[Gnumed-devel] (no subject)

[syan tan]

Ian Haywood wrote:

Message: 2 Date: Sun, 5 Oct 2003 08:17:31 +1000 From: address@hidden 
(Ian Haywood) Subject: Re: [Gnumed-devel] test-area/sjtan To: 
address@hidden Message-ID: <address@hidden> 
Content-Type: text/plain; charset=us-ascii On Sat, Oct 04, 2003 at 
04:55:19PM +0200, Karsten Hilbert wrote:

> > Also consider this complex example (which, however, does not
>  
[snip]

> > use of the function-hook API in those cases that warrant that.
>  
Fair enough, we *do* need MatchProvider subclasses for this sort of
thing instead of the GuessXXX functions I was writing, to provide
extra hooks for inserting new entries, learning and so on.

How should match providers know the context of their matching?
(for street names, we need to know the postcode, urb, etc.) 

One solution is to provide a dictionary of <widget name>:<widget contents>
problem is, we need to know the context when searching, not at widget setup.
So the dictionary could be an object implementing __getitem__, so we
always get the "current" values of widgets.

Ian


-----------
The dumb machine rendering in a testcase gmEditArea3 
is that one has to select the suburb first before 
giving the number and the street , othewise if the user selects the suburb
the street gets rubbed out, and if the street happens to be the same text
as a street in another suburb, triggering a selection event on it rubs out
the suburb !! Another score for logic -> user unfriendliness !