[Dcmlib] TestVR back again

[Jean-Pierre Roux]

>
>
>>> Hello,
>>>
>>> I just modify the TestVR to fail on a very bizarre case. Looks like 
>>> the switch in gdcm::VR that read on the fly the dict is not working. 
>>> Whereas if you open gdcmDefaultDict.cxx in the gdcm-bin dir you'll 
>>> see that there is not '\000/' VR.
>>>
>>> Instead of fixing the parser how about we always use the other case: 
>>> FillDefault ? This is a very nice feature in gdcm to load on the fly 
>>> a new dicom dict but how about a new VR dict ? Does it really make 
>>> sense ? How often does it change ? Did it ever change ?
>>
VR list didn't change since *ages*§
-One of the few things permanent in Dicom-
Sure we'll never try to create VR on the fly !
When a image is supposed to be 'Explicit VR' -we  *know* some Entries 
are *not*-, the only way we have to guess if the length is coded on 2 or 
4 bytes is to check the 2 first bytes of the 'area' taht may contain, 
either [any VR but OB, OW, SQ + length, coded on 2 bytes] or [OB, OW or 
SQ + 0x0000] or [ length, coded on 4 bytes]

An pray -or throw curses on %$@ Dicom- we never get, in an illegal 
Implicit VR Entry, a length starting with 0x5352
(this entry will be considered as a Sequence :-(  )

JP

>>>
>>> Mathieu
>>>