02:08 <firefrommoonligh> Absolutely. The answer is bump the min

09:00 <jannic[m]> One good reason to keep compatibility with an older rust version is if you want your crate to be included in linux distributions: They often have the reasonable requirement that all included software needs to be built with the compiler version provided by the same distribution.

09:00 <jannic[m]> This usually doesn't apply to crates that are only meant to used for embedded development, but if you create a library that can be useful both on embedded and on non-embedded targets, it may be worth considering.

09:12 <thejpster[m]> that's a good point. "This crate is for me, and I will set it according to my needs" is a valid take. But "I want as many people to use my software and so I will make it easy to use, even on old rust compilers" is a different, but equally valid, take.

09:12 <thejpster[m]> * many people as possible to use

09:24 <therealprof[m]> Does anyone know a driver that will use the I2C transaction method?

09:24 <therealprof[m]> I've seen a lot of todo!() in the HALs the last couple of days so I'm curious about a driver to test that.

09:27 <mkj[m]> https://github.com/ftdi-rs/ftdi-embedded-hal implements it, not sure if that's much use

09:28 <mkj[m]> oh, use it.

09:34 <thejpster[m]> https://github.com/search?q=embedded_hal%3A%3Ai2c%3A%3AOperation&type=code turns up https://github.com/Phosfor/bhi160/blob/51dde9d1bad0614cbfc79937c26ec341b2b76167/src/interface.rs#L44. Is that useful?

12:01 <therealprof[m]> <thejpster[m]> "https://github.com/search?q=..." <- Good call searching for the `i2c::Operation`. 👍️

16:53 <MichaFita[m]> <M9names[m]> "I'm doing this stuff in my spare..." <- This is valid point of view (kind of especially in my case), but then I had contributors who don't follow `rustc` updated that fast.

16:55 <MichaFita[m]> <jannic[m]> "One good reason to keep compatib..." <- No, I don't want PACs distributed with Linux... aside the fact I probably wouldn't want any of my crate distributed with Linux as their distribution is broken for Rust crates (they try to use it the same way as dynamic C/C++ libraries).

19:37 <whitequark[cis]> for those of you who know the pain of recovering a netlist of a board you're reverse engineering: you can do it much faster with boundary scan! https://mastodon.social/@whitequark/114564412970898277

20:05 <Mathias[m]> Catherine: I saw your messages on Mastodon, and I was wondering how one could map the boundary scan to specific pins. Looking at the gist, the answer is you need a BSDL file for the SoC.

20:06 <Mathias[m]> Pretty cool new application for Glasgow Interface Explorer.

20:44 <whitequark[cis]> yep, I had to write a BSDL parser, which was distinctly unpleasant

20:44 <whitequark[cis]> I've avoided it for like 5 years :D

23:50 <KevinPFleming[m]> This is a bit surprising... compiling with Rust 1.87 and 'release' profile with maximum optimization, for thumbv6m. The 'memcpy' builtin is 420 bytes of object code.

23:50 <KevinPFleming[m]> ```134232668 00000420 t compiler_builtins::mem::memcpy::hb1afff1216b6a636```

23:50 <KevinPFleming[m]> 'memmove' is even worse, it's 938 bytes.