[Orbit-python-list] ORBit-Python Roadmap

[Jason T. <ta...@li...>]

Over the last couple of months, ORBit-Python has received a considerable
amount of attention by myself and the other active developers and users.
Hacking has been steady, especially since the crew in #orbit-python
(OpenProjects Net) has been keeping me motivated. :)  There have been
some pretty interesting developments in the community, particularly
Johan Dahlin's work with OAF and Bonobo bindings.

I am especially curious to hear comments from people using O-P.  If
you're using it, have used it, or know someone that's using it, please
let me know how it held up.  Were there bugs that didn't get addressed?
Missing features?  Voice your opinions!

I've been mentally sketching out a roadmap for the next few versions.
Here are a few of my thoughts.  Comments welcomed.

--- snip ---

0.3.0
-----

Quite a bit has changed since 0.2.0 was released, and I'd like to get
0.3.0 out the door by next month.  Many areas have received more polish
and much closer spec compliance, especially with respect to system
exceptions, object references (implemented _is_a, _is_equivalent,
_hash_, _non_existent, _narrow), and the POA.

Some pretty large changes went in recently to make object references
actual class instances, derived from CORBA.Object.  (Originally all
object references were created from a type object.)  Objrefs now behave
like real instances, so you can do things like objref.__class__.__dict__
and it will behave as you'd expect.

Johan Dahlin's work has also involved creating an external ORBit-Python
API, similar to PyGtk's API.  This release should make the API useful.

Before 0.3.0 is released, these changes will have to stabilize and be
tested.  Also planned for 0.3.0 is a comprehensive test suite, to
validate and stress test _all_ of O-P.  I would also like to include a
number of small examples to compensate for the lack of decent
documentation.  Submissions graciously accepted. :)


0.4.0
-----

For 0.4.0, the ORBit module will be created, and all O-P proprietary
functionality will be put there (_load_idl(), for example).  It will
also let multiple ORBs coexist.  e.g.:

   from ORBit import CORBA
   import omniORB.CORBA

There are a few small features I have already planned, and those that
don't make it into 0.3.0 will be implemented for 0.4.0.

Any other medium to large changes or features that get requested will
likely be implemented for 0.4.0.  For this release, I want ORBit-Python
to be as spec-compliant as possible.


0.5.0
-----

This release will focus on performance and stability.  Performance is
already pretty good (still the fastest Python ORB that I'm aware of),
but some areas can be tightened and optimized.

Stability means _zero_ known bugs, and _zero_ known memory leaks.  I
will not release 0.5.0 with an unresolved bug or memory leak.  Some
memory leaks and bugs will be obviously addressed for 0.4.0, but for
those that are left, they will be closed for 0.5.0.  Period.

Basically, if 0.5.0 crashes or has problems, they're going to be
ORBit's. :)  That's the goal.  0.5.0 will be as close to production
quality as is possible.

--- snip ---

So, that's the game plan.  You'll notice that no dates were attached to
these versions.  That's because I have no clue. :)  I'd like to see
0.5.0 released by August, however.  It depends entirely on my own
schedule, and the amount of activity by the other developers.

Jason.