[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[DotGNU]Fullback webservices

From: Peter Minten
Subject: [DotGNU]Fullback webservices
Date: Sun, 05 Jan 2003 10:30:00 +0100

Hi folks,

as you know I'm currently working on a first example implementation of the
butler. It appears to me that the current mode of operation for webservices, put
a frontend on the client for GUI stuff and connect it to the backend on the
server, is not the future. The basic idea of webservices is run everything on
the server with no program code on the client.

Though this is not possible at the moment it will be in the future. The idea I'm
having is this: in dynamic languages like Python (IIRC), Ruby, and perhaps C#
(if Rhys and co make Reflection really worthwhile) classes can be constructed at
runtime, so it's possible to create Fox widget clases at runtime (and of course
realize them). This can be used to start a program purely on some translator
code and then let the translator and some helper classes construct an
FXMainWindow descendant and realize it. Get the idea?

Say this works. The GUI can be created dynamically, no program GUI code required
anymore on the client. But how about other code that needs to be on the client,
most importantly I/O code. With frontended webservices I/O is done using the
languages standard I/O interface, a grave security risk. We could solve this
problem by creating Fox widgets on the client side that do the I/O, the
application could then use the usual mechanism to do I/O.

The big advantage of this (besides removing application frontend code from the
picture) is safety. Since we supply the I/O classes we can use them to implement
a DotGNU Security System (DSS). I'll explain the DSS in another message.



reply via email to

[Prev in Thread] Current Thread [Next in Thread]