On Sun, Jul 9, 2023 at 6:20 PM Peter Eisentraut <pe...@eisentraut.org> wrote: > So I don't think this code is correct. AFAICT, there is nothing right > now that can possibly define HAVE_MBSTOWCS_L on Windows/MSVC. Was that > the intention?
Yes, that was my intention. Windows actually doesn't have them. The autoconf/MinGW test result was telling the truth. Somehow I had to make the three build systems agree on this. Either by strong-arming all three of them to emit a hard-coded claim that it does, or by removing the test that produces a different answer in different build systems. I will happily do it the other way if you insist, which would involve restoring the meson.build and Solultion.pm kludges you quoted, but I'd also have to add a compatible kludge to configure.ac. It doesn't seem like an improvement to me but I don't feel strongly about it. In the end, Solution.pm and configure.ac will be vaporised by lasers, so we'll be left with 0 or 1 special cases. I don't care much, but I like 0, it's nice and round.
