[patch] save ~600B

[Paul Fox]

wharm wrote:
 > Bernhard Fischer wrote:
 > > On Mon, Jun 05, 2006 at 01:04:06PM +0200, walter harms wrote:
 > >> hi bernhard,
 > >> just a minor question.
 > >> Does exit(-1) make sense ? so far i remember the shell expects 0-255.
 > > 
 > > IMO it does make sense. It it a means to make sure to return ~0, no
 > > matter what return type the shell happens to use.
 > > So, for "the shell" -- which likely happens to use unsigned char -- we
 > > return 255, while we wouldn't have to change all explicit calls to
 > > {_,}exit(255) iff the type would change, for whatever reason.
 > > Makes sense?

only if that's what the programmer intended.  in my experience,
people that write "exit(-1)" usually are confusing the exit
status or the program with the return code from a C function, and
are forgetting that the traditional error exit status for a
program is "1", or some other small non-zero integer.

 > personally i am the fan of
 >    exit( EXIT_FAILURE ) ;

and "EXIT_FAILURE" is cleverly defined as "1", right?  why
not just "exit(1)"?  this keeps it simple, concise, and transparent.

 > we should define a way in the style guide.
 > and get rid of stuff like:
 >    exit(ERROR);

i agree.  but "s/ERROR/EXIT_FAILURE/" accomplishes nothing.

paul
=---------------------
 paul fox, pgf at brightstareng.com