danmar
extracted hard-coded platform definitions to actual platform files
This breaks cppcheck-gui since it relies on the built-in platforms. Seems I need to change everything first and create smaller parts from that.
Since I just wanted to get my backlog'd changes into the tree and those are incomplete this PR will unfortunately sit a while.
The tests are affected by no platform data being set on Windows. The platform file loading is part of the command-line parsing and is not executed before the tests. When the windows platform non-builtin causes no data to be set. Before we can merge this we need to remove those first. Fortunately there are already scheduled to be removed in the next dev cycle.
For some reason the validation is not able to read <default-sign>�</default-sign> because it says that 0 is an invalid xmlChar. The application code handles it fine.
But I think we should not have this. The built-in configurations currently default to the signedness of the binary which I think is wrong. Same as we were defaulting to a Windows platform in Windows binaries. Those should possibly be signed by default. We already have unsigned for unix so you could already fixate that.
We also should probably have a command-line option similar to -fsigned-char/-funsigned-char instead. This is rather specified by the compiler than the actual platform.
This is something I have to think about and I am gonna finish #5414 first as this also plays into it.
Having \0 in the character sign field makes no sense at all. We are provide a fixed platform and then we are basing that value on the compiler - that just seems wrong. I will drop that and go for signed by default as the common compilers do and add an additional option to control that in a different PR first.