13:32 <thejpster[m]> Now everyone knows I’ll be in Montreal for RustConf and in Vienna for EuroRust.

13:32 <thejpster[m]> https://fosstodon.org/@eurorust/112806436599573492

13:33 <thejpster[m]> jamesmunns accused me on his podcast of having the embedded rust project that’s had the most talks. One solution to that is for more people to submit more embedded rust talks.

14:27 <ilya-epifanov[m]> does anyone know what's the progress of #[ramfunc]? https://github.com/rust-embedded/cortex-m-rt/pull/100 What are the workarounds being currently used?

14:46 <Gnome[m]> Would the libs team accept a PR that made `heapless::Vec<T, N>` generic over the length type used? For the extreme example of a `Vec<u8, 2>` on a 64 bit system it would balloon to `(2 * 1) + size_of<u64>` which is `2` bytes of data and `8` bytes of length type.

14:46 <Gnome[m]> This could be smoothed by implementing a default length value for common values of `N`, with some macro generated trait implementations and would only really affect users using weirdly high `N` values or generic libraries which would only need to add a concrete or generic value of `LenT`.

15:33 <t-moe[m]> is someone also using the nordic ppk2?

15:33 <t-moe[m]>

15:33 <t-moe[m]> I'm trying to set it up to catch power consumption regressions in CI/CD automatically. But I'm struggling to get the same current integral as in the GUI when using it in the terminal (via python library or also rust crate)

15:39 <diondokter[m]> <Gnome[m]> "Would the libs team accept a..." <- > <@gnomeddev:matrix.org> Would the libs team accept a PR that made `heapless::Vec<T, N>` generic over the length type used? For the extreme example of a `Vec<u8, 2>` on a 64 bit system it would balloon to `(2 * 1) + size_of<u64>` which is `2` bytes of data and `8`... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/buVcDZSCDIyCcFngfeCpTHoH>)

16:02 <Gnome[m]> <diondokter[m]> "> <@gnomeddev:matrix.org..." <- Okay, I'll look into implementing it

16:03 <diondokter[m]> Gnome[m]: I'm not a libs maintainer though! So I don't know if something like this would be merged

16:03 <Gnome[m]> That's okay, if there's some precedent I can imagine it's wanted.

16:03 <JamesMunns[m]> it might be worth looking back in history, IIRC, heapless USED to be generic over the len type?

16:04 <JamesMunns[m]> and IIRC it was removed?

16:04 <JamesMunns[m]> I could be remembering wrong tho

16:04 <diondokter[m]> Gnome[m]: Just make sure to measure the change too. Padding could make the smaller fields not actually take less space

16:05 <JamesMunns[m]> yeah, v0.6 -> v0.7 went from typenum length generics to just `usize`/`u8`

16:05 <JamesMunns[m]> I can't remember why tho

16:07 <diondokter[m]> Maybe they thought the complexity (both for the implementation and having to carry around the generics) vs saved ram tradeoff wasn't worth it?

16:07 <JamesMunns[m]> yeah, I can certainly imagine a lot of reasons why

16:08 <diondokter[m]> Gnome!: Probably good to figure that out first before spending lots of time on this

16:13 <Gnome[m]> I can see v0.6 was based on generic_array, but I'm not talking about using typenum-style generics. I'm talking about changing ```rs... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/tCcCpUlDknvJkMKoBVgjQdvP>)

16:13 <Gnome[m]> * I can see v0.6 was based on generic\_array, but I'm not talking about using typenum-style generics. I'm talking about changing ```rs... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/MQeKXaDSVtMRMcrZkDGVOcxk>)

16:14 <Gnome[m]> * I can see v0.6 was based on generic\_array, but I'm not talking about using typenum-style generics. I'm talking about changing ```... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/sFMhKqErKpJwnPGmMxAggBOD>)

16:14 <Gnome[m]> * I can see v0.6 was based on generic\_array, but I'm not talking about using typenum-style generics. I'm talking about changing... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/TFuVSQPgyhqoRpeEnAuZDDTX>)

16:15 <Gnome[m]> * I can see v0.6 was based on generic\_array, but I'm not talking about using typenum-style generics. I'm talking about changing... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/cREXohwgBdDjAKntsCcLhIwu>)

16:39 <JamesMunns[m]> yep, I understand

16:39 <JamesMunns[m]> im just saying:... (full message at <https://catircservices.org/_matrix/media/v3/download/catircservices.org/tFkQMYUjCmCnADOJkrRTzxJK>)

16:41 <JamesMunns[m]> I'd probably bet "it made heapless vec painful to use" was the case, but that's just a guess :)

16:45 <Gnome[m]> The pain would be significantly mitigated via the DefaultLenType, I can imagine.

16:51 <Gnome[m]> Oh dear, this is going to be absolutely hell to implement with the view stuff....

16:54 <thejpster[m]> Does anyone know anything about FPUs and targets in Rust?

16:54 <thejpster[m]> I'm getting weird errors from the assembler again: https://rust-lang.zulipchat.com/#narrow/stream/122651-general/topic/Error.20writing.20.20inline.20asmfor.20the.20FPU.20on.20Armv8-M/near/452392031

20:21 <Gnome[m]> Turned out it wasn't too annoying, can someone take a look and review please :3

20:21 <Gnome[m]> https://github.com/rust-embedded/heapless/pull/504

20:24 <danielb[m]> if only we could calculate the type in comptime from N

20:24 <diondokter[m]> danielb[m]: That'd be nice haha. OTOH, zig doesn't have the generics Rust has

20:24 <Gnome[m]> I could have a nightly feature to do that, tbh.

20:25 <danielb[m]> maybe not for mainstream heapless :D

20:26 <danielb[m]> this PR and a nightly-only auto variant sounds complex in the same codebase

20:26 <dirbaio[m]> that wouldn't work with VecView hto

20:26 <dirbaio[m]> s/hto/tho/

20:27 <Gnome[m]> See the PR, I added the length generic to VecView but noted that I could also only allow `VecView` usage with `Vec<T, N, usize>`.