From: Erik Veenstra <google@...>
Date: 2005-01-18T10:01:06+09:00
Subject: Re: Ignoring RUBYOPT ?

> > > > How can I let Ruby ignore RUBYOPT?
> > > >
> > > > Both -T1 and unsetting the environment variable aren't
> > > > possible in my situation. I don't want the rest of the
> > > > script to run in safe mode 1 and I don't have control
> > > > of the environment...
> >
> > Mmhh... I might use an intermediate Ruby script to unset
> > the environment variable... Not pretty, but it might
> > work....
>
> It's the last line of the second test which is interesting.
> Promising?

Consider the following:

$ ruby -T1 test1.rb test2.rb

In which test1.rb (aka bootstrap script) does something like:

ENV.delete("RUBYOPT")
system("ruby #{ARGV.shift}")

.... and test2.rb is the real stuff.

It does work, really. But...

There's one potential problem, according to the Pragmatic
Programmer's Guide, $SAFE==1 means: "Can't start processes from
$PATH if any directory in it is world-writable.". So, it's not
guaranteed that the inner ruby can be launched.

I've tested this and it results in an "Insecure operation -
system (SecurityError)".

Still there?...

What about this:

$ ruby -r test1.rb -T1 test2.rb test3.rb

test1.rb is the above mentioned bootstrap script, test2.rb is
empty and test3.rb is the real stuff. Because "-T1" occurs
after "-r test1.rb", this test1.rb is run in safe mode 0, so it
can run any process it wants. test2.rb is there, because Ruby
needs an application script, not just "-r"'s and parameters.

The result of all this is that -T1 is encounterd after parsing
the commandine part which does the real work, but before the
end of the commandline where RUBYOPT is parsed! (Well, *not*
parsed, because, in the meanwhile, we run in safe mode 1.)

Bingo!!! It works!!!

What a hack!
What the heck!

I've to rethink it tomorrow... It's still not pretty...
gegroet,
Erik V.