Updating Carbon's safety strategy (#5914)

Carbon is accelerating and adjusting its safety strategy, specifically
to flesh out its memory safety strategy and reflect simplifying
developments in the safety space.

This proposal replaces the previous directional safety strategy with a
new concrete and updated framework for the safety design. It includes a
specific framework for memory safety, simplified build modes, specific
"safety modes", and terminology.

This proposal also provides a _directional_ suggestion for temporal and
data-race safety specifically.

In addition to fully building out the above directional component, there
are several other aspects of our safety design that will follow in
subsequent proposals. The hope is to establish the initial framework
here.

---------

Co-authored-by: Dana Jansens <danakj@orodu.net>
Co-authored-by: josh11b <15258583+josh11b@users.noreply.github.com>
Co-authored-by: Mike Forster <michael@forster.pro>
Co-authored-by: Richard Smith <richard@metafoo.co.uk>
This commit is contained in:
Chandler Carruth
2025-08-20 23:31:30 +00:00
committed by GitHub
co-authored by Dana Jansens josh11b Mike Forster Richard Smith
parent 30b8a93fde
commit 223d0397c0
7 changed files with 810 additions and 31 deletions
-1
View File
@@ -25,6 +25,5 @@ wants to pursue, as well as those we want to exclude.
- [Information accumulation](information_accumulation.md)
- [Low context-sensitivity](low_context_sensitivity.md)
- [Prefer providing only one way to do a given thing](one_way.md)
- [Safety strategy](safety_strategy.md)
- [One static open extension mechanism](static_open_extension.md)
- [Success criteria](success_criteria.md)