[6712] in Athena Bugs
vax 7.1H: are exit codes signed or unsigned?
daemon@ATHENA.MIT.EDU (daemon@ATHENA.MIT.EDU)
Wed Dec 26 18:35:56 1990
Date: Wed, 26 Dec 90 18:35:43 -0500
From: "Jonathan I. Kamens" <jik@pit-manager.MIT.EDU>
To: bjaspan@ATHENA.MIT.EDU
Cc: bugs@ATHENA.MIT.EDU
In-Reply-To: Barr3y.Jaspan@ATHENA.MIT.EDU, bjaspan@mit.edu's message of Thu, 20 Dec 90 19:41:28 EST <9012210041.AA15755@yaz-pistachio.MIT.EDU>
Date: Thu, 20 Dec 90 19:41:28 EST
From: Barr3y.Jaspan@ATHENA.MIT.EDU, bjaspan@mit.edu <bjaspan@ATHENA.MIT.EDU>
It has a bug that causes it to exit with return status 11111111b.
Yesterday, I was on an RT when it happened, and the shell reported
that xrn exited with status 255; today, on a vax, the shell reported
an exit status of -1.
I assume that one of them is *wrong*. This could potentially break
some programs/shell scripts that depend on return values being
correct.
I believe that your assumption is wrong.
The only definition of exit status is that it is an eight-bit field in
the wait union. The definition does not say whether those eight bits
represent a signed or an unsigned value, just that they represent
eight bits of data.
The numbers 255 and -1 are equivalent when interpreted as an eight-bit
value. Therefore, the shell is free to report an exit status of
11111111 as either -1 or 255, at its discretion.
Shell scripts that rely on the signed/unsigned nature of a valud that
is not specified as being signed or unsigned are going to lose in some
situations no matter what. A shell script that really needs to get
the "real" exit status of a process should add 256 to whatever it gets
and take the result mod 256 as the exit value.
Jonathan Kamens
Project Athena Quality Assurance
