| 12 May 2021 |
 dirbaio | oh true, this is going to be a pain | 18:24:54 |
| dirbaio | 🤣 | 18:25:05 |
 Bob McWhirter | for drogue-device, we examples/$chip/... | 18:25:07 |
| Bob McWhirter | and SPI v2 isn't what's in the F4 | 18:25:25 |
| dirbaio | yeah | 18:25:30 |
| dirbaio | no examples for now then | 18:25:50 |
| dirbaio | =D | 18:25:52 |
| Bob McWhirter | also, went with your change-word-size-on-the-fly idea | 18:26:06 |
| Bob McWhirter | biggest changes are spi.rs and adding some fns to AnyPin | 18:26:27 |
| Bob McWhirter | plus random .py changes | 18:26:31 |
| dirbaio | neat | 18:26:50 |
| Bob McWhirter | https://github.com/embassy-rs/embassy/blob/07db3ed7c10915f16eeda08909f04afd92902b1b/embassy-stm32/src/spi.rs | 18:26:52 |
| dirbaio | i'm finishing some crap, will review later | 18:26:57 |
| Bob McWhirter | no worries | 18:27:08 |
| Bob McWhirter | some questions I have... (a) should I be using some critical sections? I own everything so I think not? and (b) when do you decide to force inline'ing? | 18:27:33 |
| dirbaio | force-inlining ideally never, can make code size worse | 18:28:08 |
| Bob McWhirter | 'k, we trust rustc for inline decisions | 18:28:27 |
| dirbaio | ah you mean #[inline] | 18:28:31 |
| Bob McWhirter | yah | 18:28:45 |
| dirbaio | I thought by "force" you meant #[inline(always)] | 18:28:59 |
| Bob McWhirter | I probably did :) | 18:29:10 |
| dirbaio | #[inline] doesn't force inlining, it allows inlining across crates but the compiler may still decide it's not worth it | 18:29:28 |
| Bob McWhirter | ah | 18:29:49 |
| dirbaio | within a crate everything can be inlined | 18:29:59 |
| dirbaio | across crates only #[inline] stuff can be inlined (or generic stuff always, I think?), or everything if building with LTO | 18:30:28 |
| dirbaio | so I guess the rule of thumb is put #[inline] in hot, small fns that will be called often from outside the current crate | 18:31:55 |
| dirbaio | people who care about performance+codesize will build with LTO anyway | 18:33:03 |
| Bob McWhirter | so don't worry about private/crate-level fns? they'll optimize regardless? | 18:34:41 |
| Bob McWhirter | e.g., my set_word_size(...) which is private? | 18:34:52 |
| dirbaio | yeah I think so | 18:36:13 |
