On 16/11/2012, at 11:13 PM, Hugh Brock wrote:
On Fri, Nov 16, 2012 at 09:50:08PM +0100, Justin Clift wrote:
<snip>
> Do you reckon it's worthwhile for the Aeolus Release Cabal to
look into having
> Torquebox be a "supported" upstream platform for Aeolus in a future
release?
>
> Obviously not near term, but down the track if/when it makes sense.
I think this is a Tech Cabal matter to start with.
Now that you mention it, that definitely makes sense. There's scope for
discussion, but yeah, they'd be the ones to say if it's going to be too
much effort and so forth.
To explain why I'll
restate what I said on IRC earlier re jruby/torquebox:
1. Torquebox and jruby are awesome and I would love to support
them. I've never heard a single negative report about compatibility and
performance and I think it would be a huge win.
2. However we have to be really careful that our source code remains
compatible with a bog-standard Ruby environment (i.e. MRI + standard
rubygems) or we will have undone all the hard work we've done making our
upstream palatable to normal Ruby developers. To be specific, any gem
with c extensions will not (by default) work in torquebox, for obvious
reasons. We would have to make sure that any torquebox services we use
are pluggable such that our code will use native Ruby alternatives when
we're not running in Torquebox.
3. I've been told that these kinds of compatibility problems are
lessening, and the fact that Charles, bobmcw, and mgoldmann are awesome
and will be anxious to help us succeed is a point in torquebox/jruby's
favor.
4. We will need to think about how to deal with production deployments
when running on torquebox, which requires EAP (JBoss).
All good points. :)
Regards and best wishes,
Justin Clift
--
Aeolus Cloud Evangelist
http://www.aeolusproject.org