[nateberkopec]

Rubyspec is a great project for all the reasons Brian outlines.

However, it's also a failure - in part, due to Brian and his attitude towards contributors. See this twitter conversation: https://twitter.com/the_zenspider/status/547527644535726080 He's been grinding on Rubyspec for years, bless him, but I think there's a reason why he was unable to rally the community behind his effort, both in terms of gathering more contributors and in making it "official" in terms of the language spec.

JRuby, currently the only non-MRI ruby implementation that you can seriously consider for production use, runs the MRI test suite against JRuby. RubySpec is far from The Only Solution, though Brian would like you to think it is.

[jfaucett]

"RubySpec is far from The Only Solution"

you're right, but I remember learning quit a bit of ruby by working through RubySpec back in the day. and I remember liking how well structured, logical and organized the suite was, very exemplary. so I'm sorry to see it go. I have no perspective on Brian or the mri devs so I can't say anything about the situation around it, I just liked the code and spec and found it very well done.

It would certainly be nice to have such a ruby specification and test suite that was endorsed by ruby community at large (unlike what I saw in those example MRI tests).

[headius]

I came to the thread to make sure someone was spreading truth rather than FUD, and this was the first comment I saw. Bless you.

[rmoriz]

>However, it's also a failure - in part, due to Brian and his attitude towards contributors.

Looks like he seems to be burned out - maybe he just wants something in return for his time investment e.g. a paid job or some other gratification in return for his work.

Btdt - you put a lot of work into something, people/companies start benefit from it but they don't return anything important (monetary compensation is not a bad thing -it allows one to pay the bills/go on holiday/get a life).

We've seen this "pattern" a lot in the OSS world, but saying a different attitude would "fix everything" is a lie. We (as in the Ruby community) need to crowd-fund independent developers who are not paid by companies for their OSS contribution IMHO.

[headius]

Brian has been paid to work on Rubinius and RubySpec full time for at least 6 years.

[rmoriz]

oh - didn't know that. Okay that makes a hughe difference IMHO.

[btown]

> JRuby, currently the only non-MRI ruby implementation that you can seriously consider for production use

Are there concrete shortcomings when using Rubinius in production, or is it just not battle-tested yet?

[cheald]

Every time I've tried to use it, it's either been unstable (segfaults are fun) or orders of magnitude slower than MRI.

It's a promising project, but as far as "no GIL, true concurrency, mature GC" implementations go, JRuby delivers the goods.

Edit: For kicks, I'm trying to boot my primary app with RBX right now. RTLK threw some "id for nil" exceptions due to some initializer stuff, so I disabled that, and now, trying to run my test suite, after 2 minutes of sitting there doing who-knows-what, it finally attempts to boot my app and dies with a "Missing constant" error despite the fact that I put in debug code to ensure that the file containing said module is a) loaded, b) executing properly, and c) that said constant actually exists in the module tree.

This code boots and runs just fine under MRI 1.9, 2.0, 2.1, and JRuby 1.7. I might be able to get it running with some more work and explicit load path management, but I'm off to a party for now, so I'll poke at it a bit when I get back.