SQLite Forum

Sum precision ?
Login
You made the change to func.c to use the LONGDOUBLE_TYPE instead of double in func.c

If LONGDOUBLE_TYPE is not defined then it is set `long double` so the easiest way is just to make sure the compiler is passed `-DLONGDOUBLE_TYPE=__float128` to override that default on the command line that builds the executable.

I build on WIndows, so I use MSVC and makefile.msc to make the amalgamation, then have all the options I want to use in config.h (including the define for LONGDOUBLE_TYPE) and set `-D_HAVE_SQLITE_CONFIG_H` on the gcc command line (the GCC build doesn't use the makefile).