Re: [Gnumed-devel] Document part does not exist in the database

[Karsten Hilbert]

On Sun, Aug 31, 2008 at 08:19:09PM -0700, Jim Busser wrote:

>> Jim, do you think empty documents should be exempted from
>> generating pseudo-notifications as to their unsigned status ?
>
> I guess it depends on why anyone would add an empty document part to a 
> patient's record. If it is only now happening because people are playing 
> with the interface, and uploading whatever file(s) happen to be handy -- 
> even if these are empty -- then if the scenario would ever happen in a 
> production system, should it come to attention if only to educate the 
> user to stop adding empty parts?
>
> Can the detection and therefore review / signature of empty parts be a 
> warning sign of a mishandled document?
>
> Is there any use-case for deliberately adding an empty document part?

Maybe there is a misunderstanding here.

Empty documents (as in "pageless") were introduced to allow
people to "index" a treeware document store. That was your
suggestion.

Documents with pages but where the pages do not hold *data*
aren't useful IMO (unless someone argues for a finer-grained
version of the above). 

There currently is a not-NULL constraint in place on
blobs.doc_object.data. This does not protect against the
data being set to the empty string, however.

It may be more semantically valid to change that constraint
to data either being NULL (explicitely not available) OR of
non-zero length. What do you think ?

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