09:04 <re_irc> <nihal.pasham> I'm using the "log" crate for some quick testing on the raspberry pi 4 (which I believe rpi4 has support for atomics). The "impl" is a copy-paste from the crate's documentation.

09:04 <re_irc> use log::{Level, Metadata, Record};

09:04 <re_irc> ... long message truncated: https://psion.agg.io/_matrix/media/r0/download/psion.agg.io/29cc8774d85848f61e004036eeda382c3052a74b0a059b0f38977b81e7ca99a5 (31 lines)

09:04 <re_irc> use rustBoot_hal::info;

09:04 <re_irc> use log::{LevelFilter, SetLoggerError};

09:22 <re_irc> <David Cabanis> ls

09:27 <ni> Cargo.lock Cargo.toml README.md src/

09:27 <ni> :)

09:57 <re_irc> <nihal.pasham> the thread-unsafe version (i.e. "set_logger_racy") works as expected. I wonder why the safe version does not work. My guess is - probably has something to do the MMU being disabled. I'm writing some bare-metal stuff that doesn't need the MMU.

12:57 <re_irc> <chmanie> I'm looking at the embedded_time library here and am trying to format a frequency: https://docs.rs/embedded-time/latest/embedded_time/rate/trait.Rate.html#formatting but it seems the "format!" macro is not available to me. How could I obtain it?

13:00 <re_irc> <Martin> maybe you could use "write!" instead

13:01 <re_irc> <chmanie> Hm, where would that come from?

13:03 <re_irc> <chmanie> Ah it only says it's ambiguous

13:04 <re_irc> <Martin> "use core::fmt::Write;"

13:10 <re_irc> <chmanie> Okay thank you!

13:13 <re_irc> <chmanie> It seems I have to implement the formatter: https://stackoverflow.com/a/50201632/1525722

13:14 <re_irc> <chmanie> I mean it's weard, embedded_time is specifically targeting no_std, why would format! be in the docs

13:14 <re_irc> <chmanie> * weird,

13:15 <re_irc> <Martin> also in the super cool heapless crate you have a String type to allows you:

13:15 <re_irc> let mut out: String<U256> = String::new();

13:15 <re_irc> let my_var = Type::new()

13:15 <re_irc> write!(&mut out, "{}", my_var).unwrap();

13:18 <re_irc> <chmanie> Ah yeah I forgot about heapless!

13:20 <re_irc> <chmanie> But I also just saw that there's an "integer()" method on the Rates

13:21 <re_irc> <chmanie> Thanks Martin!

16:28 <re_irc> <ryan-summers> Question: If you have an array of "[MaybeUninit<u8>; N]", is it sound to transmute that memory into an array of "[u8; N]" since all memory values are valid for "u8"?

16:30 <re_irc> <ryan-summers> I essentially just want that uninitialized u8 array as a "&mut [u8]" slice so I can pass it to an external function. I don't have control to make that external function take "&mut [MaybeUninit<u8>]" instead

16:33 <re_irc> <ryan-summers> I guess you can do "let array = uninit_array.map(|x| unsafe { x.assume_init() });" as well. Unsure if that compiles into any actual instructions

16:41 <re_irc> <ryan-summers> Follow up: No, it's not valid to do, and I always forget this

20:28 <re_irc> <dngrs (spookyvision@github)> it sure is slightly irritating that you get nonsensical missing stdlib build errors when you just haven't done the required "rustup target add whatever-target-triple"

20:29 <re_irc> <dngrs (spookyvision@github)> ryan-summers: hmm, I suppose because the compiler will be like "yeah this is invalid memory so I can do UB in any way I want" when you try to access it?

21:17 <re_irc> <jamwaffles> They really meant min when "min_const_generics" was stabilised :((((

21:17 <re_irc> Is there a way to do "let buf = [0u8; N / u8::BITS]" where "const N: usize" on stable rust?

21:19 <re_irc> <dirbaio> no :(

21:20 <re_irc> <jamwaffles> https://psion.agg.io/_matrix/media/r0/download/matrix.org/vPbPnhhdUadVyZQRLerOzajj/image.png

21:21 <re_irc> <jamwaffles> It's beautiful on nightly with "#![feature(generic_const_exprs)]" 😍

21:21 <re_irc> <dirbaio> you can kinda workaround it by having the user pass in both N and N/8

21:22 <re_irc> <jamwaffles> Yeah that's what I'm doing for now. This struct will end up with 4 const params which is a lil uggo

21:22 <re_irc> <dirbaio> and maybe do some hax to assert the user did it right at compile time..

21:23 <re_irc> <dirbaio> or have the top-level API be a macro that passes all the params

21:23 <re_irc> <dirbaio> I did that here, it wasn't that bad https://github.com/embassy-rs/atomic-pool/blob/main/src/lib.rs#L236

21:25 <re_irc> <jamwaffles> Hm that doesn't look too bad actually, thanks for the tip 🙂

21:26 <re_irc> <dirbaio> it doesn't work if the user passes _another_ const generic to the macro, of course...

21:28 <re_irc> <jamwaffles> I think I can get away with "PackedBuffer<const W: u32, const H: u32, const N: usize>" where "N = (W * H) / 8", no need for a fourth one (which was the cause of my original question). I suppose a macro would let me elide "N" but I'm doing an assertion to check the invariant anyway

21:29 <re_irc> <dirbaio> fun