[ruby-core:74355] Ruby+OMR: how should we proceed?

Hi everybody,

I'm looking for your advice on how and where best to publish source code 
for the Ruby+OMR work that was presented at Ruby Kaigi. Hopefully you'll 
forgive my long note! I wanted to make sure everyone had enough background 
for the questions at the end.

I am the lead for the Eclipse OMR project which builds runtime components 
for use in all kinds of language runtimes (like Ruby!). Many of these 
components originally came from the IBM J9 Java Virtual Machine.  We 
recently kicked off the Eclipse OMR project at the Eclipse Foundation with 
code hosted at GitHub:
        https://github.com/eclipse/omr

Back in December, we presented at Ruby Kaigi on our experiments using 
these components inside CRuby:

    * Matthew Gaudet's "Experiments in sharing Java VM technology with 
CRuby"
      
http://www.slideshare.net/MatthewGaudet/experiments-in-sharing-java-vm-technology-with-cruby

    * Robert Young's and Craig Lehmann's "It's dangerous to GC alone. Take 
this!"
      
http://www.slideshare.net/craiglehmann/the-omr-gc-talk-ruby-kaigi-2015

Since many of the OMR components are now available as open source, we are 
in the process of updating our Ruby+OMR Technology Preview release that we 
made available in mostly binary form in December at 
https://github.com/rubyomr-preview/rubyomr-preview to now include all the 
source code needed to build with the open OMR components. We'll contribute 
the "Ruby glue" code, which connects the language agnostic OMR components 
into the CRuby runtime, to the OMR project initially under its license.

We had the thought to put all this code on GitHub (unless you'd prefer it 
somewhere else?), but we don't want anyone to think we're forking CRuby. 
We are NOT forking CRuby. We will donate all the source code needed to 
consume the OMR components to the CRuby community, if that's something 
you're interested in. The open code will already be licensed to enable 
your use anyway, if you choose.

But we thought you might like to take a look at it first :) . That's why 
we created the Ruby+OMR Technology Preview project in the first place. 
Also, since not all of the OMR components are in the open yet (for 
example, the OMR JIT compiler is not yet available in source form), we 
still want people to be able to try those parts as used for Ruby via our 
Docker images for Linux on x86-64, OpenPOWER, and LinuxONE.

We're really keen to get your feedback!

Before we do anything, we thought we'd reach out here to see how you would 
prefer us to proceed. Our thinking was to create a git repository with the 
baseline Ruby 2.2.3 release we used along with a single commit for the OMR 
changes needed to build Ruby+OMR.

Moving forward, we'll update the base Ruby code to be more current as well 
as adding and augmenting the source components for the remaining OMR 
components as they become available in source form at the Eclipse OMR 
project. 

What do you think? Is this of interest to the CRuby community? Would you 
prefer that we proceed any differently than I outlined above?

--mark




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