Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: $Revision: 1.6.2.16 $; site ISM780B.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!gamma!epsilon!zeta!sabre!petrus!bellcore!decvax!ISM780B!jim
From: jim@ISM780B.UUCP
Newsgroups: net.bugs.usg
Subject: Re: Re: Sv.2 suid bits (why a RESTRICTED
Message-ID: <29800007@ISM780B.UUCP>
Date: Fri, 13-Dec-85 18:36:00 EST
Article-I.D.: ISM780B.29800007
Posted: Fri Dec 13 18:36:00 1985
Date-Received: Mon, 16-Dec-85 05:36:28 EST
References: <699@abic.UUCP>
Lines: 27
Nf-ID: #R:abic:-69900:ISM780B:29800007:000:1116
Nf-From: ISM780B!jim    Dec 13 18:36:00 1985


>    Does anyone know why the stupid test for 'r' in the progname.
>    I thought I'd left that sort of junk behind with fortran!

The real problem is an error in the design of #!, and that error appears
to have been propagated everywhere (so at least it is consistent).
When excuting a #! file, the kernel passes (the last component of) the
file name as argv[0] to the program being run (e.g., sh).
Thus, not only does any script containing an r invoke a restricted shell
because the shell checks its argv[0] to decide whether it is to run
restricted, but also when programs which use their argv[0] to determine
how to identify themselves in error messages are run via #!, you get things
like
 
foo: syntax error in foo, line 47

instead of

<name of command>: syntax error in foo, line 47

And generally, any program which uses the name it is linked to to determine
its behavior cannot be used effectively with #!.

Whoever came up with this behavior no doubt thought at the time that it was
clever.  Such is the result of failure to think things through.

-- Jim Balter, INTERACTIVE Systems Corporation (ima!jim)