mol wrote:Maybe we could bring HMG4 to life?
We should left backward compatibility in my opinion...
IMHO, if some coding effort must be done, it should be in the web direction, since it gives you access to ANY operating system (present and future).
The new GUI should be flexible enough to allow handle desktop and mobile browser clients (obviously, we should forgot backwards compatibility).
HMGSCRIPT (in its current design) is not the answer. simply because the HMG community does not embrace it.
I guess that the problem with HMGSCRIPT is that it requires that all client-side be coded in JavaScript. JavaScript is not particularly difficult, but, it is when you compare it with xBase.
So, we should create a web solution that allow us to code in Harbour and a in new language for web GUI.
The starting point could be around the Harbour web server (the same HMGSCRIPT uses) using server procedures written in Harbour, then, we should be able to include the GUI code in such procedures using HMG style commands on top of simple and elegant classes.
The way it will work is simple: These GUI classes will send to the client all the required HTML/JavaScript code to create the required widgets.
From the original HMGSCRIPT paradigm, the synchronous AJAX should be kept, since it allows us to think that application in the way we know and love, without to take care of callback procedures and other things not required for the kind of applications we code.
Another benefit of this approach, is a better source code protection, an issue I've not considered in the original HMGSCRIPT design.
It could appear to be extremely difficult, but it is not, since a lot of HMGSCRIPT current code can be easily reused with minimal modifications.
The implications are fantastic: You can bring to the user an application that he can run in any device at any place.