<gchristensen>
mostly I'm just a bit miffed because I'm trying to fill out various "paper"work so I can pay myself, and there are *so* many web portals I have to fill out
<gchristensen>
I would happily pay more in taxes if there was just one really high tax to cover all the taxes :)
<gchristensen>
PFML, WTH, UI, EMAC, ... things I didn't know about 4 hours ago
<aaronjanse>
It's very minimal at the moment. For example, it does *not* correctly implement all of nix (e.g. it doesn't attempt to correctly calculate /nix/store path hashes), but what I have is enough for a proof-of-concept
lunc has quit [Ping timeout: 240 seconds]
<pie_>
i think im running out of allocated vcpus on my worldsimulation instance or something. the deeper i get into learning about something the more the world breaks. the audio quality on this lecture recording is so garbage my ears and attention are bleeding
<siraben>
aaronjanse: what would it take to make it work with Emacs?
<siraben>
Guess it's time for me to pick up more Emacs Lisp heh
<siraben>
cole-h: wow, the graphical one is already at 98.77%
<siraben>
aaronjanse: yeah, I think I'd need a bit of glue to be able to properly nix-eval-lsp
lunc has joined #nixos-chat
cole-h has joined #nixos-chat
<ldlework>
TypeScript's type-system is pretty crazy powerful with mapped and conditional types. I've been thinking about a site called Type Tetris which would be a code playground affair where you try to solve a series of abstract type challenges by satisfying some unit tests by utilizing those type features in pointless but creative ways.
waleee-cl has quit [Quit: Connection closed for inactivity]
<LinuxHackerman>
Does anyone know a tool that can navigate a list of paths interactively? I'm imagining maybe something visually like `tree`'s output, with collapsible nodes. And instead of being produced from actual directories, it's produced by piping in the output of `find` or similar.
<gchristensen>
LinuxHackerman: maybe there is a generic "interactive tree" tool? tree can read from stdin
<LinuxHackerman>
gchristensen: yeah that's what I'm hoping for
<LinuxHackerman>
DigitalKiwi: hm, also not quite what I'm looking for — I want to be able to step through a tree, which is generated from a list of filenames
<gchristensen>
ncdu can take an input file that was created with ncdu ... maybe you can fake its format
<__monty__>
You could kludge that into ranger. Not sure whether it's worth the trouble though.
<LinuxHackerman>
it's basically the output of find, but filtered according to somewhat complex criteria, and I want to navigate it as if it were just a filesystem tree
<gchristensen>
surely this must exist
<LinuxHackerman>
gchristensen: oooh funky idea
<LinuxHackerman>
__monty__: yes, that's the sort of thing I'm looking for — ranger, but with a list of files as input, rather than the filesystem
<LinuxHackerman>
gchristensen: that's what I thought!
<DigitalKiwi>
probably not what you want either but it's cool lol
<__monty__>
LinuxHackerman: Maybe you can fake it till you make it? Basically running mkdir -p on all the paths after making them relative? Then you can "just" browse that using ranger. (Add symlinking in the setup script if worth the trouble.)
<LinuxHackerman>
yeah, that popped into my mind too… but it's also kind of awful :p
<gchristensen>
agreed
<gchristensen>
what you really need is a FUSE filesystem to do it for you
<LinuxHackerman>
hehehe I knew you'd say that
<__monty__>
Yeah, that'd be ideal.
<DigitalKiwi>
can you do it with symlinks
<DigitalKiwi>
heh idk but visidata might actually work for this...
<DigitalKiwi>
tree|visidata
<LinuxHackerman>
oooh tree has a --fromfile option
<LinuxHackerman>
DigitalKiwi: ooh this feels very close, now if only I knew how to work visidata
<gchristensen>
LinuxHackerman: that is what I meant by tree can read from stdin :)
<LinuxHackerman>
Uh. Somehow I missed that part of your message.
Dotz0cat has quit [Ping timeout: 260 seconds]
<LinuxHackerman>
hm. I could replace all the leading characters that each line shares with its predecessor with spaces, then use vim's folding with foldmethod=indent
MichaelRaskin has quit [Ping timeout: 240 seconds]
endformationage has joined #nixos-chat
luxemboye has quit [Ping timeout: 240 seconds]
luxemboye has joined #nixos-chat
tomberek100 has joined #nixos-chat
tomberek100 has quit [Client Quit]
tomberek100 has joined #nixos-chat
tomberek is now known as Guest29953
tomberek100 is now known as tomberek
<abathur>
LinuxHackerman: hmm, I don't know of anything but I agree it seems like it should exist; I did find this https://github.com/hhggit/tree2tui
<pie_>
LinuxHackerman: thanks for asking this ive wanted something similar for a while
<LinuxHackerman>
gchristensen++ abathur++ __monty__++ DigitalKiwi++ this is why I love #nixos-chat
<philipp[m]>
It was all the way but you can't see it on the pic because that was just a smartphone camera.
<cransom>
still amazes me what you get with 'just a smartphone camera'. the first digital camera i had access to wrote to a floppy disk and i'm not sure those images would be suitable for a thumbnail even anymore.
<joepie91>
even my sub-$100 phone shoots pretty good pictures
<philipp[m]>
__monty__: I can even get you a testimonial from a currenly very happy synoptic meteorologist that this rainbow was indeed double all the way.
<pie_>
i can shoot terrible pictures with anything i think
<__monty__>
I trust you : ) Double rainbows aren't that rare here.
<gchristensen>
cransom: I had one of those, what a campera!
<cransom>
aww. the website that i 'designed' in ms frontpage in hrmm... 1997 for my employer finally was taken down to get refreshed. the main photo was something taken by the illustrious sony mavica.
<pie_>
i was thinking of refreshing a friends website
<pie_>
but even though its cheesy it has so much more personality than some generic modern thing
<pie_>
and has everything it needs on it
<philipp[m]>
Here I see a proper double rainbow every few years.
<pie_>
so i ended up thinking maybe not
<cransom>
the load speed was so amazingly fast. it was even quick on a modem.
<colemickens>
>I have zero hope
<colemickens>
big mood.
<aleph->
Heh
<colemickens>
* > I have zero hope
<colemickens>
big mood.
<colemickens>
lol, I guess Matrix is just going ot completely rewrite my markup
<cransom>
yeah. matrix client eats my inputs in strange and unexpected ways frequently.
<colemickens>
everytime I complain about it and it affecting my ability to promote Matrix, I wound up personally bit by the bridge too.
<colemickens>
I don't even bother with auth'd or ident'd rooms anymore since I know I won't manage to stay in them.
<drakonis>
you cant see edits and it ruins multiline messages in the irc side
<drakonis>
its really good at finding new and spectacular ways to mangle messages
<__monty__>
And the reply feature is such a PITA and they simply don't bother trying to solve it because some users insist that the context is useful and necessary on the IRC side of the bridge and not having it would be disastrous.
<cole-h>
ldlework: feedback: "click the button" -- what button? just by looking, it's not immediately obvious that you're telling me to click on the code boxes
<cole-h>
(also cert is for github, not your website)
<__monty__>
Anything would be better than including the context *before* the actual message.
<ldlework>
so they were not written with the new site in mind
<cole-h>
oh, ok -- so not your fault :P
<colemickens>
gnome 40 merged, lets goooo
<__monty__>
f0x: But I'm sure most IRC users don't like the reply context most of the time. Maybe if > X messages have occurred in the mean time and at least Y time has passed, context could be appended. Otherwise a reply should simply be a mention, IMO. When I shared this opinion I was met with hostility though.
<f0x>
hostility, really :((
<colemickens>
I've pretty much only gotten silence when even trying to set expectations properly around the bridge.
<f0x>
I've heard that workaround before, maybe I should look again into implementation, shouldn't be that difficult actually
<__monty__>
Maybe that's too harsh. I didn't feel welcomed though.
<colemickens>
The introduction of yet-another-irc-bridge with no framing around it doesn't help anything.
<f0x>
colemickens: I have a poorly maintained fork with quite some QoL improvements already, it's what I use to speak here
<f0x>
including things like much less spammy media uploads
<colemickens>
cool! nice!
<colemickens>
I am sympathetic, it feels a bit like xwayland in a way almost (seriously irc fans dont at me)
<__monty__>
I don't mind the Matrix side being nice and matrix-y, everything working through the GUI, that's up to individual matrix users. I do like the idea of trying not to impact existing communities though.
<f0x>
colemickens: hmmm, interesting
<joepie91>
honestly there are a handful of big issues with the bridge that are easily solvable (edits, replies), and one issue that's unsolvable due to IRC constraints (mass disconnects during network issues)
<joepie91>
I don't think the problem really is the bridging model
<joepie91>
some people like to claim it is, but that doesn't match the actual frustrations I've observed
<f0x>
so what does the heisenbridge actually do different
<samueldr>
mass disconnect with network issues could be *as solvable* as irc network links
<samueldr>
so unsolvable :)
<samueldr>
and yeah, I think it's the details of how the messages are poor lossy encoding of the original matrix messages
<samueldr>
the main issue
<joepie91>
both poor and lossy together is the problem yeah
<joepie91>
lossy is fine so long as it makes sense
<f0x>
the matrix-appservice-irc codebase does seem vastly overcomplicated tho
<f0x>
and the typescript nonsense doesn't help either..
<samueldr>
joepie91: what I assume is a trivial improvement already would be to "complete" the usernames with the irc side nicknames instead of expanded full names on the matrix side
<samueldr>
that's lossy *and* poor :)
<joepie91>
samueldr: that's actually already meant to happen but there's a bug when it combines with HTML
<samueldr>
haha
<joepie91>
f0x can tell you more
<samueldr>
multiline text probably could have an easy win by sharing the first line if it's not a pre-formatted block, so you have context when people send a long message starting with a question
<joepie91>
and yeah if it weren't for the TS use I would've PR'ed the edit/reply stuff long ago
<samueldr>
TS use?
<joepie91>
typescript
<samueldr>
ah
<samueldr>
edit/reply I don't have any trivial idea in mind, but that's also a way more complex issue
<joepie91>
the codebase was converted to typescript in an IMO misguided attempt to solve architectural issues with type annotations
<samueldr>
mass join/leave to me is a non-issue considering how netsplits exists on "healthy" irc servers like here
<joepie91>
that unfortunately moved it from the "sure let me PR that" bucket to the "ugh I don't want to deal with typescript's bullshit" bucket for me
<samueldr>
maybe it could be helped if the bridge bridged as an IRC server so the netsplit happens as a detectable IRC extension netsplit
<joepie91>
samueldr: see above links for formatting solutions
<samueldr>
though I'm sure there could be something worked out to make this work
<samueldr>
something like a less-privileged irc link for the server implementation
<samueldr>
since *anyway* AFAIUI networks already have to special-case the big matrix bridges?
<samueldr>
I'm talking in absolutel ideals terms :)
<samueldr>
hmmm. not absolute... practical ideals
<f0x>
samueldr: uhh basically pills/matrix mentions are HTML, they get transformed to the proper irc-side nick, some specific html gets parsed (italics, bold) then it runs a html sanitizer with an instant fallback to the plain body if any other non-handled html was found :/
<samueldr>
absolute would be the irc server acts less like an irc server and has more in-depth knowledge about users, but that doesn't exist
<f0x>
basically undoing the pill > ircnick conversion as well
<samueldr>
heh
<f0x>
and clients will put the displayname in the plain body
<f0x>
that code was such a fucking mess to understand
<joepie91>
this is kind of the problem with typescript; it introduces a ton of those little "ok, but why do I need to prove this to the compiler, and *how* do I do that" cases while providing no real benefit
<samueldr>
f0x: it wasn't obvious, but I wasn't serious
<joepie91>
because the type system isn't very good
<f0x>
ah :P
<joepie91>
it's sold as "JS, but with types!" but in practice lots of idiomatic JS constructs are difficult to express with it because its inference is pretty shit
<samueldr>
I guess it's good I never got the chance to actually start looking into it
<joepie91>
and I haven't verified this because I gave up on TS fairly early, but someone I know who does know both JS and TS quite well has assured me that there are constructs which are impossible to express in it :P
<__monty__>
Could everyone please stop nerd sniping us PLT wannabes? We do have work to do you know 😤
<samueldr>
it must mean they're not typesafe, joepie91, obviously ;)
<samueldr>
oh, I was wondering what not being able to push to github in one go, but in jamaican accent, meant
tomberek has quit [Quit: Connection closed]
<MichaelRaskin>
I would have a guess for cuban accent, is it close enough?
<adisbladis>
Not sure how a cuban accent sounds tbh
<samueldr>
I know only in french
<MichaelRaskin>
The accent would be applied to the process, not to sound of anything
slack1256 has joined #nixos-chat
<elvishjerricco>
Hm. lspci doesn't show my intel iris pro gpu in my macbook, just my nvidia gpu.
<__monty__>
Macbook, nvidia, 🤨
<samueldr>
elvishjerricco: should it?
<samueldr>
I mean, I don't know :)
<elvishjerricco>
__monty__: This is a 2013 model, before they ditched nvidia
<samueldr>
vintage
<elvishjerricco>
samueldr: I mean it has an iris pro gpu in it. I would assume it should show up
<__monty__>
Didn't know they ever went nvidia.
<__monty__>
My mac's from before nvidia I guess.
<elvishjerricco>
__monty__: Back then they flip flopped between amd and nvidia
<samueldr>
when they didn't strive for monocultures I guess
<samueldr>
some call it flip-flopping, I call it keeping things open and fostering competition
<samueldr>
;)
<__monty__>
I call it a ploy to get better prices.
<MichaelRaskin>
Oh, I guess I know what's the problem… apparently trying to push a git commit with date set to Unix Epoch doesn't work with GitHub
<gchristensen>
oh no
<samueldr>
epooooooooooch!
<gchristensen>
epoch+1
<elvishjerricco>
Hm. MacOS reports both gpus obviously. But under "Bus", it has PCIe for nvidia, but "Built-In" for the intel gpu. Furthermore, it says the display is connected to the intel gpu, which is the opposite of what Linux thought.
<__monty__>
MichaelRaskin: Github stores *everything* in zips? 🤔
<samueldr>
that's why I thought "should it?"... I don't really _know_ much, but I assumed that the GPU might not be technically on the PCI[e] bus
<adisbladis>
__monty__: Somehow that wouldn't surprise me..
<samueldr>
I would assume an `if date` where 0 is falsy
<MichaelRaskin>
__monty__: not zips, just zip model of world stored in MySQL
<adisbladis>
samueldr: I think that's such a design mistake :/
<samueldr>
adisbladis: conflating types?
<adisbladis>
Yeah
<samueldr>
woah there, you with your controversial opinion, stop it
<adisbladis>
Truthy/falsy non-bools
<elvishjerricco>
samueldr: Well the NixOS wiki page for nvidia prime says it should be a pci device, but maybe this one is just old and wasn't always like that
<samueldr>
elvishjerricco: ah, I see, and probably!
<adisbladis>
samueldr: That's hardly a controversial take?
<samueldr>
adisbladis: oops, I dropped all my /s on the floors :)
<adisbladis>
And my brain is in reduced functionality mode :)
<samueldr>
uh, yes, floors, multiple :)
<elvishjerricco>
The fact that the display is connected to the iris graphics on macos is throwing me for a loop...
<adisbladis>
elvishjerricco: I had the misfortune of using such a laptop many years ago
<elvishjerricco>
I added a systemd-boot entry that just runs this. Had to manually select it and then boot nixos. Is there a way to do that automatically?
__monty__ has quit [Quit: leaving]
<ar>
hm
<ar>
if the emulated pendrive was writable, and one used systemd-boot, it shouldn't be too hard to make a rotary dial with options based on ESP:/loader/entries/ contents and serving ESP:/loader/loader.conf file
<ar>
though it'd be a race between selecting the boot option, and systemd-boot starting
slack1256 has quit [Ping timeout: 260 seconds]
Dotz0cat has joined #nixos-chat
Synthetica has quit [Quit: Connection closed for inactivity]
supersandro2000 is now known as Guest37374
Guest37374 has quit [Killed (hitchcock.freenode.net (Nickname regained by services))]