gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
jasongrossman has quit [Ping timeout: 264 seconds]
jasongrossman has joined #nixos-chat
pie_ has quit [Read error: Connection reset by peer]
endformationage has quit [Quit: WeeChat 2.4]
drakonis_ has joined #nixos-chat
drakonis has quit [Quit: WeeChat 2.3]
jtojnar has joined #nixos-chat
jtojnar has quit [Client Quit]
jasongrossman has quit [Ping timeout: 268 seconds]
<colemickens> MFW renting a server for a dime an hour and the specs are: (2x28-Core@2.2GHz + 384GB of RAM) = :O
<srhb> colemickens: The question then becomes: How quickly can you provision/destroy it? :D
<colemickens> srhb: takes a good while to bring back up and repopulate the store/cache, 20 minutes maybe
<colemickens> but it's fine when it costs $2.40 to keep the thing up for a day
<colemickens> plus, when you're building chromium, there's no downtime anyway /upside-down-smiley
Myhlamaeus has quit [Quit: WeeChat 2.2]
<jD91mZM2> joepie91: Sorry, that seems like an oversight. Mind making a GitHub issue?
johanot has joined #nixos-chat
<jD91mZM2> joepie91, emily: Yeah, rnix is slow because of rowan. I made a rowan-like library that uses an arena-based allocator that is a lot faster (forgot how much, at least 2x but probably more), but the rowan dev thinks there are other, more pressing issues at the moment
<jD91mZM2> Writing lots of allocations in JS is fine because of the garbage collector. Writing it in Rust brings it down to JS in speed.
<jD91mZM2> Especially now when they changed the default allocator to the system one and not jemalloc.
<andi-> I remember writing custom allocators in C++. Does Rust offer that yet without touching all the sources?
jasongrossman has joined #nixos-chat
<andi-> Ahh I see it is but very fine grained. All or nothing.
goibhniu has joined #nixos-chat
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.1)]
jasongrossman has joined #nixos-chat
cransom has quit [Ping timeout: 258 seconds]
__monty__ has joined #nixos-chat
<joepie91> jD91mZM2: aha, so I'm not holding it wrong! :)
<joepie91> yeah, I'll create an issue for the identifiers thing later then
waleee has joined #nixos-chat
drakonis has joined #nixos-chat
jtojnar has joined #nixos-chat
johanot has quit [Quit: WeeChat 2.4]
<emily> jD91mZM2: heh, I'm torn -- on the one hand, writing Rust like it's JS is kind of weird, on the other hand I kinda wish Rust had a modern GC so you could opt out of the bondage & discipline with less penalty >_>
<simpson> emily: It's almost as if Rust is only appropriate for certain tasks~
<emily> shrug; it has strong enough abstraction facilities that it could extend its range without too much effort
<emily> it had far more high-level niceties and intention to implement fancy GC years back
<simpson> Yeah. I remember those times. Good words, good intentions.
<emily> general-purpose programming languages are hard
<emily> it's a lot more of a high-level functional language than C++ at least :p
<jD91mZM2> Somebody wrote a garbage collector in Rust actually!
<emily> yeah, but it's not exactly world-class <.<
<jD91mZM2> The rowan dev says parsing performance doesn't really matter - and he's not wrong. rnix should implement cached parsing and stuff like that to *actually* be faster and not just get some nice perf numbers.
<emily> I disagree; parsing performance matters for many applications.
endformationage has joined #nixos-chat
waleee has quit [Quit: WeeChat 2.4]
<etu> Hmm
<etu> What should I talk about on the opening speech on my emacs meetup on wednesday? :D
<srhb> Clearly evil-mode
* srhb ducks
yl has joined #nixos-chat
waleee has joined #nixos-chat
drakonis has quit [Quit: WeeChat 2.3]
<mdash> etu: declarative emacs config with with home-manager :)
<etu> mdash: adisbladis covered that the last time :D
<mdash> etu: oh well i guess it was obvious :)
<sphalerit> I love `zfs inherit`. It's like cleaning, but with the results and without the effort.
<etu> mdash: I might just choose EXWM as a topic :D
<etu> I first thought is was too niche, but well :D
<goibhniu> are there videos of those talks by any chance?
<etu> goibhniu: Nope :(
<goibhniu> ah, pity
<elvishjerricco> sphalerit: How's that?
<sphalerit> elvishjerricco: it removes the "this is different" data :p
<elvishjerricco> Fair enough :P
<mdash> etu: emacs has a lot of niches
<mdash> etu: "org-mode and calc as a spreadsheet replacement" is another one I like ;-)
waleee has quit [Quit: WeeChat 2.4]
<etu> mdash: yep :-)
<etu> mdash: Last time I spoke about magit
<mdash> oh, i forget some people don't know about magit yet
waleee has joined #nixos-chat
<etu> yeah
<elvishjerricco> magit is amazing. It's at least 50% of the reason I won't use other editors.
<srhb> I miss magit.
<srhb> It really is great.
<waleee> do the emacs clone of gitgutter (or magit) support adding hunks yet? Using interactive always feels a bit clunky
<waleee> (it does)
<elvishjerricco> Yea hunking with magit is 1000x better than on the command line
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 246 seconds]
drakonis_ has joined #nixos-chat
drakonis has quit [Read error: Connection reset by peer]
<ldlework> magit is top-shelf
<ldlework> etu, maybe do a tutorial on using Hydra to "magit-ize" everything
<ldlework> (or magit-popup, w/e)
<etu> ldlework: Already writing my slides on EXWM :-)
<ldlework> etu, something about using a single-threaded WM gives me the jeebies.
<mdash> ldlework: ... isn't that most WMs
<ldlework> EXWM will just lock up for periods of a time
<ldlework> no other WM I've used in the past has that problem
<mdash> that would be a problem yes
<ldlework> tbh, even then it was very hard for me to decide whether that was a big enoug problem to leave EXWM behind because it is otherwise pretty awesome
<elvishjerricco> What do hydra and exwm have to do with magit?
<ldlework> are you gonna tell us?
<etu> ldlework: Well, yeah, that's one of the weak points
<etu> ldlework: But emacs since 26.1 has some kind of light threading
<ldlework> etu, yeah but it is really annoying to use
<ldlework> oh wait I'm thinking of the async package I think
<etu> the async is different
<etu> because that launches a emacs in batch mode with a payload :D
<etu> But yeah, since 26.1 they have some threads ish
<ldlework> yeah
<ldlework> if EXWM ever stops freezing up I'll definitely take another crack at it
<ldlework> it's just too cool to have Emacs as your WM
waleee has quit [Quit: WeeChat 2.4]
goibhniu has quit [Remote host closed the connection]
<noonien> does anyone know what relation the 16 terminal colors have to eachother? and which ones are used for text and such?
<joepie91> noonien: that is dependent on ansi escape code that the application uses
<joepie91> terminal colors are a palette that the application can take colors from basically
<samueldr> though it is (was?) kinda expected to have the "bright" colours be the same as the non-bright one, but brighter
<ldlework> it's sort of standard as far as terminal emulators go
<ldlework> the first color is background
<ldlework> the last color is foreground
<noonien> yes, i know, i'm more curious as to what they're usually used for
<samueldr> (and the colours to match a general palette)
<ldlework> the first 7 are "bright" the second 7 are "dark"
<samueldr> worst that will happen when the palette isn't followed is you'll get blue text when the author of the program expected red (for example)
<noonien> i'm using wpgtk/pywal to genrate palettes, an am trying to understand why it choses the colors it does, my vim theme seems to choose an odd sets of colors for text and i'm trying to understand if the theme i'm using is wrong, or the color palette generator just spewed somme bad values
<noonien> it seems the error texts are red, which implines the palette/theme is correct
<noonien> but i get some text that is yellow, that shouldn't be (directories in my case)
<noonien> and it looks rather odd
<noonien> the colors also seem to be identical for the light and dark "row"
<ldlework> noonien: I just shuffle the palette a few times until it looks good
<ldlework> I have global key bindings to all of wpgtk's functions so I can modify the current theme from anywhere :)
<noonien> yeah, i'll probably do that as well, thanks!
<ldlework> huh my dns died
<noonien> i made https://github.com/muxro/cf-ddns for cloudflare, maybe it'll help
<noonien> i think cloudflare might even have some more standard ddns clients that it works with, if you're running on a router
<ldlework> dns is working again :)
<__monty__> I thought the first colors were the dark/regular ones? So programs that aren't compatible with brighter versions don't look all neon.
<__monty__> base16 perverts the 16 colors btw. Found this out recently, the "bright" colors are replaced by shades of grey apparently. Don't ask me why.
<__monty__> Bright is most often used by programs instead of boldface because most terminals don't support boldface.
<jD91mZM2> joepie91: Skip the issue if you haven't already made it: I pushed a fix to master (although GitLab is down)
<joepie91> jD91mZM2: ah, I hadn't - thanks :)
* joepie91 has been swamped in work today
<jD91mZM2> Also notified the rowan guy about your performance concerns (not so much yours as emily's), and plan to make time to look into ways to optimize it myself (without doing the whole arena-based allocator that didn't lead anywhere last time)
drakonis has joined #nixos-chat
drakonis_ has quit [Read error: Connection reset by peer]
<joepie91> good to hear!
<andi-> oh-ho, http://nix.dev does exist now..
<andi-> A few days ago it was sold but didn't point anywhere.
<sphalerite> wtf, it's not nix-related at all :(
<samueldr> someone thinking unix relatives as *nix I guess, and hopping on the short domain name bandwagon
<andi-> yeah, but 500$? O.O
<samueldr> "I bet I can sell it for even more!"
<andi-> never worked for my collection of names.
<samueldr> (oblivious to the fact that it's a reverse dutch auction and no one bid beforehand)
<andi-> even the 4 letter (and one dot) domain I had wasn't selling :D
<flokli> never sold any of my pet domains either :-D
<sphalerite> quick someone buy nix.tech
<samueldr> looks like it's 150$CAD today for a 3 letters .dev
<sphalerite> (I already have nixos.tech, but nix.tech was expensive :p )
<samueldr> it's already taken
<sphalerite> oh
<samueldr> ah
<samueldr> >> This domain has been reserved by the registry, and is not available for registration.
<samueldr> (nix.tech)
<andi-> I currently only own nixos.{cloud,dev}, the first I only once used for a nixos workshop where I hosted VMs on subdomains.. The other one I decided to "secure" in case some troll buys it. Available for good projects.
<sphalerite> samueldr: huh, last time I checked it was available
<samueldr> just now used `whois` to check, and google domains too said it was unavailable
<sphalerite> samueldr: that was a month or two ago though
<sphalerite> andi-: redirect them to nixos.org until you have a better use for them?
<andi-> sphalerite: might do
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 258 seconds]
<gchristensen> andi-: ah ha, you got nixos.dev!
<andi-> ^^
<andi-> As I said up for grabs / good projects.
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 245 seconds]
drakonis has quit [Read error: Connection reset by peer]
drakonis has joined #nixos-chat
drakonis has quit [Ping timeout: 264 seconds]
mkaito has quit [Remote host closed the connection]
mkaito has joined #nixos-chat
jasongrossman has quit [Remote host closed the connection]
aanderse has quit []
__monty__ has quit [Quit: leaving]
jackdk has joined #nixos-chat
jackdk has quit [Remote host closed the connection]
yl has quit [Ping timeout: 244 seconds]
jasongrossman has joined #nixos-chat
<elvishjerricco> Hm. I cached 4GB of random data into a file in a tmpfs (with plenty of memory leftover), then dd'ed it onto my root file system, and found that it wrote at ~80 MB/s. Shouldn't a Samsung 960 Pro write way faster than that?
<elvishjerricco> Obviously it started way faster (like 2.2 GB/s) because of the OS caching the writes, but it slowed to 202 MB/s over time. I used `conv=fsync`, and once the final sync was done, it reported a final average of 80 MB/s
<tilpner> elvishjerricco: What blocksize did you choose?
<elvishjerricco> 8M
<emily> do you have filesystem-level compression turned on? hopefully shouldn't slow it down that much, but maybe if you have it set to gzip or something...
jackdk has joined #nixos-chat
<elvishjerricco> Hm, yea I do have `compression=on` in ZFS
<emily> you could try something less pathological than random data, although it still seems iffy to me that it'd end up that slow, I think ZFS defaults to lz4?
<emily> also you should probably do it twice: the first time it might have had to fsync a bunch of other stuff
<elvishjerricco> Yea I've tried it a few times. The CPU usage doesn't appear to be high during sync
<elvishjerricco> Just tried a large video file. Roughly the same write speed
drakonis has joined #nixos-chat
<elvishjerricco> I wonder if LUKS has anything to do with it