[Dcmlib] Checkin the 'Value Multiplicity'

[Mathieu Malaterre]

Jean-Pierre Roux wrote:
>   Hi, Mathieu, Benoit.
> 
> Something you probabely forgot, is the entry 'Type' :

Nope, see the comment I put in gdcm::DataEntry::IsValueCountValid.
This is definitely on my TODO list. We should not pretend being able to 
write DICOM file without support for this. Otherwise we'll have to 
support yet another broken DICOM file format: the one produce by gdcm. 
So as long as we are not able to check that, I would say that the file 
we write are invalid (which seems to be the case, since dciodvy also 
complains about gdcm file).

> 
>   // Entries whose type is 1 are mandatory, with a mandatory value
>    // Entries whose type is 1c are mandatory-inside-a-Sequence,      
> //                          with a mandatory value
>   // Entries whose type is 2 are mandatory, with an optional value     
> // Entries whose type is 2c are mandatory-inside-a-Sequence,
>   //                          with an optional value
>   // Entries whose type is 3 are optional
> 
> 
> It's *not* in the Dicom Dictionary;

This is again why I wanted a more XML like dictionary: so much more 
flexible (by design). And as a side note I am working on it, there is 
only one tag from ACR-NEMA that is causing trouble (where VM depend on 
VR...).

> For a  given Tag, it may change, depending on the Modality and/or 
> whether it's or not embedded inside a Sequence.
> (Have a look at the various conformance statements ...)
> 
> 
> Still wishing to check it at reading time?

Again it depends what you call read time. If we reread gdcm written 
file, then yes we should definitely check that the file written are 
valid, otherwise we'l just spread some more 'deceased' DICOM files.

2 cents,
Mathieu