Age | Commit message (Collapse) | Author | Files | Lines |
|
available trait
|
|
known bounds
Potential types from trait bounds are enumerated and, if there's no
possible coercions/unsizing, are searched for a single type that meets
all known conditions (i.e. trait bounds on the ivar, and method
existence)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Generate MIR when required for constant evaluation
|
|
|
|
|
|
|
|
but it's less ugly)
- Also includes some MIR optimisation changes to reduce some compile times (hopefully)
- Removed duplicated MIR consteval and now-unused HIR consteval
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
test harness
|
|
|
|
|
|
debugging)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Also adjust macOS comment. FreeBSD also uses clang and lld (current dev
version even uses lld as /usr/bin/ld), function-sections/gc-sections are
very well supported. It's probably a Mach-O issue on Macs.
|