Re: [Gnumed-devel] Comments on 0.2.......................

[Karsten Hilbert]

On Mon, Jun 19, 2006 at 10:45:06AM +1000, Richard wrote:

> There is something fundamentally wrong with the way you are approaching the 
> writing of gnuMed and the paradigm of medical records you are attempting to 
> implement.
...
> I think your paradigm is wrong wrong wrong.
Richard, I am going to give you an account of how I
developed the UI parts for the document system. Please tell
me where my approach goes wrong (not the result, that might
well be wrong):

I think to myself: I want to handle documents. How would I
want to see documents listed ? How have I seen documents
listed in other products ? OK, a tree that can be sorted by
different criteria seems to serve well.

So I start wxGlade and click together a panel holding a
tree. I then add event handlers to the stubs wxGlade
provides. In those handlers I access the backend and pull
out the data I need.

I notice that I want to *display* documents from the tree.
So I make the "activate node" event handler use the
middleware to get the document and hand it over the
appropriate viewer. Works. Done.

Then I notice that I want to be able to add a review to a
document from within the tree. So I add code to the
right-click event handler to popup a menu and offer an item
"Review document". Then I go to wxGlade, click together a
dialog allowing me to edit the review and some other
metadata. Then I add save code for that to the new dialog's
event handler. Test. Works. Done.

Then I notice I would like to have the review popup after
displaying a document which has not been reviewed yet. So I
add appropriate code to the end of the display event handler
to show the already-known dialog.

So, Richard,

1) where does this approach go wrong fundamentally ?





2) where is the *result* as it is inefficient ?






Karsten
-- 
GPG key ID E4071346 @ wwwkeys.pgp.net
E167 67FD A291 2BEA 73BD  4537 78B9 A9F9 E407 1346