I understand your point is

* CRuby uses test-all. It°«s not good for other implementations.
  * someone need to reimplement tests
* test-all (and rdoc) doesn°«t describe the behavior well.
  * test re-implementer needs to read CRuby source
* ruby/spec is better than test-unit
  * ruby/spec encourages to write the description
  * ruby/spec°«s example is shorter and usually atomic
  * ruby/spec has tags for known failures
* CRuby people should write ruby/spec
  * change CRuby and ruby/spec atomically
  * single person write implementation and spec
  * use ruby/spec instead of test-unit

But
mspec doesn°«t have enough utility in envutil, like assert_separately.

Like above mspec is not as robust as miniunit.
Even though ruby/spec°«s example is smaller than test-all,
mspec has more abstraction layers than minitest+test-unit wrapper.
If it hits mysterious SEGV, it makes hard to debug.
(I experienced such hard debugging some times through the experience)

Therefore it is not acceptable to use mspec as primary framework of
testing CRuby especially for core classes.

Just as a note
* test-all also has tag feature; JRuby uses that for these days
* CRuby already has test-rubyspec to ease running rubyspec.
  (But if people actually change rubyspec, they need to run it with
other version of rubies)


I agree with tests should be more descriptive.

-- 
NARUSE, Yui  <naruse / airemix.jp>

Unsubscribe: <mailto:ruby-core-request / ruby-lang.org?subject=unsubscribe>
<http://lists.ruby-lang.org/cgi-bin/mailman/options/ruby-core>