05:05 <mkj[m]> is there a way to mark fields as safe for FieldWriter using svd2rust? things like timeouts or lengths which are multi-bit but not intended as enums

05:20 <mkj[m]> <mkj[m]> "is there a way to mark fields as..." <- ah, `writeConstraint`s on the fields. https://github.com/rust-embedded/svd2rust/issues/844

05:44 <ericseppanen[m]> Is there a circular buffer / ring buffer / spsc channel implementation that can be used on rp2040 for sending data between cores?

05:50 <ericseppanen[m]> <EricSeppanen[m]> "Is there a circular buffer..." <- I could just put data in a heapless::Deque behind a lock, I guess? But I'm a bit spoiled by lockfree channels in userspace so that's what I reach for by instinct.

06:07 <mkj[m]> <EricSeppanen[m]> "Is there a circular buffer..." <- https://github.com/jamesmunns/bbqueue maybe?

06:08 <M9names[m]> heapless spsc is also an option if you don't need a deque

06:10 <M9names[m]> <MurrayToddWillia> "Okay, so for my own understandin..." <- > <@murraytodd:matrix.org> Okay, so for my own understanding, let me play things back. This has to do with the fact that the rp2040 has two cores, so rather than assuming `portable-atomic` should use it's own critical-section based on single-core... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/WvAdWTDDBcTuQMHQwzhIgTUI>)

06:11 <ericseppanen[m]> bbqueue uses cortex-m 0.6 and hasn't been released since 2021. Hard to tell if it would play nicely with the rest of the ecosystem?

06:12 <M9names[m]> bbqueue should be fine. but it's really for "I want to DMA to/from the buffer", if you don't need that then it's probably not a great fit

06:12 <M9names[m]> and it's not like it's unmaintained, if it hasn't changed it's because no-one has needed it to change

06:12 <ericseppanen[m]> heapless::spsc would probably be OK but it's a bit hard to call split() in a const context.

06:31 <ericseppanen[m]> * const context. Maybe StaticCell<Producer> would work though...

10:59 <henrik_alser[m]> Eric Seppanen: you can also use embassy-sync zerocopy_channel for example (with CriticalSectionRawMutex)

11:00 <henrik_alser[m]> https://github.com/embassy-rs/embassy/blob/main/examples/rp/src/bin/zerocopy.rs

11:03 <henrik_alser[m]> Or a normal channel if copy is ok

11:03 <henrik_alser[m]> Here’s an example of that https://github.com/embassy-rs/embassy/blob/main/examples/rp/src/bin/multicore.rs

11:03 <henrik_alser[m]> s/channel/Channel/

11:03 <henrik_alser[m]> s/channel/Channel/, s/copy/copying/

11:05 <henrik_alser[m]> (They can be used outside of embassy aswell if you don’t use it)

11:10 <JamesMunns[m]1> <M9names[m]> "and it's not like it's unmaintai..." <- It's a little unmaintained. Ithinel has started helping out but I haven't had time to give it love for a while.

11:17 <henrik_alser[m]> (Note that zerocopy_channel is spsc only, if you need mpsc you can use for example atomic-pool or heapless::Pool and send Box on a channel as above

11:17 <adamgreig[m]> Error in core on its way :o https://github.com/rust-lang/rust/issues/103765#issuecomment-2155244052

11:19 <diondokter[m]> adamgreig[m]: Yes I saw!

11:19 <diondokter[m]> So ehh, when do we embedded-hal 2.0 with this trait bound on all error associated types?

11:19 <diondokter[m]> :P

11:19 <diondokter[m]> * do we release embedded-hal 2.0

11:21 <diondokter[m]> Doesn't this also unlock getting the std::io traits to core?

13:22 <henrik_alser[m]> * (Note that zerocopy_channel is spsc only, if you need mpsc you can use for example atomic-pool or heapless::Pool and send Box on a normal Channel as in the above multicore example

13:22 <henrik_alser[m]> * (Note that zerocopy_channel is spsc only, if you need mpsc you can use for example atomic-pool or heapless::Pool and send Box on a normal Channel as in the above multicore example)

16:43 <dirbaio[m]> <diondokter[m]> "Doesn't this also unlock getting..." <- Nope 🥲

16:43 <diondokter[m]> what else is in the way then?

16:45 <dirbaio[m]> the std::io traits need the std::io::Error struct

16:45 <diondokter[m]> Ah, right

16:45 <dirbaio[m]> which contains a `Box<dyn Error>`

16:45 <dirbaio[m]> for the wrapped

16:45 <dirbaio[m]> * the wrapped error

16:46 <dirbaio[m]> which you can't do on core

16:46 <diondokter[m]> https://doc.rust-lang.org/std/io/struct.Error.html#method.new

16:46 <diondokter[m]> Yep, I see. Annoying!

17:03 <takkaryx[m]> started looking for a static allocated protobuf implementation for Rust, and came across this: https://github.com/dflemstr/femtopb just recently, anyone have any experience with it? looking to use USB with protobuf on an STM32F4 to report data periodically.

18:15 <henrik_alser[m]> takkaryx: https://github.com/embassy-rs/noproto

18:15 <takkaryx[m]> henrik_alser[m]: oh hey, another one! thanks!

18:16 <henrik_alser[m]> Yeah i haven’t tried either but just FYI :)

18:18 <takkaryx[m]> well, I'm working on my USB layer, guess I'll give both a shot and see which one works better XD

20:12 <barafael[m]> Is there an alternative to micromath when better precision is required? Finally time to learn to use fixed point math?

20:19 <JamesMunns[m]1> <barafael[m]> "Is there an alternative to..." <- libm?

21:03 <firefrommoonligh> Hmm. Well... Cmsis-dsp. Hardware cordic if available. Num_traits, which I think abstracts over mm and some other options; don't recall

21:04 <firefrommoonligh> Both fixed and floating point have their own traps. I think you are more likely to hit fixed point traps, but it depends

21:04 <firefrommoonligh> f64 can be useful too if you understand the performance implications of software floats, or if you have double prec Fpu

21:10 <barafael[m]> I got no fpu, rp2040 :)

21:16 <barafael[m]> Maybe the core intrinsics have some floating point ops

21:23 <henrik_alser[m]> <barafael[m]> "Maybe the core intrinsics have..." <- Yeah, they are enabled by default in embassy-rp at least

21:33 <AlexandervanSaas> <henrik_alser[m]> "Yeah, they are enabled by..." <- Are they? Not on the version on crates.io as far as I can see.

21:33 <AlexandervanSaas> Not in the git version either https://github.com/embassy-rs/embassy/blob/e5495b51b43f71cdc489e92f4deb0cbc0e27e6aa/embassy-rp/Cargo.toml#L24

21:39 <henrik_alser[m]> <AlexandervanSaas> "Are they? Not on the version..." <- Oh, i recall they were at some point back in time last time i needed them, but i might be wrong

21:41 <henrik_alser[m]> But yeah just enable the feature barafael

21:45 <henrik_alser[m]> And thanks for the heads-up Alexander van Saase 😅

21:49 <barafael[m]> So I still have to use libm for e.g. sqrt to become available right? I'm defining the math in a separate crate