Thanks, matz,

On Sun, Jul 6, 2008 at 12:22 PM, Yukihiro Matsumoto <matz / ruby-lang.org> wrote:
> According to Nobu, those tests mix buffered and unbuffered IO
> operations, which exact behavior rely on underlying stdio.  If it's
> true, I think the tests should be updated.

The offending spec were updated to not rely on paltform-specific behavior.

And with that, RubySpec run with 1_8_6 branch is 100% clean for me on
Linux, and I hear the guys who run the specs on Mac (Arthur, Federico)
say that they have clean run on MacOS X.

So, from rubyspec perspective, we're good to go with upcoming release of 1.8.6.

I see 6 failures for the latest MRI 1.8.7 (4 for Set and 2 for
CGI#Cookie), we'll investigate.

Thanks,
  --Vladimir