Path: utzoo!telly!ddsw1!mcdchg!rutgers!njin!princeton!notecnirp!drh
From: drh@notecnirp.Princeton.EDU (Dave Hanson)
Newsgroups: gnu.gcc.bug
Subject: Re: Bug with functions as parameters?
Message-ID: <12426@princeton.Princeton.EDU>
Date: 5 Oct 88 22:22:10 GMT
References: <8810051813.AA01101@uther.cs.purdue.edu>
Sender: news@princeton.Princeton.EDU
Reply-To: drh@notecnirp.UUCP (Dave Hanson)
Distribution: gnu
Organization: Dept. of Computer Science, Princeton University
Lines: 22

In article <8810051813.AA01101@uther.cs.purdue.edu> spaf@PURDUE.EDU (Gene Spafford) writes:
>Index: 
>    gcc 1.28 on a Sun 3/50, 3/60, 3/280 all running 3.4 SunOS
>
>Source to reproduce problem:

	int bar () { printf("hi!\n"); }
	void foo (x) int x (); { x(); }
	int main() { foo(bar);}
	
	As I understand both K&R C and the new draft, this program should
	not compile because the declaration of "int x ()" should conflict
	with the parameter "x" provided as an argument.

not quite. foo is defined in `old-style' and hence there's no argument
checking done for calls to foo like the foo(bar).

also, the type of argument x is automatically converted from
`int function' to  `pointer to int function' as stipulated at the
top of p. 218 in K&R2.

hence, there are no errors.