I am attempting to run a set of simuilnk models using model references. I have both MinGW and Visual Studio installed, and switch between them regularly using mex -setup. I typically leave the Simulink model configset 'Toolchain' parameter at 'Auto', which has worked for many years.
However, my current model ALWAYS seems to want to use visual studio to compile the top-level model accelerator target, even when mex -setup is set to MinGW (both C and C++), and I've even manually set the 'Toolchain' parameter to 'MinGW64 | gmake (64-bit Windows)'!
Curiosly, the referenced models (which are sharing the same referenced config set) ARE built with the MinGW toolchain, but when the top-level accelerator target is attempted to update, it bombs out with the message:
The model 'FSIM_FsimTestModel' is configured to use the compiler 'Microsoft Visual C++ 2015 (C):TMF', however the referenced model 'FSIM_FsimTestModel_referenced' was built with 'MinGW64 Compiler (C):Toochain' which is not compatible for a model reference hierarchy build. Consider configuring all models to use the same compiler. The Build Process will terminate.
The message politely suggests to "Consider configuring all models to use the same compiler", but I've exhausted all of my knowledge of how to accomplish this? Any thoughts?
Why is the top level model claiming to be congifured for Visual Studio when it's 'Toolchain' parameter is explicitly set to 'MinGW64 | gmake (64-bit Windows)'?