ilya-biryukov added inline comments.
================
Comment at: lib/Tooling/InterpolatingCompilationDatabase.cpp:201
+
+ CommandTraits computeTraits() const {
+ CommandTraits Result;
----------------
jfb wrote:
> It's not clear to me that this entire function is safe to call from multiple
> threads at the same time. Even if it's safe now, I'm willing to bet it won't
> always be that way. `getTraits` should therefore use a magic static or
> `call_once` and avoid the headache entirely.
Thanks, `call_once` is both simpler and more reliable.
================
Comment at: lib/Tooling/InterpolatingCompilationDatabase.cpp:258
+ /// Null before Traits were computed, non-null otherwise. All accesses to
this
+ /// must be atomic.
+ mutable std::shared_ptr<CommandTraits> Traits;
----------------
jfb wrote:
> The comment should say why accesses need to be atomic. Or better yet, this
> should only be usable "the right way".
Clarified how it's used.
Repository:
rC Clang
https://reviews.llvm.org/D51314
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
