Fixed VectActive::from() returning incorrect value

[skibon02]

Fixes #499

Fixed VectActive::from(u8) now returns correct value for Interrupt variant Refactored SCB::vect_active() to use VectActive::from(u8) and get rid of code duplication

[newAM]

Can you rebase this? The CI failure is from an old MSRV

[skibon02]

@newAM, Could you please explain a little more? I forked the repo and made this commit from the v0.7.x branch.

[newAM]

Oh, sorry, I didn't see that this was targeting v0.7.x! That makes more sense then. The problem is that serde_json isn't pinned, and the latest version requires edition 2021. I think specifying >=1.00,<=1.0.100 for serde_json should fix the build failure.

[skibon02]

Oh, unwrap_unchecked seems to not be available on rustc version 1.38.0. It forces me to write a reaction to the case, when icsr register returns unexpected value. Is panic appropriate in this case? Or maybe we should execute udf instruction?

Assuming correct implementation for VectActive::from for all cortex-m processors, this problem can only appear when incorrect target was selected by user, and it is impossible to check this in compile time.

[skibon02]

Adding these ugly lines in Cargo.toml to fix deps versions did help compiling on 1.38

Or should we better just bump MSRV to 1.60?

[jonathanpallant]

cortex-m-rt says:

This crate is guaranteed to compile on stable Rust 1.59.0 and up

[skibon02]

But this is cortex-m, not cortex-m-rt latest update 0.7.7 was published 1 year ago

From v0.7.x branch Readme: https://github.com/rust-embedded/cortex-m/blob/bb4a78208323260a161e68b2498438867f971bc5/README.md?plain=1#L14

[jonathanpallant]

But how many people use cortex-m without using cortex-m-rt?

[skibon02]

cortex-m has twice as many downloads compared to cortex-m-rt It seems that there isn't much sense in supporting these old versions of the Rust compiler. Moreover, a significant portion of people even uses Nightly.

Should I bump the MSRV to 1.60 and remove fixed dependency versions?

[jonathanpallant]

1.59 makes sense, so they are the same.

unwrap_unchecked first appeared in 1.58 so that's fine.