stuart@bms-at.UUCP (Stuart Gathman) (12/15/88)
To take advantage of prototypes but still remain compatible with older compilers, we do the following: For ANSI compilers: int func(/**/ type, type /**/); For K&R compilers: int func(/* type, type */); A convert program (e.g. sed) can easily turn '(/**/' into '(/*' and vice versa. This works well, but there are still some things on my wish list. 1) many compilers take: int func(/**/ type name1, type name2 /**/); but we can't use this feature (to make header files more self documenting) because of the ones that don't. 2) Ansi requires '...' in a prototype to indicate varargs functions. Many compilers won't take '...' but use trailing comma instead. 3) The '/**/' is annoying when working with an ANSI compiler. Does anyone have a parser that can recognize function prototypes and 1) replace them with comments or 2) strip the optional identifiers or 3) add or remove the trailing '...' By making some reasonable assumptions about coding style, it should work without having to process typedefs or run the preprocessor. I have started such an animal, but don't have a lot of time to spend on it. Has it been done already? -- Stuart D. Gathman <stuart@bms-at.uucp> <..!{vrdxhq|daitc}!bms-at!stuart>
davidsen@steinmetz.ge.com (William E. Davidsen Jr) (12/17/88)
In article <136@bms-at.UUCP> stuart@bms-at.UUCP (Stuart Gathman) writes: | Does anyone have a parser that can recognize function prototypes and | | 1) replace them with comments | or | 2) strip the optional identifiers | or | 3) add or remove the trailing '...' I will share with you a little technique I use for allowing procedure declarations to hold what I want without having to rewrite the source code. Be warned, this addresses (1) and (2), and most of (3), but it's really ugly. Also, some compilers which take prototypes with type info don't accept elipsis, while others do. The comment shows where I change this if need be. ________________________________________________________________ #define NoProto 1 #define TypesOnly 2 #define FullANSI 3 #ifndef P #define P NoProto #endif #if P == NoProto #define Param(x) () #endif #if P == TypesOnly #define Param(x) x #define Ptype(x, y) x #define Pelipsis /* is this one right for you? */ #endif #if P == FullANSI #define Param(x) x #define Ptype(x,y) x y #define Pelipsis ,... #endif int someproc Param((Ptype(int *,x),Ptype(long,ypos))); int myprintf Param((Ptype(char *, format) Pelipsis)); ________________________________________________________________ This will doubtless start a macro war, in which ten people will tell me there's a better way, and another ten will ignore the comment about variable arguments and tell me Pelipsis has the wrong value for TypesOnly. Oh well, I hope this actually helps someone. -- bill davidsen (wedu@ge-crd.arpa) {uunet | philabs}!steinmetz!crdos1!davidsen "Stupidity, like virtue, is its own reward" -me
pjh@mccc.UUCP (Pete Holsberg) (12/17/88)
On a somewhat-related topic: Some authors call the first line of a function definition a declaration. Some don't call it anything. Is there an official name? Thanks, Pete -- Pete Holsberg UUCP: {...!rutgers!}princeton!mccc!pjh Mercer College CompuServe: 70240,334 1200 Old Trenton Road GEnie: PJHOLSBERG Trenton, NJ 08690 Voice: 1-609-586-4800
gwyn@smoke.BRL.MIL (Doug Gwyn ) (12/17/88)
In article <494@mccc.UUCP> pjh@mccc.UUCP (Pete Holsberg) writes: > Some authors call the first line of a function definition a >declaration. Some don't call it anything. Is there an official name? "Function header" is what I usually hear it called.