Re: [Gnumed-devel] Time for a major re-think in 2005 - opinions please.

[Horst Herb]

On Thu, 6 Jan 2005 14:02, Richard Terry wrote:
> doesn't say 'hey I don't like that - why don't we do x,y,z, or 'I don't
> think that will work'. Do so, but let the gui-designer then experiment and
> make the modificications, and if at the end of the day he decides, 'no we
> will do it this way', then so be it, the backend coders will have to work
> around that.

I have always advocated clean separation of GUI design, GUI functionality, and 
non-UI backend code.

You can use wxGlade to "paint" the user interface (elements),
then a coder *inherits* the autogenerated (and *never* manually modified) GUI 
code and fills it with functionality.

If it is done that way, designers can do their work, and coders can do theirs, 
without standing each other in the way ....

... except if you require weird window manager magic which to you is 
appealing, but I cannot cope with a user interface where I don't see all 
elements in a way that I can consistently access them at any time regardles 
of context. My brain is obviously wired different from yours, and Karstens 
and Syans brains must be wired different yet again - point I am trying to 
make is that there won't be one single design that makes everybody happy, 
because we think and perceive too differently. Meaning we need to keep the 
code and interface design flexible enough to make any user interface element 
redesignable and interchangeable without much coding, meaning we need to 
stick to an architecture that allows this.

Under Karsten's leadership, the backend has come a long way. The middleware is 
getting there. Some GUI is functional, other isn't quite there yet - but the 
GUI is only a small fraction (albeit the only visible one to the end user) of 
the work, and ***NOT*** the most significant one!

The main purpose of medical software is making data collectible and accessible 
in a *safe* and *reliable* way, and in a way that will not prevent us from 
reusing existing data in any perceivable way. Just hacking a GUI and filling 
it up with functionality will not get us there - this is how "Medical 
Director" emerged, Australias "market leader", and look how crappy it is 
under the hood - wouldn't entrust it any data at all, regardless how much 
people like the user interface (I know you don't like it at all, but the 
majority of Australian GPs seems to like it).

Horst