what sense cannot mess?
<MichaelRaskin>
Not sure it actually worked in the AIGLX setting
<ashkitten>
i don't know exactly what that means, but i've tried drawing on other windows in x11 and it works
<ashkitten>
there's zero security
<MichaelRaskin>
Yes, and _some_ of it is even used
<ashkitten>
yeah
<MichaelRaskin>
(mostly drawing on the root window to set background)
<ashkitten>
like for instance wallpapers can be set by drawing on the root window
<ashkitten>
right
<ashkitten>
but that's a security risk
<ashkitten>
also, clients can send inputs to other clients
<MichaelRaskin>
As I said, that's actually opt-outable
<MichaelRaskin>
In X11.
<ashkitten>
but anyway the point being, in wayland there's no assumption of being able to do whatever you want, even if that's sometimes partly true in practice
<ashkitten>
i may look into if a window can be excluded from being able to be captured, that would be very interesting
<MichaelRaskin>
If there is no such assumption, it's just because not enough helper tools have been implemented.
<ashkitten>
android has that concept too, if you try to screenshot, say, the netflix app, it will tell you that you can't do that. if you get clever and try to screenshot it in multiwindow mode, it'll show up as a black rectangle
<MichaelRaskin>
Well, _some_ screenshotter will be able to capture everything
<ashkitten>
what do you mean?
<MichaelRaskin>
Users want to take screenshots
<MichaelRaskin>
Unlike Android, they are not actually hostage
<gchristensen>
iirc there is a way for the data to remain encrypted until it arrives at the physical screen controller
<MichaelRaskin>
I am not sure it actually works under Linux with _all_ drivers
<ashkitten>
so you're talking about a potential "privileged" screenshot tool?
<MichaelRaskin>
Well, either there is a privileged tool everywhere, or there is a nonprivileged tool that gets network effect and becomes a problem
<ashkitten>
i mean, really like i said i don't see wayland as giving most users a different experience than they would on x11. however, there are some use cases that require a concept of security boundaries, and that cannot be done on x11 neatly
<ashkitten>
that's really all i have to say
<MichaelRaskin>
It can be done in X11 the protocol
<MichaelRaskin>
A secure protocol should not even let the normal apps talk about security boundaries
<MichaelRaskin>
(harder to grab what you cannot even describe)
<MichaelRaskin>
Yes, Xorg became too much effort to maintain
<ashkitten>
x11 in my opinion is too big and overgrown to even try to mess with like this
<samueldr>
>> We cannot but note that the neural network managed to learn simple arithmetic only from training on images. It is possible to train a model that first turns an image into numbers, performs the arithmetic and then renders an image of the result. This is not how our model works. We do not have explicit arithmetic step in our network, but it is still able to generate correct answers.
<samueldr>
p. 110
<infinisil>
Neat
<samueldr>
Sorting with Human Intelligence, p. 258 is good too
<tilpner>
Oh no, this is horrible
<tilpner>
If I wanted to sneak a bug into nixpkgs...
<pie_[bnc]>
why are people obsessed with formatting
<__monty__>
Can't say I'm looking forward to apps telling me I can't take a screenshot.
<nix-build>
__monty__: 19 hours, 39 minutes ago <cole-h> btw Iosevka has a 0 that looks different from empty set: cv93 https://i.imgur.com/41wE4Fg.png
<pie_[bnc]>
cant thez just reformat stuff in their editor automaticallz
<__monty__>
pie_[bnc]: Not really.
<__monty__>
You'd have to format back to the project's format before submitting code.
<tilpner>
(Ideally, the author provides reproduction steps, and we can all check the final nixpkgs hashes to the same)
<__monty__>
Otherwise the diffs get even gnarlier.
<pie_[bnc]>
__monty__: hm
<pie_[bnc]>
tilpner: oh thats a good idea
<__monty__>
cole-h: I don't know of any fonts where they are identical. I was mostly just messing with you. But it's undeniable slashed zero is inferior to dotted zero >: )
<cole-h>
>:(
<cole-h>
Puttem up >:(
<MichaelRaskin>
We just need an AST-aware diff
<cole-h>
(I hate dotted zero because the dot is usually so antialiased it looks like a blob)
<__monty__>
MichaelRaskin: That's not even enough. Most ASTs don't include comments with enough precision to keep them where they are supposed to be.
<pie_[bnc]>
stop making crappy parsers lennyfacee
<pie_[bnc]>
not that i have any idea how
<MichaelRaskin>
Well, we do not need AST to know enough to keep the comment in place, only to know what content it has (and thus whether to highlight it as move-around change or content change)
<samueldr>
text files are a messy lossy abstraction over ASTs :(
<__monty__>
samueldr: And vice versa : /
<samueldr>
yep
<samueldr>
it would be more interesting to use IDEs that worked on ASTs rather than byte soups
<samueldr>
not saying scratch-like drag and drop hellscapes
<__monty__>
MichaelRaskin: Comments could change as well. And you'd have to figure out whether a comment move is semantically relevant.
<samueldr>
but aware of the ASTs
<pie_[bnc]>
samueldr: haha scratch like dnd hellscapes
<samueldr>
tell me you'd want to write real software™ using a drag-and-drop puzzley piece interface
mrCyborg has quit [Quit: WeeChat 2.7.1]
* pie_[bnc]
waits for all the haskell libraries to build...
<joepie91>
I remain convinced that it's quite possible (and viable!) to build a node-based visual programming interface that doesn't suffer from the "this is just a toy" problem like scratch does
<MichaelRaskin>
samueldr: like paredit ?
<samueldr>
I actually don't know, but I'm sure new paradigms for more ergonomical editing can be thought of once you free your mind from the preconceptions of byte soups
<samueldr>
one of these is the visualisation of the code you're editing, instead of being a grid of characters, could be... anything!
<samueldr>
the templeos editor may have some good inspiration, albeit weird, like integrating a preview of the resources in-line
<samueldr>
paredit seems like a step forward, yeah, from my minimal understanding from looking at overviews
<MichaelRaskin>
Well, you might want to find and interrogate people who like it enough to never enter unmatched parentheses
<__monty__>
I think what's holding visual programming back is the physical interface rather than the graphical interface.
<__monty__>
It's just so much slowere to work with a mouse than a keyboard.
<MichaelRaskin>
Well, paredit is specifically keyboard-driven
<joepie91>
samueldr: XOD is the closest thing I've seen that tries to be a serious implementation of node-based general-purpose programming (though it still has a loooong way to go): https://xod.io/
<joepie91>
but it does work as advertised, it's just super limited for now
<samueldr>
I never implied mouse orientated programming was the solution, but to instead lift all pre-conceptions about what code editor should be
<joepie91>
__monty__: there's absolutely no reason why visual programming would inherently require a mouse :)
<samueldr>
and am even not implying _visual_ programming!"
<samueldr>
it could just as well be aural!
<joepie91>
people love to make this association between terminal<->keyboard and graphics<->mouse but they're really two entirely separate axes of UI
<samueldr>
the main thing is that character-soup (let's not forget about unicode) based editing of code is a dead-end we've been working around since forever
<MichaelRaskin>
Well…
<samueldr>
(obviously, my opiniong, and that's, like, my opinion, man)
<MichaelRaskin>
I think you might have an easier time pushing for something that us a) truly hypertext b) not called Wikipedia.
<MichaelRaskin>
In the following sense: people are scarily good at failing any attempts to go beyond mostly-linear flow of more or less words
<MichaelRaskin>
Wikipedia actually stands out w.r.t. density of non-linear functions
<MichaelRaskin>
Which is a shame, in the sense that nonlinear features should be used more
<MichaelRaskin>
And with text one can at least point at Wikipedia and show that the direction to go _exists_
<MichaelRaskin>
(well, for code one can point at paredit in Emacs used for editing Elisp, but it's less widely used; but I am a Vim user, am biased against Emacs, right)
<MichaelRaskin>
And there is a nontrivial number of idioms for a non-linear structure of interconnections between ideas/notions/observations for fit together, so maybe one could write a bingo-winning pitch for doing all this…
<pie_[bnc]>
damn
<pie_[bnc]>
mps-youtube's api key broke
<MichaelRaskin>
Extract the API key from Google's Youtube app for Android and recompile
<pie_[bnc]>
hmm >:)
<pie_[bnc]>
does that work though
<pie_[bnc]>
id imagine they do something speial coupled with your google account
<samueldr>
it could, but might not, depending on how they made their backend
<samueldr>
not-google, but as an example, twitter removed abilities from their own API keys compared to the public API keys
<pie_[bnc]>
incidious exists but no cli client
<samueldr>
(and rely on new private to those keys features, compared to the public API)
<pie_[bnc]>
that i know of anyway
<MichaelRaskin>
youtube-dl just scrapes
<pie_[bnc]>
yeah
<pie_[bnc]>
mps gives you a nice ui
<pie_[bnc]>
wewell, i never bothered to learn the mps-youtube ui so im not sure if its really anyy good but it was ok
<MichaelRaskin>
Well, once you scrape a playlist and have the entry names, running a UI on top of that is an exercise to the reader
<MichaelRaskin>
(youtube-dl leaves you literally with a directory of correctly named video files)
<MichaelRaskin>
(or audio, if you prefer)
<pie_[bnc]>
can youtube-dl do searchesi suppose i could look stuffup withlinks2 or something but meh
<MichaelRaskin>
youtube-dl 'ytsearch:octocopters'
<pie_[bnc]>
huhh
<MichaelRaskin>
Seems to grab just the first one by default
<MichaelRaskin>
Hmm. If anything, Moon is supposed to be less yellow, not more
<gchristensen>
hah
<gchristensen>
nice catch
<cole-h>
"Fix it, gchristensen not-ofborg. You're my only hope."
<gchristensen>
the color?
<pie_[bnc]>
MichaelRaskin: well, not the worst thing; mpv $(youtube-dl -g -f best $(youtube-dl ytsearch10:gesugao -e --get-id | fzf --tac)) , its oddly slow at getting the list though so i think it might be downloading the page of every result...
* pie_[bnc]
shrugs
<cole-h>
Please. I cannot go on knowing that it is a slightly different color.
<cole-h>
You must save me
<infinisil>
Hm github is having a bit of trouble for me
<infinisil>
Doesn't update the PR with my new commit
<infinisil>
And it's super slow
<cole-h>
I think it's stumbling. I've noticed the slowness as well.
<MichaelRaskin>
Can it please stop stumbling and fall over already?
<MichaelRaskin>
It kind of does somewhat load for me
* etu
just pushed some cool commits
<infinisil>
Vermin Supreme's dream is coming true
<infinisil>
Oh wait, he wanted to give everybody a pony, not a unicord
<infinisil>
unicorn*
<MichaelRaskin>
OK, changed the lua label colour
<MichaelRaskin>
By copying, so there is a nonzero chance I did it correctly
<cole-h>
<3 MichaelRaskin My hero!
<nix-build>
MichaelRaskin's karma got increased to 23
<cole-h>
I can go on with my life, thank you.
<infinisil>
We sure are rather dependent on github eh
<etu>
unicorn!
<infinisil>
I wonder if githubstatus sends a ping to people if the network traffic spikes
<cole-h>
I always get unicorns when I try to visit GrahamcOfBorg's gist page, unrelated to anything else
<samueldr>
general slowness and unicorns here too
<etu>
git says I'm all pushed but it doesn't show up in the github ui and I can't tell people about all the great changes :(
<MichaelRaskin>
etu: well, you should be able to construct an URL for your commit!
<MichaelRaskin>
Then I can tell everyone that the great changes exist but only I am allowed to see them, somehow
<infinisil>
etu: Yeah same
<__monty__>
joepie91: I wasn't saying visual programming implies the mouse. I was lamenting the current state of affairs where that seems to be the case.
<__monty__>
I guess it's migrate-to-azure day? : >
<MichaelRaskin>
> Preliminary user reports include indications that the behavior is ‘‘very unstable’’ or ‘‘does not work at all.’’ Clearly, a wider-scale test deployment is needed.
<nix-build>
error: syntax error, unexpected $undefined, expecting ')', at (string):293:67
<MichaelRaskin>
It's actually from SIGBOVIK2020, but oddly applicable
<joepie91>
MichaelRaskin: is that like the ignoble awards but as a tech conference
<__monty__>
Wow, everything but Github Pages and Actions seems affected.
<MichaelRaskin>
Come on, Ignobel awards do require papers to be actually published in venues enforcing some methodological soundness
<samueldr>
joepie91: kinda, but specifically made to be weird
<samueldr>
and they're usually all really working things
<joepie91>
right :P
<adisbladis>
Come on github!
<adisbladis>
I want to push some stuff
<LnL>
rip I did
<samueldr>
don't know if you saw my earlier quote of that computer vision thing that does computations?
<MichaelRaskin>
I did
<joepie91>
so I don't think I've suggested this here before, but if anyone is looking for something thought-provoking to watch in these covid-y times, I can strongly recommend Utopia (the UK drama, not the reality series)
<joepie91>
CW: violence that makes you very uncomfortable, but it serves a purpose, as well as conspiracies
<joepie91>
it's a very, very good series that hasn't gotten anywhere near as much attention as it deserved
<nix-build>
#83909 (by OmnipotentEntity, 1 day ago, open): Per RFC45, remove all unquoted URLs
<cole-h>
Probably related to GH going limp for a second
lovesegfault has quit [Quit: WeeChat 2.7.1]
<MichaelRaskin>
I wonder if these nulls in previously non-nullable fields were from early buildup to the failure
<gchristensen>
interesting
<MichaelRaskin>
OK, I would assume from titles that the longest chess game write-up in SIGBOVIK would take a ton of space by including the entire game. But no, game takes just a few pages. A bogus list of references of some quasi-ML paper…
nix-build has quit [Remote host closed the connection]
{^_^} has joined #nixos-chat
<samueldr>
tom7 (the author) did another chess writeup last year, I think it builds up on that one
<qyliss>
tom7 is fantastic
<samueldr>
didn't dwell much on that one since it went over my head
<gchristensen>
danderson: congratulations on the launch!
<gchristensen>
oops
<danderson>
gchristensen: thanks!
Jackneill has quit [Ping timeout: 240 seconds]
<danderson>
fielding an interest spike from that launch now, and it's funny seeing a few people trying to run it by hand on nixos, and being surprised when I tell them it's built in, just `services.tailscale.enable=true` and you're done :P
<gchristensen>
<3 nice
<danderson>
(also points out that the "static" binaries we distribute seem to be dynamically linked now - woops how did that happen)
<gchristensen>
nix users are good at finding that
<danderson>
yup. But also a good reminder that I need to write a KB article about tailscale on nixos
<danderson>
so people stop trying to run it by hand :)
<lovesegfault>
How on earth did we ever use displays >20" with 1080p?
<lovesegfault>
Even with 4K I feel like I need more pixels
<gchristensen>
samueldr: hilarious.
<danderson>
hah
<samueldr>
someone had a wrong remote setup?
<danderson>
if you can reach it, there's a live branch or merge commit somewhere keeping that ref alive
<MichaelRaskin>
Or GitHub is up in the un-split-brain mode?
<samueldr>
that was first looked at yesterday
<samueldr>
so not from today's hiccups
<danderson>
github's eventually consistent with garbage collection of commit hashes, but it tries really hard to deny access to hashes that aren't reachable from a live branch/tag
<danderson>
(so that you can push -f to remove secrets and have it actually be useful)
<lovesegfault>
That's pretty nuts
<gchristensen>
danderson: since you can browsea fork's commits on another fork's browser, I suppose anyone could have pushed it to the repo ever
<danderson>
so, one of the gazillions of branches or tags is probably a Nix repo that someone pushed
<samueldr>
yeah
<samueldr>
that confused me a lot when I realised I wasn't looking at the nix repo
<pie_[bnc]>
nix is sometimes hit and miss on batteries included
<pie_[bnc]>
i often forget to check if something exists already
<pie_[bnc]>
i think we need a paperclip that suggests the executable we are ttying to run exists in some package
<pie_[bnc]>
*trying
<MichaelRaskin>
I wonder, if I push Nix master into a Nixpkgs fork, will later Nix commits still be accessible via Nixpkgs URL?
<danderson>
yeah, and in this case it's completely fair that someone trying "that thing from this random company that just launched" assumed it wouldn't be in nixos
<pie_[bnc]>
MichaelRaskin: probably?
<danderson>
I'm glad I broke that assumption :P
<pie_[bnc]>
:D
<gchristensen>
I don't suppose anyone has a spare macbook pro or mac mini in their house they could erase and hand over to me for the next few weeks?
<gchristensen>
hand over virtually -- install NixOS and hand me an ssh key
<lovesegfault>
They are just so nice and really great to use
<colemickens>
wtaf
<lovesegfault>
But I can't pay that much for a damn board
<colemickens>
sorry, that just came out
<lovesegfault>
I just can't
<gchristensen>
that looks like it would destroy my floor lol
<cole-h>
Make your own :P
<lovesegfault>
gchristensen: They sell a mat, for like $10,000 ofc
<colemickens>
I finally got a real adult desk chair. Need to actually read the manual and adjust it though.
<lovesegfault>
(It's like $30)
* cole-h
sitting here on his Costco mesh desk chair
<cole-h>
thisisfine.jpg
<jtojnar>
it is possible to stand on the balance ball, but you need much higher table
<samueldr>
I'm using one of those rubberish pad that interlocks, with a memory foam mat that people disgustingly place in their bathrooms
<lovesegfault>
cole-h: You're a student, you're indestructible
<gchristensen>
maybe at 10 years of wfh I'll get a herman miller
<lovesegfault>
Truth: The day you get out of college your youth goes away and all of a sudden you need to spend money just to not break your back
<gchristensen>
oh they're havinga 15% sale ......
<cole-h>
👀
<colemickens>
One of my top regrets is not taking better care of my back after leaving college.
<gchristensen>
+1
<colemickens>
This chair was expensive and it's already clear it will pay for itself very soon. Not quite HM territory, but still. If this chair lasts a year, it's a cheap daily expense.
<gchristensen>
yeah. even at 15% off, an aeron isn't in the cards lol
<danderson>
gchristensen: ... can you install nixos on a mac?
<gchristensen>
oh sure!
<gchristensen>
all the NixOS mac builders are mac minis, running NixOS, running macOS in a VM
<samueldr>
generally the question is "can you install Linux on $thing" and nixos generally will
<lovesegfault>
gchristensen: I live next to a store that sells hermann miller chairs and I went to try them out
<samueldr>
with generally the same pitfalls
<danderson>
huh.
<lovesegfault>
IMHO: totally worth it, I sat on a chair and it was mind-blowingly amazing
<danderson>
I assumed apple made it annoying and impossible to boot non-macOS stuff on their machines
<danderson>
(other than approved stuff like bootcamp for windows)
<gchristensen>
their newer models make that increasingly true
<samueldr>
oh, T2-using apple hardware may differ
<samueldr>
but the same still applies, if linux will run, nixos will
<gchristensen>
lovesegfault: yeah .... I know ... just not in the cards this year :P
<samueldr>
at some point in the past, I don't know if it changed, Linux would boot on T2-using hardware, but had not driver to interact with the storage, which has to go through that T2 chip
<samueldr>
so no storing data on your probably only storage
<MichaelRaskin>
I have used a laptop with only a (minimal-formfactor) USB drive as a storage for weeks in a row
<lovesegfault>
It's where my next paycheck is going
* cole-h
keels over at the price
<gchristensen>
cole-h: no kidding
<cole-h>
"Hey, it's only USD 1.3k instead of USD 1.6k!"
<MichaelRaskin>
(not a mac, eww, it was that time when Asus chose a HDD supplier with a busted model)
<lovesegfault>
$1500 is totally worth it IMHO
<joepie91>
what's the warranty on that?
<lovesegfault>
these chairs last forever
<cole-h>
12y
<joepie91>
it better be 10+ years...
<cole-h>
joepie91: ^
<joepie91>
ah okay, 12 years is reasonable
<lovesegfault>
joepie91: 12 effing years!!
<joepie91>
then it's actually not even that expensive, assuming the warranty is actually honored
<cole-h>
Wait, the all-black one is even cheaper
<cole-h>
~USD 1.1k
<lovesegfault>
It is, Herman Miller rocks
<gchristensen>
joepie91: it is the rolls royce of tech worker office chairs, you bet it is honored :P
<cole-h>
But still 😰
<lovesegfault>
cole-h: Yeah, cause that one has more plastic
<lovesegfault>
the silver one is all cast-alu
<joepie91>
so 12 isn't that remarkable :P
<joepie91>
lovesegfault: I mean, keep in mind that even *IKEA* gives you 10 years warranty on their office-tier chairs
<cole-h>
Sounds bad for my cold feet
<joepie91>
gchristensen: so from all I've seen of the Very Expensive Supercar brands, warranty is actually one of the things that sucks a lot with them lol