So let me try my crystal ball on this one ;-)
By my experience, "script" is usually something that is run through an interpreter, so in a 64-bit environment a script might be using the environment's native script interpreter. So the interpreter should be 64-bit-compatible by default.
One possible source of errors could then be that the script does not get the response it expects from the system. For example, if the script is using "uname -m" or something similar to identify the architecture and does not know about the new 64-bit architectures (x86_64 or ia64), it will return an error.
There *is* a command that allows the system to "lie" about its architecture identifier to a given process and all its children. It is useful only on systems that have some multi-architecture capability. That command is "setarch".
On my home machine (Debian Etch, x86_64) I can do this:
$ uname -m
x86_64
$ setarch i386 uname -m
i686
You can also use just the name of the desired architecture as a command. This invokes "setarch" through a symbolic link. Setarch does the right thing automatically:
$ uname -m
x86_64
$ i386 uname -m
i686
Apparently you can also use the setarch command to escape from the fake 32-bit mode, if necessary:
$ i386 bash
$ uname -m
i686
$ setarch x86_64 uname -m
x86_64
$ x86_64 uname -m
x86_64
If running the script with "setarch i386" makes it work, it might be easy to modify the script to automatically identify the compatible 64-bit architectures.
MK
MK
