Re: [Gnumed-devel] Re: Info re way data retrieved - 4

[Karsten Hilbert]

> Ok, checked it out again and sure enough the little box comes up - minus 
> names 
> in the case of hilbert, but with names in the case of leo.
Reason being that "hilbert"s aren't in v_basic_person. Reason
for this being that they don't have a NAMES row with
active=true. Reason for this being that we haven't yet solved
how to force unique-per-patient active=true rows in NAMES.

> The search is stil odd. In my database I have a number of persons with 
> christine names containing the letter 'i'. Just typing that and hitting enter 
> brings up nothing at all, the 'i' just sits there.
See above, problem with missing in v_basic_person due to
missing active=true in names.

> Now I have richard, ian, christine, hilmar, cillar sebastian, julian, which 
> should all appear on the list.
Correct.

> Think about the sequence. You are sitting in front of your computer. You have 
> decided to look for a patient, ie you have dismissed either in practice or in 
> front of you the last patient. If nothing else, removing the previous record 
> at the start of search or at the end of search if nothing is found at least 
> gives a visual signal to the user that something is happening.
Ah, *now* you are getting sensible.
You have "dismiss previous record":

- at start of search
- at end of search if nothing found

This can be collapsed into:

- at start of search

resulting in

- new patient if patient found
- no patient if nothing found

BUT NOT (as I thought you previously suggested):

- when *indicating likely desire to soon start search* (eg by
  clicking into the search field)

What you suggest now makes sense. Or perhaps you were
suggesting that all along and I wasn't getting it right.

Technically this is the same as:

- dismiss at end of search *regardless* of search result

This is still a change to the current design concept "always
have an active patient". If people really think that is worth
the increased trouble in coding they better speak up and/or
start coding.

> We must remember we are designing this sytem not for you or
> I but hopefully for many many thousands of users.
Hopefully so. However, even though you have shown to some docs
your way of dismissing patient records there still are a bunch
of commercial programs in widespread use here in Germany that
do adhere to "always have an active patient". They even go so
far as to load the last patient from the previous run or the
first patient in the database upon startup. I can tell you that
that does NOT confuse a few thousand German doctors. There's
evidence either way. In face of that evidence *I* would prefer
the option that's safer to code (since simpler).

> > > 2) Immediately show a busy cursor to indicate it is doing something
> By convention the busy cursor is the first event in any search. The routine 
> should be something like:
> 
> def Whatever:
>       wx.wxBeginBusyCursor()
>       try:
>               look for something. do your search
>       except:
>       wx.wxEnd BusyCursor()]
Changed code to likely match suggestion.

> (Hey, I'm leaning eh! I can argue the point with a bit of python code!!!!!)
:-)

> > > 3) If no records found  this should be handled
> > It is, a popup appears saying so. Except that the parent=NULL
> > bug prevented that from happening. Fixed.
> Ah, just tried this and noticed it. Tell me which file, which line and I'll 
> pretty it up,
client/wxpython/gmPatientSelector.py

but I am OK with that. The part that needs prettying is the
pick list on multiple patients.

> but note that if I type in terry - nothing apparen happens - no 
> busy cursor, no data, no indication anything has happened.
This is again due to the missing active=True in names hence no
v_basic_person rows.

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