ode-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From James Strachan <james.strac...@gmail.com>
Subject RE: Ode / BPEL Donation of BPEL 2.0 Engine
Date Mon, 20 Feb 2006 15:07:31 GMT
On 20 Feb 2006, at 11:04, ode-dev-help@incubator.apache.org wrote:
> From: "Noel J. Bergman" <noel@devtech.com>
> Date: 18 February 2006 06:12:10 GMT
> Subject: RE: Ode / BPEL Donation of BPEL 2.0 Engine
>
>
> Considering that people have expressed the clear desire to develop  
> joint
> work, an umbrella-style project housing multiple implementations of  
> the same
> domain, although probably workable for a period while the  
> community's active
> focus is on the joint solution, if persisted into the long term  
> usually
> leads to a balkanized project that breaks up.

If the project has to split up into some different pieces thats fine  
too. There's no reason why there has to be just one engine for all  
our orchestration/BPEL/workflow needs. But if we can end up with just  
one, hey thats great.


> The ASF has, as I noted on general@, no need to deal with independent
> releases of the vendor products.  For that matter, we couldn't care  
> less
> about anyone's delivery schedules.  Our releases occur when  
> determined by
> our PMCs.  Vendors may have their own schedules, but our focus is  
> building
> an ASF project.  If someone wants to do a release on their  
> schedule, they
> are free to pull source code and use it in accordance with our  
> license.
>
> James mentioned that ServiceMix has some needs to work with  
> Sybase's code.
> I am sure that no one, *especially James*, wants anyone to have the
> impression that ServiceMix is trying to co-opt the direction of this
> separate project.

Agreed - particularly as we've been working with PXE for over 6  
months...


> The Sybase code will be available for anyone, including
> ServiceMix, to use as necessary, so you should feel free to pursue  
> whatever
> direction is jointly chosen by this community.

Agreed.

I'm now taking off my Ode hat and putting on my ServiceMix hat just  
to make things completely 100% explicit so you can understand exactly  
where I've been coming from since I've been raising my little red  
flag...


Lets assume that the Sybase code is imported into Ode (I'll call it  
Foo from now on) and the PXE code is imported (lets call that Bar);  
so Foo and Bar are both Ode code bases in the org.apache.ode  
namespace, anyone can use them at Apache and they have nothing to do  
with vendors releases etc. The first day they are imported Foo and  
Bar will not reuse any of each others code; then over time we figure  
out how to refactor Foo and Bar to reuse code etc.

Now with just my ServiceMix hat on, I don't really mind if Foo and  
Bar merge together 0%, 10%, 50% or 100%. So whether or not Foo and  
Bar completely merge doesn't really interest me a whole lot right now  
- I'm totally happy if further down the road we decide that Foo and  
Bar should not be completely merged and instead focus on what can be  
shared.


Now as soon as Foo arrives at Apache we'll be using it in ServiceMix  
and contributing patches to ease the integration and fix any issues  
we find. We've been using PXE for over 6 months & currently use a  
slightly old version of PXE. We tried to use the latest greatest and  
found some problems. So the very day that Bar is checked into Apache  
we will immediately move from PXE to Bar and work on the code to make  
sure it works inside ServiceMix which will probably result in some  
patches for Bar.

So before the real meat of the unification gets done on Ode, we'll be  
working on both the Foo and Bar codebases and providing an  
integration test to both libraries and their integration with JBI.  
This integration test will be useful as the unification process takes  
place - plus there's no better way of providing input on 2 codebases  
than by using them both :)

The little red flag I've been waving lately - which in the grand  
scheme of things is no big deal - is purely that as an end user of  
both Foo and Bar, I'm going to want milestone releases of Foo and Bar  
fairly soon, plus at significant points of the unification. Up to now  
an incubating podling tends to just have 1 release; whereas I think  
it makes sense for the Ode project to support multiple releases of  
Foo and Bar while we go on the unification journey and see where we  
end up.


BTW Dims, please don't see this little wave of a red flag as me being  
negative; I'm really excited by the Ode project and am sure its gonna  
be a great success - I just want the project to start on the right  
footing so we don't end up with unnecessary friction or worse,  
another Avalon.

James
-------
http://radio.weblogs.com/0112098/


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message