Re: [Gnumed-devel] schema verification

[Jim Busser]

On 2011-10-25, at 2:50 AM, Karsten Hilbert wrote:

> I now better understand exactly what you were aiming at.
> 
> It's just not like that would be without drawbacks either.

By analogy, when one presents

        General case
        Specific case
        Attribute

in a fictitious example

        Hotel chain: Hilton Hotels
        Hotel: Hilton Honolulu
        Status: Bankrupt

to what does the status 'bankrupt' apply? Perhaps the english-language 
convention – wherein an attribute takes, as object, the last-previous-named – 
does not apply in other languages. But

        db      v15
        size    3.2 Gb
        hash foo

is certainly open to question as to what the hash pertains. Reversing the 
position of the 'size' line and the 'hash' line, and better labels, seemed to 
me a no-brainer but maybe my brain lacked something <grin>.

While there is a risk that others may consider this thread to be angels on the 
head of a pin, I predict that as soon as GNUmed would gain greater uptake, you 
are going to have people breaking their databases whether through fiddling or 
debugging or applying a local fix pending the release of one and then 
forgetting or losing track of the change that they had made.

If information that is being presented to degugger /troubleshooters can easily 
be improved (if it confused me, it could confuse others) then we at least 
consider to do so.

At the risk of diminishing value of pursuing the thread… given that I was 
saying nothing about the order (sequence) of column names returned by the 
function

        select gm.concat_table_structure_v3()

rather only to improve presentation of the information *above* that list, 
*what* drawback to such a change?