[joe_the_user]

As I recall, Rubinius did things like attempt to document the Ruby Language through a series of test suites rather than the standard formal methods (Edit: they did have the quandary that Ruby had without much exact specifications but still, no number of test by themselves can formally capture a language).

Having play with "Test Driven Development" just a bit, I feel like it tended to involve a lot of arrogant posturing with a flavor of "who needs design, who needs formal specs, if you've got the balls to write test and dive into the code, you can do anything without knowing anything".

It's not entirely unpleasant to see such reasoning break on the shoals of reality.

Edit: This sounds mean spirited but it seems like tail of the problems of 2007-8 programming styles.

[zem]

i have no idea either why you are moved make this comment, or why you are making it here. rubinius needed a test suite to ensure that they were compliant with a language whose 'spec' consisted of the mri implementation. they decided to do so in a way that would be useful for everybody, and started what is now the rubyspec project.

both rubyspec and rubinius are alive and well, so there's really nothing that has "broken on the shoals of reality" or would justify your evident schadenfreude at the prospect.

[evanphx]

RubySpec is a huge success in the Ruby community amongst implementations, I'd hardly call EY ending sponsorship a blemish for RubySpec.

[astrodust]

Creating a formal spec, while not trivial, is not nearly as hard as writing strict tests that validate an implementation against the spec.

If you've just played with TDD and never used it in a serious capacity, you're in no position to comment and your ignorance is almost so bad as to be offensive, especially your seeming delight at this strategy being perceived as having failed.

Code without tests is code that cannot be trusted.

[nwmcsween]

Because formal specs are open to interpretation, have you seen how much POSIX formal specs differ between operating systems?

[pjmlp]

Sadly yes, as I wrote UNIX portable code in 1997 - 2002 timeframe.

Unfortunely many on HN tend to think all UNIX are alike, which just shows how little have experiece writing portable UNIX system software.