In the same vein as my last change. I'm not convinced about using inline in the header to force the function to live in a COMDAT, so, suggestions welcome.
Details
Diff Detail
- Repository
- rL LLVM
Event Timeline
LGTM. Thanks!
include/llvm/Analysis/CFLAliasAnalysisUtils.h | ||
---|---|---|
23 ↗ | (On Diff #104061) | Now that I look at it, this should probably live in the llvm::cflaa namespace, rather than the global namespace (unless you're planning on making this more general?) As an added bonus, that may let us get rid of the using namespace llvm above? I'm indifferent about whether that's included in this patch, or if it's done in a follow-up. |
46 ↗ | (On Diff #104061) |
Honestly, given how small this function will end up being and that it should only ever be used in 2 TUs, I'd be content if we just made it static inline. I'm happy with either approach ¯\_(ツ)_/¯. |
56 ↗ | (On Diff #104061) | If we expand these namespaces to include FunctionHandle, please add // namespace cflaa and // namespace llvm to their respective close braces. |