Re: [Gnumed-devel] Re: Medication viewing

[Jim Busser]

On 2009-11-19, at 5:11 AM, Karsten Hilbert wrote:

> > Can there be a Grouping = none and what would be the best resulting
> > sort... alpha on Stoff (and within Stoff by decreasing numeric value
> > of Dosis)
>
> :-))
>
> You are saying you want Grouping = None and then define a grouping ?
>
> (Note that we've been ungainly mixing grouping and sorting
> in the above discussion.)

I presume the above is ;-)

Grouping need not be a "special sort" in which – among items that are  
identical on an attribute – the sort function essentially does not  
apply or, if applied, makes no difference. It is also possible to  
group things based on a property that links indirectly to the one  
under consideration and may be a non-linear property.

How would rows within columns be understood to be anything other than  
"Sorted" since you presumably apply no random function to sequence  
them? You either fetch them "unsorted" by sequence of raw creation,  
which is essentially sorted by datetime, but as it is possible to  
backdate when the patient started on a medication, the underlying sort  
on creation will be inapparent.

Accordingly (apparently) "Unsorted" can mean sequenced on a random  
value, or just sorted on a property whose value is not displayed.

Ungrouped may therefore mean clustered according to a random property,  
or just grouped on a property whose value is not displayed.

If the list is displayed with its rows sequenced by substance name and  
within substance name by dosis and within dosis by plan what would you  
suggest is the grouping that should be offered? By substance?