[technology-pmc] Eclipse OMR project's product is "source code"

["Mark Stoodley" <mstoodle@xxxxxxxxxx>]

Making the technology-pmc group aware of
this discussion, as requested by Wayne.

Wayne's motivation: "It's a little
weird (for us) that a project not produce a binary, so I'd like the PMC
to be aware of the nature of the project and have a chance to discuss it."

Gist is in this comment:
On 18/03/16 02:08 PM, Mark Stoodley
wrote:
It does build: we designed a standalone
binary build that can be created and tested (make && make test).
But the resulting binaries are simply tests for the various core components.
So those could be considered "exemplary" implementations but
they aren't really very useful other than as tests. No other project, for
example, would want those binaries for any purpose that I can (naively?)
imagine.

The project is designed to be absorbed as source code components into other
language runtimes for them to build. We're working with the Ruby community
right now to create a pull request against CRuby to bring in parts of the
OMR project. IBM will be pulling the OMR code base in on every accepted
merge request to build our J9 JVM as well as some of the experimental projects
we're working on for other runtimes.

Of course, that's only the way it works for the components that are currently
in OMR.

It's possible that future components may be distributable in binary form,
so we may have to revisit this model at some point.

But for now, our "product" is source code.

Basically, OMR is a platform and so
it doesn't do anything by itself.

Some of these tests are pretty straight-forward
examples of how to use the facilities in the project (for example, the
thread and port libraries). The GC tests also show how to bootstrap, exercise,
and shutdown the GC component.

So they could be considered "exemplary"
implementations.  But as binaries, they are not really very useful
for anyone to consume other than to show that the test code does run and
works on different platforms. No one would use those binaries to actually
do anything.

Questions welcome.

Mark Stoodley		8200 Warden Avenue
Senior Software Developer		Markham, L6G 1C7
IBM Runtime Technologies		Canada
Phone:	+1-905-413-5831
e-mail:	mstoodle@xxxxxxxxxx
We cannot solve our problems with the same thinking we used when we created them - Albert Einstein

----- Forwarded by Mark
Stoodley/Toronto/IBM on 2016/03/21 02:52 PM -----

From:      
 Wayne Beaton <emo@xxxxxxxxxxx>
To:      
 Mark Stoodley/Toronto/IBM@IBMCA
Cc:      
 Dorra Bouchiha/Toronto/IBM@IBMCA,
Gary Liu/Toronto/IBM@IBMCA, Irina Rada/Toronto/IBM@IBMCA, Gunnar Wagenknecht
<gunnar@xxxxxxxxxxxxxxx>, David M Williams <david_williams@xxxxxxxxxx>,
Mike Wilson/Ottawa/IBM@IBMCA
Date:      
 2016/03/21 02:00 PM
Subject:    
   Re: Fw: The
Eclipse technology.omr project is complete!

---

Hi Mark.

I would like to involve the Technology PMC in this discussion. In fact,
I'd prefer that you move it to the technology-pmc@xxxxxxxxxxxmailing list.

It's a little weird (for us) that a project not produce a binary, so I'd
like the PMC to be aware of the nature of the project and have a chance
to discuss it.

I've copied Gunnar Wagenknecht, the PMC lead as an FYI.

Thanks,

Wayne

On 18/03/16 02:08 PM, Mark Stoodley wrote:
It does build: we designed a standalone
binary build that can be created and tested (make && make test).
But the resulting binaries are simply tests for the various core components.
So those could be considered "exemplary" implementations but
they aren't really very useful other than as tests. No other project, for
example, would want those binaries for any purpose that I can (naively?)
imagine.

The project is designed to be absorbed as source code components into other
language runtimes for them to build. We're working with the Ruby community
right now to create a pull request against CRuby to bring in parts of the
OMR project. IBM will be pulling the OMR code base in on every accepted
merge request to build our J9 JVM as well as some of the experimental projects
we're working on for other runtimes.

Of course, that's only the way it works for the components that are currently
in OMR.

It's possible that future components may be distributable in binary form,
so we may have to revisit this model at some point.

But for now, our "product" is source code.

Mark Stoodley		8200 Warden Avenue
Senior Software Developer		Markham, L6G 1C7
IBM Runtime Technologies		Canada
Phone:	+1-905-413-5831
e-mail:	mstoodle@xxxxxxxxxx
We cannot solve our problems with the same thinking we used when we created them - Albert Einstein

From:        Wayne
Beaton <emo@xxxxxxxxxxx>
To:        Mark
Stoodley/Toronto/IBM@IBMCA
Cc:        Dorra
Bouchiha/Toronto/IBM@IBMCA, Gary Liu/Toronto/IBM@IBMCA, Irina Rada/Toronto/IBM@IBMCA
Date:        2016/03/18
01:41 PM
Subject:        Re:
Fw: The Eclipse technology.omr project is complete!

---

We don't put any restrictions on how you build project code. We do, however,
require that project code be buildable (i.e. a mere mortal should be able
to pick up and build your code).

I'm curious... if you're not producing a binary release, what are you building?

Is there no opportunity for an exemplary product from this project?

Wayne

On 18/03/16 01:34 PM, Mark Stoodley wrote:
Hi Wayne,

Another question: there is interest within the OMR community to leverage
Travis-CI for builds. Given the OMR project does not actually produce a
binary release (just source drops), is there any reason we shouldn't or
can't use Travis-CI ?

Mark Stoodley		8200 Warden Avenue
Senior Software Developer		Markham, L6G 1C7
IBM Runtime Technologies		Canada
Phone:	+1-905-413-5831
e-mail:	mstoodle@xxxxxxxxxx
We cannot solve our problems with the same thinking we used when we created them - Albert Einstein

-- 
Wayne Beaton on behalf of the Eclipse Management Organization
@waynebeaton
The Eclipse Foundation