See, iirc, Matz years ago specifically said "fork Ruby". He wants lots of different versions which I feel to some extent means diverging from a formal spec, and then occasionally maybe coming back to it.
And that's even worse. I'm working on a Ruby compiler. While it is far from being at a stage where this is a real problem, currently there's no sane way of knowing whether or not I'm "close enough" to MRI other than extensive testing of every single piece of Ruby code I want to work.
"Diverging from a formal spec" is one thing. As it is, many versions of MRI has diverged from how the core team thinks it works. If MRI is "the spec", then Ruby changes from revision to revision as the test coverage is nowhere good enough to prevent regressions or changes in behaviour.
Basically: Nobody knows Ruby.
Frankly, I very much hope that another Ruby implementations overtake MRI sufficiently in quality to out compete it enough to shift the initiative of definining the language to a more responsible team.
"Diverging from a formal spec" is one thing. As it is, many versions of MRI has diverged from how the core team thinks it works. If MRI is "the spec", then Ruby changes from revision to revision as the test coverage is nowhere good enough to prevent regressions or changes in behaviour.
Basically: Nobody knows Ruby.
Frankly, I very much hope that another Ruby implementations overtake MRI sufficiently in quality to out compete it enough to shift the initiative of definining the language to a more responsible team.