Once a translation has been made, running the limbo (1) compiler on the resulting output should pick out the areas where hand editing is required.
C2l normally puts all mapped C code (plus that from included files) into a single .b file.
The options to c2l are:
-p Use an ANSI preprocessor in place of the internal one.
-D name=def
.ns
-D name Define the name to the preprocessor, as if by .LR #define . If no definition is given, the name is defined as .LR 1 .
-I dir An .L #include file whose name does not begin with slash or is enclosed in double quotes is always sought first in the directory of the file argument. If this fails, or the name is enclosed in <> , it is then sought in directories named in -I options, then in /sys/include , and finally in /$objtype/include .
-m Put the mapped code of any included .h files into its corresponding .m file instead of the .b file.
-i Send the mapped code of any included .h files to /dev/null .
-l Send the mapped code of any non-local included .h files to /dev/null .
-c Just generate code corresponding to the C code ie don't include any prologue or epilogue code such as an implement header, include declarations, module declarations or an init function.
-v Outputs any warnings to standard error as well as putting them in the output source.
-s Map C strings to NUL-terminated arrays of bytes in Limbo. This just about preserves the semantics of strings and makes the process of hand editing much easier. It is useful as a first attempt at translation. In this case the module /module/libc0.m is used in place of the standard one /module/libc.m.
-S Map "char*" in C to string in Limbo. Incompatible with the -s option.
-M Indicates this file is the one containing the C main program. Used with the -q option below when c2l does not always know this until it's too late.
-q This reduces the number of passes that c2l makes over the C code. It makes it faster but more liable to miss some transformations. Cyclic data structures might not be detected.
-a For functions which are passed the address of a scalar typed (ie not a structure or union) expression as a parameter, pass the expression itself and rewrite the function and all calls of it to return the expression. For example :-
.EX int f(int x, int *y) { *y = x*x*x; return x*(*y); } void g() { int p3, p4; p4 = f(1729, &p3); }
becomes
.EX f(x: int, y: int): (int, int) { y = x*x*x; return (x*y, y); } g() { p3, p4: int; (p4, p3) = f(1729, p3); }
C2l runs the preprocessor on the C code before starting translation. As a special case it will convert definitions of constants into Limbo constant declarations. It makes no attempt to convert any definitions into function declarations.
Identifier names that clash with Limbo keywords have letter x appended so, for example, a structure member called type would become typex .
Warning messages preceded by the acronym TBA (to be addressed) are issued for NUL bytes in strings, ... as an argument, array indices in declarations, use of void type, use of unions, bit fields, use of address operator, negative array indices, element specifiers, initial values in Limbo modules, labels, gotos and case statement fall through.
The C types char and "unsigned char" are mapped to the Limbo byte type. The C types short, unsigned short, int, unsigned int, long and unsigned long are mapped to the Limbo int type. The C types long long and unsigned long long are mapped to the Limbo big type. Finally the C types float and double are mapped to the Limbo real type.
Anonymous C structures and unions map to a name of the form <module>_adt_<num> where module is the name of the module which is, in turn, derived from the file name. Anonymous member names in strucures and unions have a name of the form anon_<num>. Finally,temporary variables generated by c2l have a name of the form tmp_<num>. In all cases <num> is a unique identifier.
C2l may well crash if given invalid C code.
c2l -a does not always do all possible conversions.