Silence msvc warning about an unused named parameter #2289
Conversation
Warning C4100 may cause compile failures under strict warning regimes.
| @@ -2446,6 +2446,7 @@ constexpr int get_arg_index_by_name(basic_string_view<Char> name) { | |||
| } | |||
| if constexpr (sizeof...(Args) > 0) | |||
| return get_arg_index_by_name<N + 1, Args...>(name); | |||
| (void)name; | |||
There was a problem hiding this comment.
Looks like a bug in MSVC because name is used in the line above. I recommend reporting to Microsoft.
There was a problem hiding this comment.
I don't think so: if the control flow in an instantiation of
template <int N, typename T, typename... Args, typename Char>
constexpr int get_arg_index_by_name(basic_string_view<Char> name)
reaches line 2449, then both if constexpr statements before were evaluated as false and their first (in this case: only) substatements become so called discarded statements [stmt.if]/2 which leave no use of name in that template specialization.
There was a problem hiding this comment.
Looks like it was also discussed on Microsoft Developer Community, and resulted in joining the other compilers (GCC and Clang, at least) for these cases.
There was a problem hiding this comment.
Thanks for both links!
Regarding the supposed GCC bug: the original poster gave a wrong description of the nature of the problem. He was missing the difference between a runtime if and a compile-time if constexpr as layed out in the standard. And "This is a valid program which should not produce a warning" misses the point of -Wunused-but-set-parameter explicitly requesting a warning.
This is exactly what the last poster on MS DevCon is rightfully complaining about.
There was a problem hiding this comment.
I've referenced this PR on DevCon. According to my knowledge of the release schedule this will certainly not make it for 16.10 which will probably be around for a long time.
There was a problem hiding this comment.
Sure, whatever is preferred here. It's just the same as a few lines later.
There was a problem hiding this comment.
Thanks for investigating and I'm glad that this nonsense has already been reported. Merged this as a workaround for now.
There was a problem hiding this comment.
"This is a valid program which should not produce a warning" misses the point of -Wunused-but-set-parameter explicitly requesting a warning.
I disagree, using a parameter in one branch of if constexpr is not conceptually different from using it in one branch of if (that may be optimized away). It's great that it was fixed in GCC.
There was a problem hiding this comment.
The "misses" is referring to the blanket statement. A valid program may have many good reasons to issue warnings.
Warning C4100 may cause compile failures under strict warning regimes.