🚀 go-pugleaf

Hello,

I presume since you are reading this message that you are either
developing distributed applications already, or are interested in doing
so...

For my undergraduate project in the coming year I am planning to design
& implement a class-library which facilitates the rapid development of
distributed applications, and ensures that they are portable and easy to
maintain.

My research for this project seems to fall into two categories :-

(a) Good practice for writing reusable C++ component libraries.
(b) Problems with conventional network programming.

For the former I am reading a lot of C++ theory about good OOA/OOD,
however for the latter I feel I need information from a wider base of
people: YOU!

In order to get a better feel for my problem - and hopefully the correct
slant on the solution - I need answers to the following :-

1. What sort of distributed applications do people want to write?
2. What makes them difficult to develop?
3. What makes them difficult to maintain?
4. What would a distributed application programmer like in an ideal
world? (That is, high/low level library, etc...)

My current thinking - which I encourage you to attack - is for a
multi-layered library. Something like the following :-

1. Transport layer
- Encapsulates sockets & name-resolution services.

2. Application layer
- Hides sockets and provides some sort of "nice" metaphor to network
communications. (A bit vague?)

3. Distributed layer
- Hides networks and allows creation of objects which exist on multiple
machines & communicate by calling remote methods. (Very vague?)

Maybe you think I am misguided, not looking far enough forward (I am on
the java list, I know what's going on there) or perhaps there's an
outside chance that I am on the  right track ?!?

ONLY YOU CAN TELL ME!!!

I would appreciate any and all responses, especially positive criticism,
pointers to previous efforts, etc... 

If you want to reply by email, could you also post to the newsgroups as
I would like this topic to generate some discussion!

Many thanks,

Matt.

P.S.#1: I hope to make something (either source or compiled libraries)
available during the projects development, and will be hosting the
project on a WWW site where people can come and look and see what I and
any other interested parties may be doing.

P.S.#2 : Please indicate if you would be interested in taking part in
any future discussions on this topic.

This project is sponsored by:-
	Coyote Ltd. (UK)
		Medical & Network technology providers.

	Novell Inc. (San Jose, CA)
		Networking Technology.



----
Matt Mower, Internet Information Services
University of North London.    Email: M.Mower@unl.ac.uk
Personal Page: http://www.unl.ac.uk/~mat/
See http://orca.unl.ac.uk/~mat for more details.