From: Meinrad Recheis <meinrad.recheis@...>
Date: 2008-04-04T18:28:32+09:00
Subject: accomplishing compatibility (was Re: RUBY_IMPLEMENTATION)

On Fri, Apr 4, 2008 at 11:02 AM, Meinrad Recheis
<meinrad.recheis@gmail.com> wrote:
> On Fri, Apr 4, 2008 at 10:37 AM, Yukihiro Matsumoto <matz@ruby-lang.org> wrote:
>
>
> > Hi,
>  >
>  >  In message "Re: RUBY_IMPLEMENTATION"
>  >     on Fri, 4 Apr 2008 12:06:20 +0900, Urabe Shyouhei <shyouhei@ruby-lang.org> writes:
>  >
>  >  |> Probably we need to introduce a new constant to represent
>  >  |> the implementation of the running VM.
>  >
>  >  |No I don't think that's a bright idea.  We need A WRITTEN SPECIFICATION
>  >  |instead.
>  >
>  >  Of course, a WRITTEN SPECIFICATION is a good thing to have.  But the
>  >  cost is not comparable to adding a single constant.  Besides that,
>  >  adding a constant would not spoil developers from accomplishing
>  >  compatibility.
>  >
>  >                                                         matz.
>  >
>  >
>
>
>
> also as history shows, the existence of the c++ standard did not make
>  the major c++ compilers fully compatible to each other ;)
>  -- henon
>
>

I know this has already been discussed some time ago, but IMHO it is
very important to point out that a suite of standardized tests which
all implementations are tested against is much more practical not to
mention much easier to provide than a written specification.

Have all test suites of all major implementations been merged into the
MRI test suite yet? I suspect that not all major implementations use
the MRI tests. Should we start a "ruby standardization test suite"
project which is maintained by all authors of ruby implementations
together? Are there any similar efforts already?

-- henon