>>>>> "RA" == Russ Allbery <firstname.lastname@example.org> writes:
RA> That's going to be a long list. It includes Linux and most SysV Unixes
RA> I'm aware of. The fact that suid scripts work on Solaris is the
RA> exception, not the rule.
That's not correct; it works fine on Linux. The kernel ignores the setuid
bit as it should and sperl properly emulates it. It works on Digital Unix,
too. That comprises all of the platforms I have available to me to test
on, which is why I continually say that someone else is going to have to
come up with a wrapper because I can't test it. I like the three-line
wrapper myself; the complex ones only make things difficult.
Alternately, the whole client-server model will eliminate the need for
setuid scripts completely. I was planning on saving it until after
release, but if someone feels compelled then I'm all for making it happen