Re: c/3246: gcc -Wconversion doesn't work properly

["Joseph S. Myers" <jsm28@cam.ac.uk>]

The following reply was made to PR c/3246; it has been noted by GNATS.

From: "Joseph S. Myers" <jsm28@cam.ac.uk>
To: <ohhara@postech.edu>
Cc: <gcc-gnats@gcc.gnu.org>,  <gcc-bugs@gcc.gnu.org>
Subject: Re: c/3246: gcc -Wconversion doesn't work properly
Date: Tue, 19 Jun 2001 13:13:33 +0100 (BST)

 On 19 Jun 2001 ohhara@postech.edu wrote:
 
 >  I use many warning options to find my small ( or big ) mistakes.
 > Therefore, I tried to use -Wconversion warning option. However, it
 > doesn't seem to work properly.
 
 -Wconversion isn't meant for finding mistakes in new code - only for an 
 initial shake-down after converting ancient (predating the adoption of the 
 first C standard in 1989) code to modern C with prototypes.  The 
 documentation could be clearer, and there could (and should) be a more 
 useful version designed to warn about cases relevant to modern code (such 
 as sign differences causing security holes) only, but that isn't what 
 -Wconversion was designed to do.
 
 > Why does it prints warning message? Even if I changed "unsigned char"
 > to "unsigned short", the same warning message would be printed.
 
 Because with an old-style function definition and no prototype - the sort
 of program -Wconversion supposes you have just converted from - the
 argument would have been passed as an int.  This may not be a useful
 warning, but the present -Wconversion isn't particularly.  See the GCC
 projects list.
 
 -- 
 Joseph S. Myers
 jsm28@cam.ac.uk