<cole-h>
omg I really hate it when rustfmt breaks-but-not-really
<jtojnar>
Ugh, I deleted passwordFile, then nerdsniped myself into researching diceware implementations, during which computer OOM froze, so I had to reboot
<cole-h>
Like why does rustfmt not like that. Send help.
<bqv>
cole-h: emacs?
<cole-h>
And outside of emacs.
<cole-h>
`cargo fmt` -> no go. `rustfmt` -> no go.
<bqv>
:/
<jtojnar>
Bless boot.debug1mounts
<cole-h>
Welp, guess I'll be putting that off until some day in the future.
<ashkitten>
the cover display on this phone is so cool
<ashkitten>
but i wish it didnt lag so much
<ashkitten>
and that it was open source
<samueldr>
it's almost ironic how it hits harder when something is close to being good (or perfect) but has some flaws, compared to a complete trainwreck
<bqv>
those of you that use wayland, can you run something in a wayland compositor from outside?
<bqv>
a-la env DISPLAY=:0 xterm
<bqv>
i think yes
<samueldr>
anyone knows a fond that is open source / freely licensed, and comic sans alike?
<samueldr>
asking for a friend
<gchristensen>
mine is not ...
<samueldr>
yeah :/
<samueldr>
it actually would have been nice for the schemeful thinking I'm doing
<gchristensen>
I wonder why these workers are not around ...
ixxie has quit [Ping timeout: 258 seconds]
neeasade has quit [Remote host closed the connection]
<ashkitten>
oh right
<ashkitten>
i had the wifi and cell antennas on without being connected to anything
<ashkitten>
that's definitely why the battery was draining so fast
<ashkitten>
i think nvm about the battery issues on the cosmo, then
<bqv>
this is gonna sound insane
<bqv>
but has anyone experienced the thing where scrolling a browser window creates sound in your speakers
<bqv>
i feel like i've had it in windows and linux, and i'd love to know how the flying fsck that's possible
<samueldr>
not insane
<samueldr>
bad isolation somewhere
<samueldr>
it's not uncommon
<bqv>
but why is it specifically the act of scrolling that causes it
<samueldr>
something that happens while scrolling
<samueldr>
most likely dumping things on the display
<samueldr>
I would bet that any animation does
<samueldr>
and even intantaneous events
<samueldr>
but that wouldn't be noticeable
<bqv>
hm, i see
<ashkitten>
bqv: you can actually capture and recreate a picture being displayed on a monitor just from electromagnetic interference it emits
<ashkitten>
it works best on crts but it is possible on lcds
<bqv>
madness
<ashkitten>
by "a picture" i mean "kinda a vague blobby mess, if anything" - ymmv based on type of monitor, what connection it's using to the computer, etc
<ashkitten>
hmm, ky0ko said her best results were with an lcd display and mostly limited by the sample rate she was capturing at
<ashkitten>
so yk, ymmv but it's definitely a thing that's possible
<bqv>
heh
<ashkitten>
she says she thinks she would've been able to reproduce readable text if she'd captured directly from the sdr instead of using a 48KHz audio stream
<ashkitten>
but she wasn't able to get much from my monitors at home
energizer has quit [Remote host closed the connection]
energizer has joined #nixos-chat
<bqv>
colemickens: i got velox working
<bqv>
this is awesome
<bqv>
though firefox-wayland is fugged
drakonis has quit [Quit: WeeChat 2.8]
<colemickens>
bqv: what's velox? and what's wrong with firefox-wayland?
<colemickens>
I was going to try to do some fixups in the firefox derivations but I'm exhausted and that probably wont go well tonight
<colemickens>
oh I see, it's related to waymonad I guess
<colemickens>
btw, is there a reason bqv-flakes isn't merged to master? I am trying to keep things buildable with and without flakes and maintaining two hm branches isn't fun
<cole-h>
My guess would be because ry*ee doesn't use flakes, and probably doesn't want to rely on a third-party for flakes fixes and stuff.
<cole-h>
But idrk
<colemickens>
alternatively, if I can import the flakey version in legacy nix, that would be fine too
waleee-cl has quit [Quit: Connection closed for inactivity]
<cole-h>
Using flake-compat, that's possible.
<colemickens>
oh I had meant to loop back around to that, thanks cole-h!
<bqv>
I set my flake up to use it for shell.nix too, the other day
<bqv>
Massively simplifies my nix
<colemickens>
https://github.com/colemickens/nixcfg/blob/flakes/shell.nix I just made it so it loads my nixpkgs branches in nix-shell, or my flakes with `devShell = import ./shell.nix { nixpkgs = inputs.nixpkgs.legacyPackges; }` or something roughly like that.
cole-h has quit [Quit: Goodbye]
<bqv>
colemickens: yeah I did the reverse
<colemickens>
I guess if you use flake-compat then the shell automatically follows your flake inputs.
<bqv>
Yup
<colemickens>
cole-h: hey random do you use flakes + nixos-config w/ git-crypt ?
FRidh has joined #nixos-chat
<colemickens>
I'm also wondering about a nixpkgs-mozilla flake that isn't impure....
<colemickens>
it would be a pretty similar setup to the type of script I run for nixpkgs-wayland
<colemickens>
get the latest actual version from mozilla's update listing, then use the overlay to import that pinned version and export that as the flake output. Run that in a CI job every hour...
<colemickens>
would that be of interest to people if I threw it together?
<samueldr>
it has been discussed on this channel in the past, you should be able to put the pieces together
<eyJhb>
samueldr: logs of DOOM
<eyJhb>
Have to work instead :(
<bqv>
Hey, how awful would it be to do a git merge as part of a nix build :D
<bqv>
Probably not worth the trouble
* eyJhb
bqv do it! (just for fun at least)
<energizer>
bqv: that sounds bad
<bqv>
You just think that because you haven't heard the context
<energizer>
you just think that because you're immersed in the context
<energizer>
:P
<bqv>
Working with two forks of the same codebase, vastly compatible, but extremely large patchsets. I want to use both features, so I take the two forks and git merge them. I could do this manually, then push it, then pull again with nix, or just have nix do the merge
<bqv>
Which one sounds smarter and more declarative :p
<bqv>
Unfortunately both forks would be around a gig with leaveDotGit, so I'll just do it the boring way for now
<eyJhb>
Seems like you need to negotiate a deal, and have them use a single codebase bqv ! :D
<bqv>
It'll be fine as soon as at least one of them is merged
ixxie has joined #nixos-chat
__monty__ has joined #nixos-chat
parsley936 has joined #nixos-chat
<sphalerite>
bqv: is the native compilation thing noticeably faster?
<sphalerite>
bqv: and are you using pgtk for wayland support too?
<bqv>
Yes, and planning on it
<sphalerite>
ah right
da_dada has quit [Quit: WeeChat 2.7.1]
dadada_ has joined #nixos-chat
ixxie has quit [Quit: Lost terminal]
<eyJhb>
NOOOO
<eyJhb>
PwnTools is broken... `ERROR: No matching distribution found for unicorn<1.0.2rc4,>=1.0.2rc1 (from pwntools==4.1.1)`
FRidh has quit [Ping timeout: 258 seconds]
FRidh has joined #nixos-chat
<gchristensen>
I wish I had a shell which had features for manipulating pipes of data around, where stages were cached
<gchristensen>
example: run `sudo zfs diff hpool/local/root@blank` -> lots of output, too much. wait for it to finish, run `sudo zfs diff hpool/local/root@blank | grep -v /home/grahamc/nixpkgs` -> first command's data is already cached, so it just runs the second. realize I wanted it sorted, run `sudo zfs diff hpool/local/root@blank | grep -v /home/grahamc/nixpkgs | sort` first two sections are already
<gchristensen>
cached, just run the sort
neeasade has joined #nixos-chat
<adisbladis>
gchristensen: Time to make a new shell ;)
<adisbladis>
The fully-resolved pipeline shell
<gchristensen>
:D
<adisbladis>
Tbh I've thought about forking Xonsh
<Valodim>
I believe I could hack that together in zsh
<Valodim>
interesting idea, might give it a shot
<gchristensen>
TIL `sudo zfs diff hpool/local/root@blank > myfile; tree --fromfile ./myfile`
<adisbladis>
Valodim: I absolutely despise bourne syntax
<adisbladis>
So for me some bash/zsh solution is no good
<Valodim>
zsh is ok for me interactively, but I recognize that's only because I've invested the time
<Valodim>
sh or bash are terrible
<adisbladis>
I think fish goes a long way in making things better for interactive use
<adisbladis>
Actually having built-in multiline editing is incredibly powerful
<Valodim>
it probably does, but it's not different enough to warrant a switch for me
<adisbladis>
I was using zsh before fish and tbh it feels like a half-baked toy in comparison
<adisbladis>
Getting rid of the bourne compatbility is a big part of that
<manveru>
gchristensen: so there was a problem with your change to verify a path exists... it can't handle names that starts with dots :)
<gchristensen>
oh interesting
<gchristensen>
could wrap it in a `builtins.path { ... name = "verify"; }` ?
<gchristensen>
hrm
<gchristensen>
is there something like `tree` but interactive?
<manveru>
ranger
<philipp[m]1>
adisbladis: Funny, I've seen that toy argument only seen the other way around so far.
<gchristensen>
oh, importantly, `tree --fromfile` ha
<philipp[m]1>
I can see how getting rid of bourne compatibility could be very freeing though.
<__monty__>
Valodim: You should give fish a chance anyway. Just the history completion in and of itself is amazing. But you have to build up your history before you can see the effect.
<philipp[m]1>
What makes fish history completion better than zsh?
<philipp[m]1>
(Call me fish-curious)
<Valodim>
__monty__: I built my daily driver mail client out of zsh. I think I'd need something with a more radical approach than fish to make the switch worth it :)
<gchristensen>
you should give it a go, philipp[m]1
<gchristensen>
I don't use fish but it is truly impressive
<Valodim>
shells are a prime example of those "I don't use it myself because I'm too invested in this other thing, but without this context it's the better choice for sure" situations
<philipp[m]1>
Very well put.
<Valodim>
although I'm fairly sure it takes thousands of hours of investment to make that true for zsh
<philipp[m]1>
Thousands of hours are not that much for a shell...
<Valodim>
they are if you discount the pure usage hours ;)
* eyJhb
just sits in silence and uses bash
<Valodim>
now that, I would recommend to noone. but maybe I've just never seen anyone use bash in a proficient manner
<adisbladis>
philipp[m]1: I mean purely for interactivity
<adisbladis>
zsh is more "competent" than fish in many ways
<adisbladis>
philipp[m]1: And re history fish has something I haven't seen in other shells, namely auto suggestions
<adisbladis>
start typing `ssh` and whatever ssh command you last typed will be suggested, press C-f or arrow-right to complete
<adisbladis>
It's different than tab completion because it's history aware
<philipp[m]1>
zsh can do that.
<adisbladis>
It's also aware of files
<adisbladis>
So it won't suggest a thing I did previously if the same file is not present
<adisbladis>
philipp[m]1: zsh can do most things fish can, if you have infinite tinker time
<adisbladis>
I've been using fish for well over a decade now and my customisations are still on the order of 30 LOCs
<adisbladis>
The only thing I have customised is my prompt
<adisbladis>
I think that's a testament to how awesome it is
<__monty__>
philipp[m]1: No, zsh can only pull of a mockery of fish's history completion but you really have to give it an honest try to see.
<philipp[m]1>
Yeah, that's really awesome.
<__monty__>
Another detail is the history also depends on the current directory.
<adisbladis>
zsh is really awesome, and I don't think fish would be nearly as awesome as it is if zsh hadn't been around to show what's possible
<__monty__>
Zsh is probably also far nicer for non-interactive use.
<adisbladis>
But I think zsh also showed what NOT to do
<__monty__>
But fish is so much faster and user-friendlier it's no-contest for interactive use.
<adisbladis>
It's very much the anti-fish in some ways, like how bare bones the OOTB experience is and how much you need to configure it to become usable
<philipp[m]1>
The fact that oh-my-zsh exists is a good argument for that.
<Valodim>
oh-my-zsh was so terrible :) the guy who made it had no idea what he was doing technically for the longest time
<Valodim>
it's ok these days I think
<Valodim>
he did have the marketing part down for sure
<philipp[m]1>
I didn't say that it was good ^^
<adisbladis>
I remember when someone made oh-my-fish and the pretty much universal reaction in the community was "huh, why do we need this"
<Valodim>
haha
<philipp[m]1>
Are more colours in a shell not automatically better?
<philipp[m]1>
Also I need at least 5 lines prompt because who cares about history?
<philipp[m]1>
Oh, nice ** globbing.
<sphalerite>
eyJhb: me too lol
<sphalerite>
I used to use zsh but gave up eventually
<sphalerite>
part of the reason is that bash is just everywhere, so I don't have that "downgrade feeling" when I ssh into some server where fish isn't the default shell for one reason or another
<__monty__>
sphalerite: That's a bit like never riding a bike because at some point you'll have to get off : D
<tokudan>
__monty__, question is what bothers you more: A) not being perfectly fluent in the only available shell on a system and thus wasting time and being annoyed at the shell or B) knowing that you know the ins-and-outs of the shell you're currently using, because it's the same on every system, but might miss out on a little comfort
<__monty__>
tokudan: Thing is it's not a "little" comfort. You probably spend more than half your time on systems where you control the shell you use. And for any complex tasks you should be writing scripts anyway. So you don't have to know the ins and outs of bash every time you log into a system.
<tokudan>
__monty__, not for me, I spend most of my time in shells that I do not want to customize too much because they are also used by others. and moving around directories is a lot of muscle memory with tabbing. zsh uses different tabbing, so if someone pushes in zsh as a root shell, that actually annoys me, because it slows me down
<tokudan>
that's why I mentioned that there are two sides. both are completely valid and just depend on your own viewpoint
FRidh has quit [Quit: Konversation terminated!]
<__monty__>
tokudan: Sure, you always have to consider your own situation. But ime most people'd rather stick with bash/zsh than give fish a chance and in the process they lose so much more time.
<__monty__>
Not changing is what's comfortable and people are absolutely horrible at integrating cost over time.
<tokudan>
that's true, maybe fish and zsh offer a lot more comfort and I don't mind having it installed on systems I use. the problem only appears where people from different areas clash. one area cannot migrate for some reason, another wants the new shell as default everywhere. then the cost boils down to "who has to type in their shell name after login"
waleee-cl has joined #nixos-chat
xd1le has joined #nixos-chat
<adisbladis>
This is also why I really, really want to use eshell way more
<adisbladis>
As it works over SSH
<adisbladis>
And I can get my same configuration everywhere regardless of what's installed on the remote
<adisbladis>
But shell has sharp edges
<adisbladis>
eshell*
kalbasit has joined #nixos-chat
drakonis has joined #nixos-chat
Jackneill has quit [Ping timeout: 256 seconds]
Jackneill has joined #nixos-chat
ixxie has joined #nixos-chat
kalbasit has quit [Remote host closed the connection]
<sphalerite>
__monty__: I think it's more like having a USB charger plug rather than a barrel plug :p
<sphalerite>
__monty__: I mean, I _would_ love to see a better shell becoming the standard that bash currently is
<__monty__>
I don't want to know how much time worldwide is wasted flipping over USB type A cables.
<__monty__>
So few people learn it's just USB logo up (modulo translation of which side means up in laptops and macs).
<__monty__>
I dun goofed.
<__monty__>
That's what I get for not using image previews.
<philipp[m]1>
Oh! Never heard of xxh, that might actually help me with an annoyance I have.
<philipp[m]1>
Ah, probably not.
<philipp[m]1>
adisbladis: I also fooled around with eshell. I don't think it's for me. A little too much like eating your toenails in public.
kalbasit has joined #nixos-chat
cole-h has quit [Quit: Goodbye]
<Valodim>
I'd be really interested looking a skilled eshell user over the shoulder for a while
<philipp[m]1>
Thet tramp feature really is pretty sweet.
<Valodim>
I believe it
<adisbladis>
philipp[m]1: Lol, nice reference
<bqv>
I mean I see no reason why xonsh can't be hacked to have trampy behaviour. Rather than execute a command, shell out to ssh
<bqv>
I've been using emacs pretty well as a shell substitute, but the most uncomfortable part is then how do I ssh into my pc, do I run emacsclient on my phone? Or fall back to a shell
<adisbladis>
bqv: Nice, "trampy behaviour" ;)
<bqv>
:p
<adisbladis>
Slut shell, connects everywhere
<adisbladis>
I've been trying to run emacs on my phone recently
<adisbladis>
But the IO is just too damn slow
<adisbladis>
It grinds to a halt when doing pretty much anything :/
<adisbladis>
Even flycheck bogs it down by a metric tonne
<bqv>
Yeah I wouldn't run it native, I use the server
<neeasade>
bqv: you can shell-mode to a tramp dir
<neeasade>
I set default-directory to a tramp value and then shell-mode there
<bqv>
neeasade: never gonna trust emacs shells that aren't vterm, comint or async-shell-command… and at the point of vterm, why use emacs :p
rardiol has quit [Remote host closed the connection]
<neeasade>
agree -- but vterm sucks :P
<neeasade>
shell-mode is comint
<adisbladis>
Show me a better terminal emulator than vterm
<neeasade>
shell-mode/comint allows me to edit the buffer as text and to muliline input edits -- a luxury I miss much in vterm
<adisbladis>
I've said this before, shells are a race to the bottom technology
<bqv>
Shell activity by nature doesn't lend itself to emacs's architecture
<adisbladis>
Lowest common denominator
rardiol has joined #nixos-chat
<bqv>
I'd rather use a shell properly, or just use emacs properly
<pie_>
adisbladis: small interjection, what is better than shells?
<adisbladis>
pie_: As an emacs zealot, emacs is better :3
<neeasade>
bqv: I disagree -- if a shell is a repl, it's fine. if a shell maks calls that anticipate a canvas (cursor movement) it is no longer "just a shell"
<bqv>
neeasade: ok good luck using anything more than busybox :p
<neeasade>
maybe we are talking about different things
<neeasade>
I'm talking about scooping shell interaction up into emacs
<energizer>
dont confuse shells with terminals
<neeasade>
> adisbladis: I've said this before, shells are a race to the bottom technology
<{^_^}>
error: syntax error, unexpected ',', expecting ')', at (string):318:34
<neeasade>
I agree
<bqv>
By using a shell properly, I mean don't try and pull text into emacs - either vterm or native term. By using emacs properly, I mean don't even abstract the shell part, natively use async-shell, comint-run, dired, etc, so emacs essentially is your shell, but way more than a repl
drakonis has quit [Quit: WeeChat 2.8]
<bqv>
I've been trying the latter, its minorly uncomfortable due to old habits, but it works
<bqv>
Just, mobile becomes hell
<bqv>
Anything inbetween those, seems a poor situation, imo
<neeasade>
bqv: I guess I'm having trouble visualizing the latter looks like -- how do you retain state? for example, if you have a shell function that you want to use in some async-shell call
<neeasade>
or are you suggesting depending on a shell function is lowering yourself to the level of the shell in the first place
<neeasade>
ie, a sin
<bqv>
neeasade: yep. I'd use emacs functions, or bust
<neeasade>
I see -- and then comint-run gives you a buffer for things you expect output from
<neeasade>
hmmm
<neeasade>
this is an interesting concept -- I think I resisted that route because I "embrace the shell" a little, to perhaps the detriment of an "all emacs" way
kalbasit has quit [Ping timeout: 258 seconds]
kalbasit has joined #nixos-chat
<neeasade>
the closest I am to what you describe right now is an eval key i have for putting things in the last shell-mode buffer and running it -- but switching that so async-shell or comint-run might be nice
<neeasade>
bqv: thanks for the food for thought, at least
<bqv>
:D
<pie_>
theres something to say for consistency in general
<pie_>
but acheiving it can be hard
<Valodim>
say the folks in the tryhard consistency software channel :)
<neeasade>
Valodim: LOL
<pie_>
and points to the usual tradeoff of whether you want a heterogeneous or homogeneous system (say, using multiple languages in a codebase)
<pie_>
Valodim: :'D arguable
<bqv>
Valodim: lmao
<__monty__>
Does anyone have a reference for how wide st_dev, st_inode and st_mtime are on popular platforms?
<pie_>
i watched two coworkers (not mine) at chaos congress have a friendly argument about using nix as glue code for a tool written in like 5 different languages, one party did not want to see something like that at work
<pie_>
the other party was definitely not profpatsch :P
<pie_>
oh damnit i cant highlight profpatsch in here because he isnt in this channel
<neeasade>
bqv: hilariously I went the other way even -- made the shell hug emacs with an 'elisp' wrapper for eval'ing on the server
<neeasade>
> __monty__: Does anyone have a reference for how wide st_dev, st_inode and st_mtime are on popular platforms?
<neeasade>
pls pingback if you do find one
<neeasade>
curious
<bqv>
Irc callbacks, heh
<__monty__>
neeasade: Going through apple headers, st_dev -> dev_t -> int32_t, st_ino -> ino_t -> u_int32_t and mtime -> timespec -> long (8 bytes) + int32_t
<neeasade>
__monty__: tybb
<__monty__>
neeasade: Though, when using sys.getsizeof(s.st_dev) in python3 on a mac I get 28 o.o
<__monty__>
Oh, whoops, that's the size of the pointer to the object.
<bqv>
Beware pointers, theyre sharp
<eyJhb>
bqv: "pointers, screwing with you since always"
<bqv>
:D
<__monty__>
neeasade: For linux I'm finding u32, long and long+long respectively.
<bqv>
u long long, me short short
<eyJhb>
I just love a language that tells me when I am about to fuck up
<__monty__>
I was inaccurate about the python thing, that's the size in bytes of a reference counter, a pointer to the class and the value.
<__monty__>
In CPython specifically.
<gchristensen>
I wonder if qemu can emulate a failing drive
<__monty__>
st_mtime is converted from a timespec of long+u32 or long+long to a double. sizeof reports 24 bytes, at 8 for the reference count and 8 for the class pointer that's 8 bytes which corresponds to a double. st_dev and st_ino otoh return 28 bytes, which leaves 12 bytes after subtracting referenc count and the point, which doesn't make sense to me.
<worldofpeace>
vocaloid would have been an upgrade
<samueldr>
hm, no preview of the fr-CA ones
parsley936 has quit [Remote host closed the connection]
parsley936 has joined #nixos-chat
ixxie_ has quit [Quit: Lost terminal]
<worldofpeace>
could've layered pitch shifter with some unique vocoder, distortion, pitch correction if u really want to just hide your voice. that would be unrecognizable
<worldofpeace>
100% digital
<infinisil>
samueldr: I have the thing setup, I could paste the result for specific voices and texts
<samueldr>
Chantal, AFAICT the only fr-CA, same text as the fr-FR sample is probably best, for comparison sake
<samueldr>
yeah, not being a native hindi speaker that's not useful :), I meant that in french (at least here) if it knows where to use french or english pronounciation it can be a killer feature
* infinisil
nods
<samueldr>
or for some technical backgrounds
<samueldr>
good luck finding a developer that knows all the néologismes (new words) from the OQLF :)
<samueldr>
if you say infonuagique (cloud computing) you're gonna get laughed at!
<samueldr>
(an english adaptation of the intent could be "cloudputering")
<pie_>
lol
<samueldr>
and now I somehow want to say cloudputering
<infinisil>
Hah
<pie_>
#shitnixoscoined
<pie_>
i never say cloud computing
<pie_>
but if im going to
<pie_>
im going to say cloudputering
<samueldr>
out with "it's in the cloud", in with "it's on the cloudputer"
<pie_>
if samueldr isnt already. i want him on the marketing team asap
<pie_>
this could be big
<samueldr>
using NixOS on the cloudputers
<samueldr>
cloudputer-enabled deployment with NixOps
<pie_>
yesssss
<pie_>
folks you heard it here first
<samueldr>
now, another neologism I will intently translate: courrier électronique ⇒ courriel, electronic mail ⇒ electromail
<samueldr>
(though e-mail is already kinda that)
<pie_>
its not cloudputering but ill take it
<samueldr>
though it feels that if you were going to make a pseudo retro-futuristic piece set in 2020, but as if it was conceived in 1990, you could use names translated from OQLF neologisms for technologies and it'd sound just foreign enough
<pie_>
just translate everything electronic or e- into electro
<samueldr>
e-waste, electrowaste
<pie_>
electroca
<pie_>
electrocar
<pie_>
electro...s? instead of electronics :P
<samueldr>
sadly we don't have a good néologisme for smartphones, it's just that, "téléphone intelligent" "smart phone"
<pie_>
the only reason smart phone sounds normal is because youve been hearing it for a long time now
<energizer>
key board
<pie_>
its "smatr <tech>" its already bad
<pie_>
*smart
<samueldr>
for smartphone, it's not that foreign, I think
<samueldr>
if you know what a phone is, and know of some technology smartphone makes sense
<pie_>
advertisement quote: "with the inter net, I can connect my smart phone to my cloudputer while driving my electrocar. In fact, with Aye Eye, I don't even need to drive!"
<bqv>
cause that makes more sense since i already tag
<drakonis>
but there's some annoyances like having to commit a flake lock bump now
<drakonis>
and it seems like doing it that way would be fairly git dependent
<energizer>
bqv: what does it say instead of "version Generation 1 NixOS 20.03.2262.2b417708c28, Linux Kernel 5.4.46, Built on 2020-06-22" under /boot/loader/entries?
<drakonis>
basalt looks hella sick
<drakonis>
gonna use that
<bqv>
oh my setup is great for that. see, i'll commit, my nixos script will run, update the flake lock, then do the full build but bail out at the end cause the repo's now dirty
<bqv>
then i just amend the commit and it's done in a second or two
<cole-h>
I think I'm getting close to switching to flakes. I kinda wanna see the blog post for setting up a system using it though.
<drakonis>
i'm importing my system config into the flake
<energizer>
drakonis: seems like more machinery than necessary for the outcome
* cole-h
is kinda sad that it was promised as "the next blog post" but actually wasn't ;'(
<drakonis>
working wonders, but no automation
<bqv>
energizer: the same, except that 2262.blah is now 20200707.36b60d8
<energizer>
cole-h: yeah i've been waiting on that blog post...
<drakonis>
energizer, well, it'd serve the purpose of being vcs independent
<bqv>
energizer: before dot is todays date, after dot is commit ref
<drakonis>
for the people who dont like git a whole lot but want something else to go
<energizer>
bqv: that's the commit of nixpkgs, tho right? not of your config
<drakonis>
to use as their vcs of choice, ie: mercurial, darcs, pijul
parsley936 has quit [Remote host closed the connection]
<bqv>
again, not that i've used this feature
<bqv>
but hey, i did it, and that was fun
<energizer>
bqv: mine is the hash of nixpkgs tho? (assuming i havent messed with that)
<bqv>
yes
<infinisil>
Question for flakes: Can flakes be pinned and fetched recursively? E.g. can foo have a flake.nix that depends on bar and bar have a flake.nix that depends on baz?
<infinisil>
And both be pinned
<bqv>
the entire flake dependency tree is pinned in flake.lock
<bqv>
oh you mean pinned in the flake.nix
<infinisil>
So bar couldn't have its own flake.lock that applies?
<energizer>
bqv: how do i get that info into mine?
<bqv>
infinisil: it will by default, you can override it in your flake.nix or at runtime with --override-input
<infinisil>
Hmm
<bqv>
at that point you'd have to create it as an input to your flake and just do .follows at the appropriate level
<bqv>
e.g. inputs.foo.follows = "yourpin"
<bqv>
e.g. inputs.foo.inputs.bar.follows = "yourpin"
<bqv>
e.g. inputs.foo.inputs.bar.inputs.baz.follows = "yourpin"
<bqv>
energizer: just set NIXOS_LABEL where nixos-rebuild can see it
<bqv>
it's in the manpages, somewhere
<infinisil>
bqv: Hm I see somewhat
<energizer>
bqv: cool, thanks
<gchristensen>
sigh, no password on the root account means I can't log in to the Emergency console
<infinisil>
So, in an ideal flake world, all things that need to be fetched for evaluation would be declared in flake.{nix,lock}, meaning even if you have a project depending on 100 Nix expressions recursively, the flake.lock would pin all of them to each of the versions they have pinned themselves