Bringing Bonobo to the masses :)

[ERDI Gergo <cactus cactus rulez org>]

Hi,

It's became clear to me that due to university duties, I will be unable to
start working on Bonobo-- and keep on working with MonkeyBeans. This means
two things: first, I will have to choose one to work on. Second, some help
would come very handy. Please read these two short descriptions of both.
(please note that `Bonobo' means `the Bonobo infrastructure' and
`GNOME Bonobo implementaton' means `the C implmenentation
and the GtkObject-based API')

About MonkeyBeans
=================
My goal in starting MonkeyBeans was research only. It is a Java-based
Bonobo API and implementation. It already has some working code (see the
`monkeybeans' module in CVS), with some low-level JNI magic I've even
embedded Evolution calendars into AWT windows. It doesn't seem possible to
have a 100% pure Java Bonobo implementation because of the BOnobo
infrastructure (including things like OAF). I do not consider this a bug,
since Bonobo is aimed at Unix systems.

TODO for MB
-----------
A lot of interfaces are not implemented yet. I would like to go quick and
dirty and do shortcuts like implementing Stream based on files and
Storages based on file systems (i.e. not having something like EFS). I'm
not sure choosing JNI for the native parts is an optimal choice, other
possibilities should be evaluated. I'm pretty sure Controls will not be
fun to be done Correctly (both on the container and the containee side).

Why should MB be continued?
---------------------------
I think Java (and running Java in JVM's) is different enough from the
GNOME Bonobo implementation that it can be used to stress-test the Bonobo
design (even with the minimal MB stuff I already have, I've found some
problems in BOnobo, all of them was reported to gnome-components-list).
Remember the whole point of Bonobo using CORBA is not to be easily
flameable by KDE people thinking we actually tunnel the GUI thru CORBA,
but for language-, machine- and network-independence.
Another aspect of why MB woudl be useful is a lot of commercial software
vendors who got into the Java hype. I would hope them to use Bonobo if
it's easily available to them, which means Java-like APIs and very good
integration with AWT and/or Swing.


About Bonobo--
==============
Recently, there has been a lot of discussion about extending GNOME-- from
its current scope (wrapping libgnomeui for C++) to various aspects of a
modern programming framework, including wrapping some other parts of
GNOME. An ideal C++ API for Bonobo would wrap some of the GNOME Bonobo
implementation where it makes sense, and create its own implementations
where it would enable a nicer, more `C++ish' API (for example, the
GNOME BonoboPersist* objects are only syntactic sugar that wouldn't be
needed when using a C++ ORB),

TODO for Bonobo--
-----------------
Please note that no actual Bonobo-- code has been written to date. I think
that pretty much sums up what's needed to be done for Bonobo-- :). First
of all, a nicer name should be found. Then, we should examine which parts
of the GNOME Bonobo implementation to wrap, and what interfaces to
implement ourselves.

Why should Bonobo-- be started?
-------------------------------
Personally, I would like to use Bonobo-- in the future. That really can't
be told about MonkeyBeans (:)). On a more serious note, if GNOME-- really
wants to become a complete application development framework, it
doubtlessly needs Bonobo support. Also, it is my perception that C++ is
used more frequently than Java for open source projects.


So, which one do you think would be more useful to work on? Who would be
interested in lending a hand in one of these projects?

-- 
   .--= ULLA! =----------------------------.  finger cactus cactus rulez org
   \      http://cactus.rulez.org           \   for PGP public key
    `----------= cactus cactus rulez org =--'
Egér: az elefánt japán változata