rochus-keller / c2obx

C header file to Oberon+ definition module transpiler

This is a tool to transpile a C header file (with includes) to a corresponding Oberon+ definition module.

It already works well with the SDL2 headers (with only minimal modifications), but consider the tool work-in-progress.

Note that since it is undecidable, whether a C pointer is indeed a pointer to the given object or in fact a pointer to an array of such objects, and Oberon+ only supports pointers to structured types, all *basic_type are converted to CPOINTER TO CARRAY of basic_type (i.e. *[]), but *structured_type are translated to Oberon+ CPOINTER TO structured_type (i.e. *). The special case of **type which is used to get a pointer out of a procedure is handled by *[]*type for structured types and void, and *[] for basic types.

Pending:

• process comments (line and block)
• consider #DEFINE constants
• use original type names for basic types (instead of plain C basic types)

c2obx's Issues

What is the purpose of -1 within an static quint32 array context for the function bool is_ident1(quint32 c).

Hi, I try to compile your code but get errors for the functionis_ident1(...) and is_ident2(...) .
I can see that -1 is some kind of end of marker in the range array used by in_rage(...) but -1 is not an quint32.
I cannot oversee side effect but is 0x00000 an option?

Tokenizer.cpp: In function ‘bool is_ident1(quint32)’:
Tokenizer.cpp:196:5: error: narrowing conversion of ‘-1’ from ‘int’ to ‘quint32 {aka unsigned int}’ inside { } [-Wnarrowing]
     };

and

Tokenizer.cpp: In function ‘bool is_ident2(quint32)’:
Tokenizer.cpp:207:5: error: narrowing conversion of ‘-1’ from ‘int’ to ‘quint32 {aka unsigned int}’ inside { } [-Wnarrowing]
     };