Code Review: 19compat

tfpt review “/shelveset:19compat;REDMOND\tomat”
Comment :
Adds LibraryPath19 option. This option allows to set paths to 1.9 libs
and is used in -1.9 mode.
Implements Kernel#trusted?, Kernel#trust, Kernel#untrust. The flag is
always settable and not checked anywhere as we don’t support Ruby’s
security model.
Fixes Array#fill overload ambiguity.
Fixes Array#== and Hash#== to handle recursive structures properly.

Tomas

I know I’m a bit late but I have a question.

Implements Kernel#trusted?, Kernel#trust, Kernel#untrust. The flag is always settable and not checked anywhere as we don’t support Ruby’s security model.

Since IR doesn’t support Ruby’s security model, does this mean that
(for example) it’s really OK to have Object#clone not preserving the
untrusted state of an object or having Kernel#trust and Kernel#untrust
not raising an exception if an object is frozen, like happen on MRI
1.9?

On Mon, Jul 19, 2010 at 22:35, Tomas M.
[email protected] wrote:

Ironruby-core mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/ironruby-core


Daniele A.
http://clorophilla.net/
http://twitter.com/JoL1hAHN

The behavior differences you describe should be fixed. This doesn’t mean
though that you should base any security related decisions upon flags
like tainted and untrusted.
We are giving no security guarantees based upon these flags. The only
security barriers are those defined by CLR security system.

Tomas

OK thanks. I supposed it was just like for the taint status but the
“always settable and not checked anywhere” made me wonder if there was
something else.

On Fri, Aug 20, 2010 at 18:15, Tomas M.
[email protected] wrote:

 Adds LibraryPath19 option. This option allows to set paths to 1.9 libs and is used in -1.9 mode.

[email protected]
http://rubyforge.org/mailman/listinfo/ironruby-core


Ironruby-core mailing list
[email protected]
http://rubyforge.org/mailman/listinfo/ironruby-core


Daniele A.
http://clorophilla.net/
http://twitter.com/JoL1hAHN