I'd just make the cast explicit:
inline constexpr vsize VPOS (vsize(-1));
It's cleaner and requires less updates as the code evolves.
However, I can't reproduce it, what compiler flags are you using?
I'm using -Wall and clang is quiet about this use case
*/tmp %* g++ -Wall -std=c++17 xxx.cc
*/tmp %* cat xxx.cc
#include <iostream>
inline constexpr size_t x(-1);
int main()
{
std::cout << x << std::endl;
}
*/tmp %* g++ --version
Apple clang version 15.0.0 (clang-1500.1.0.2.5)
Target: x86_64-apple-darwin23.3.0
Thread model: posix
InstalledDir: /Library/Developer/CommandLineTools/usr/bin
*/tmp %* ./a.out
18446744073709551615
*/tmp %*
L
On Tue, 27 Feb 2024, 12:57 Aaron Hill, <lilyp...@hillvisions.com> wrote:
> On 2024-02-27 3:32 am, Werner LEMBERG wrote:
> > Looking at some MacPorts log files from Mac OS 14.2.1 (which uses the
> > system's `clang` compiler) I see
> >
> > ```
> > flower/include/flower-proto.hh:26:30:
> > warning: implicit conversion changes signedness:
> > 'int' to 'const vsize' (aka 'const unsigned long')
> > [-Wsign-conversion]
> > inline constexpr vsize VPOS (-1);
> > ~~~~ ^~
> > ```
> >
> > for almost all C++ fies, making the log file quite noisy. Dan, any
> > idea how to fix this gracefully?
>
> Since vsize is just a typedef of size_t, would it make sense to use
> SIZE_MAX when defined VPOS instead of a manual cast of negative one?
>
>
> -- Aaron Hill
>
>
