All Hands 2020: Draft Agenda Items

[jamesmunns]

Placeholder

[japaric]

For reference, this year (RAH 2019) we had 2 1-hour meetings and one 2-hour working session. Notes and attendance can be found at https://paper.dropbox.com/doc/WG-Embedded-Rust-All-Hands-2019-Topics--Ams~zlHKeJ911vGs6mM1Vn~YAg-JrPzUM1Z3SC859MS03lfC

[Dylan-DPC-zz]

Following this issue for coordination purposes.

[jonas-schievink]

My two cents:

T-lang inquiries:

• Specify what Send and Sync mean without using the word "thread"
  • Embedded Rust uses "thread-equivalent" constructs such as interrupt handlers, and also has less equivalent constructs such as multi-core MCUs. It would be good to know what Send and Sync really mean, since their contract can't be specific to just threads.
• Reconsider RFC 2492 ("Existential types with external definition") as it would allow using the same mechanism for allocators, panic handlers, and other embedded-specific things (which would be?)
• Path to stabilization of #[link_args]? If it worked on libraries and propagated to binaries it would be useful for eg. https://github.com/rust-embedded/cortex-m-rt/issues/237.

T-compiler inquiries:

• How do we get the bug fixed where LLVM thrashes the link register? Infinite backtrace, odd backtrace behavior and "corrupted stack?" in GDB cortex-m-rt#139
  • I think there's still no upstream bug for this, so this is basically our fault

T-infra inquiries:

• How much room is there on dist-various-1 for embedded targets? Currently that builder runs all embedded tests even though it's not really meant for that I think. (failures also don't show up in PR builds, only in bors)

[jamesmunns]

All hands didn't happen this year. Closing.