C to O-O Translation: Beyond the Easy Stuff

Abstract

Can we reuse some of the huge code-base developed in C to take advantage of modern programming language features such as type safety, object-orientation, and contracts? This paper presents a source-to-source translation of C code into Eiffel, a modern object-oriented programming language, and the supporting tool C2Eif. The translation is completely automatic and supports the entire C language (ANSI, as well as many GNU C Compiler extensions, through CIL) as used in practice, including its usage of native system libraries and inlined assembly code. Our experiments show that C2Eif can handle C applications and libraries of significant size (such as vim and libgsl), as well as challenging benchmarks such as the GCC torture tests. The produced Eiffel code is functionally equivalent to the original C code, and takes advantage of some of Eiffel's object-oriented features to produce safe and easy-to-debug translations

Really Automatic Scalable Object-Oriented Reengineering

“…There, local jump instructions reach the specific point saved with setjmp within the method's body. We do not discuss the details of the translation because they refer to several low-level mechanisms discussed in [27] that are out of scope in the present paper. From the point of view of the reengineering, however, the translation expresses the complex semantics of longjmp naturally through the familiar exception handling mechanism.…”

“…This shows that the object-oriented reengineering produced by AutoOO improves the design without overhead with respect to a bare non-reengineered translation. The basic performance overhead of switching from C to Eiffel-analyzed in detail in [27]-significantly varies with the program type but, even when it is pronounced, it does not preclude the usability of the translated application or library in standard conditions. These conclusions carry over to programs reengineered with AutoOO, and every optimization introduced in the basic translation provided by C2Eif will automatically result, at the end of the tool chain, in faster reengineered applications.…”

“…As discussed in detail in [27], the raw translation from C to Eiffel provided by C2Eif does not currently support: a few rare programming patterns that rely on specific memory layouts, such as how the arguments passed to a function are stored next to one another; and a few GCC exotic extensions. Using CIL as preprocessor, while it contributes to simplifying and maintaining the translation, also carries its own limitations: K&R legacy C is not supported; and comments are stripped and formatting is lost, and hence this information cannot be used to improve the readability and formatting of the translated Eiffel code.…”

“…C2Eif and AutoOO. The reengineering techniques described in this paper are combined with our previous work on C2Eif [27] and implemented as part of the toolchain shown in Figure 1. The toolchain implements an overall transformation that inputs a C program and outputs an object-oriented Eiffel project with the same functionality.…”

“…In the second stage, C2Eif transliterates the CIL output into procedural Eiffel, whose structure replicates that of the C input program without introducing elements of object-oriented design. We described this stage in previous work [27]; its details are largely independent of the reengineering techniques implemented by AutoOO, which only uses C2Eif as a back-end. Finally, AutoOO processes the C2Eif output, introduces the transformations described in Sections 3 and 4, and outputs the reengineered object-oriented Eiffel programs that can be compiled.…”