[PATCH] D18821: Add bugprone-bool-to-integer-conversion

[Alexander Kornienko via cfe-commits]

alexfh added a comment.

Sorry for the delay. Your patch was lost in my inbox. Feel free to ping me earlier.


================
Comment at: docs/clang-tidy/checks/bugprone-bool-to-integer-conversion.rst:37
@@ +36,3 @@
+
+It turns out that the common bug is to have function returning only bools but having int as return type.
+If check finds case like this then it function return type to bool.
----------------
Prazek wrote:
> alexfh wrote:
> > It may well not be a bug. It's definitely not a bug for `extern "C"` functions and functions in C code. We definitely don't need to change the function return type, since it's a rather involved change (and clang-tidy checks currently are simply not able to make such change correctly in case of a function with external linkage). So please remove this automated fix.
> This check runs only on C++ functions. Maybe checking that the function isExternC() would be enough?
As I've said, extern "C" is not the only way to make a function a part of an API exposed to external users.

================
Comment at: test/clang-tidy/bugprone-bool-to-integer-conversion.cpp:192
@@ +191,3 @@
+  // CHECK-FIXES: bool yat2() {
+
+  auto l = []() {
----------------
Prazek wrote:
> Why it is dangerous? What I see after runnign on llvm code base, that it is one of the most frequent case.
> The only bug is the extern "C" thing.
Well, `extern "C"` is not the only way to expose functions to some external code. This fix potentially changes ABI, which is generally not a safe thing to do. I think, we should make the fix optional.


http://reviews.llvm.org/D18821