lefticus / constexpr_all_the_things Goto Github PK

The following testcase will fail:

constexpr auto number_val = JSON::number_parser()("0.0625"sv);
static_assert(number_val && number_val->first == 0.0625);

Parsed value is 0.625

GCC 7 and earlier have a bug (Bug 70167) that it treats T{} in the same way as C99 compound literals. This was recently fixed, with the result that the trunk gcc now correctly gives errors for the construct constexpr std::initializer_list<T> il{Ts...}; in cx_json_parser.h as il is not initialized by a constant expression (the standard specifies this initialization as involving pointers to a temporary object on the stack...).

Hi, I watched the talk on CppCon2017 and tried to compile the piece of code with gcc-trunk but the compiler would say something like:

variable ‘b’ set but not used [-Werror=unused-but-set-variable]

My compiler version is

Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/home/phi/gcc-trunk/libexec/gcc/x86_64-pc-linux-gnu/8.0.0/lto-wrapper
Target: x86_64-pc-linux-gnu
Configured with: ./../gcc-trunk/configure --prefix=/home/phi/gcc-trunk --enable-languages=c,c++ : (reconfigured) /media/phi/a60c856b-88ad-4199-ab95-a49fc0c9466a/codebase/gcc-infrastructure/objdir/../gcc-trunk/configure --prefix=/home/phi/gcc-trunk --enable-languages=c,c++
Thread model: posix
gcc version 8.0.0 20171123 (experimental) (GCC)

I use the command
cd build && cmake -D CMAKE_CXX_COMPILER=/home/phi/gcc-trunk/bin/g++ CMAKE_C_COMPILER=/home/phi/gcc-trunk/bin/gcc ../ to generate the Makefile.

The error message is like:

<dir>/src/test/algorithm.cpp: In function ‘void algo_tests_nonmod()’:
<dir>/src/test/algorithm.cpp:11:20: error: variable ‘b’ set but not used [-Werror=unused-but-set-variable]
     constexpr bool b = cx::all_of(std::cbegin(arr), std::cend(arr),
                    ^
<dir>/src/test/algorithm.cpp:18:20: error: variable ‘b’ set but not used [-Werror=unused-but-set-variable]
     constexpr bool b = cx::any_of(std::cbegin(arr), std::cend(arr),
                    ^
... (other variable unused error messages) ...

But when I checked the code these variables is used by static_assert like static_assert(b, "all_of fail");.

It seems so weird to me since these variables are used in the next line. Could anyone give me some insight or helps for how to go through this issue? (An obvious solution is turning off the -Werror option.)

Thanks.

Hi Jason & Ben!

First I want to start of by saing "Great talk!" (as usual).
I'm prepping a talk for myself on a compile-time lisp interpreter, is it ok to use this project as a basis for my talk?