Date: Tue, 19 Dec 1995 18:57:30 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: "A.Appleyard" <A DOT APPLEYARD AT fs2 DOT mt DOT umist DOT ac DOT uk>
Cc: DJGPP AT sun DOT soe DOT clarkson DOT edu
Subject: Re: array arg funny (2)


On Tue, 19 Dec 1995, A.Appleyard wrote:

>   I compiled this program:-
> 
> int zxcvbnm(int i, int j, double x[i][j]){int k; k=2; return 42;}
> 
>   with djgpp as C and as C++. C++ faulted thus:-
> 
> t$$.cc:1: `i' was not declared in this scope
> t$$.cc:1: `j' was not declared in this scope
> Segmentation violation in pointer 0x00000008 at d8:5cebc
> 
> EXCEPTION OCCURRED!  Information dumped to core file:
>     "t$gccdbg"
> 
>   and C thus:-
> 
> C:\WORK>c:\djgpp\bin\gcc t$$.c
> fake(.text+0xe2): undefined reference to `main'   /*** I know why that is ***/
> 
>   (1) Why the difference?

Because C and C++ have different context rules.  But this program is 
illegal in both.  You can't have array dimensions declared as 
non-constant values.  C isn't Fortran.  The segfault is due to problems 
with GNU C++ compiler version 2.6.x which sometimes crashes for illegal 
programs.

>   (2) If I include this function in a C program, would it treat the bounds [i]
> and [j] correctly in the declaration of x, as in Fortran?

No.