+2002-12-11 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
+ * src/gdcm.h, gdcmHeader.cxx:
+ - historic references to glib's g_malloc and g_free (#defined)
+ were definitively removed.
+ - gdcm.h: cosmetic changes (part of comments moved to Doc/requirements)
+ * src/gdcmElValSet.cxx:
+ - GetElement(guint32, guint32) renamed to GetElementByNumber.
+ - GetElValue(guint32, guint32) renamed to GetElValueByNumber.
+ - GetElValue(string) renamed to GetElValueByName.
+ - Added GetElementByName(string).
+ * src/gdcmHeader.cxx: added
+ - GetPubElValRepByNumber(guint16, guint16)
+ - GetPubElValRepByName(string)
+ - GetShaElValRepByNumber(guint16, guint16)
+ - GetShaElValRepByName(string)
+ - GetShaElValByNumber(guint16, guint16)
+ - GetShaElValRepByName(string)
+ - GetElValRepByNumber(guint16, guint16)
+ - GetElValRepByName(string)
+ - GetElValByNumber(guint16, guint16)
+ - GetElValRepByName(string)
+ * Doc/requirements.txt added.
+
+2002-12-9 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
+ * Test/Makefile building now depends on the one of libgdcm.so
+ * src/gdcmHeader.cxx and gdcm.h are now OB (undefined length encoded
+ pixel data) aware which enables finding the address (offset) of
+ the pixel data of JPEG encoded DICOM files. This leaves only a single
+ file in the testSuite whose pixel data address (offset) is unknown.
+ * python/testSuite.py changed accordingly.
+
+2002-12-6 Christophe Odet + Hugues Benoit-Cattin + Eric.Boix
+ * VC++ has some strong limitations when working with the STL, as stated
+ in http://support.microsoft.com/support/kb/articles/Q168/9/58.ASP :
+ "Also note that some STL containers (map, set, queue, list, deque)
+ cannot be exported. [...]
+ Some STL classes contain nested classes. These classes can not
+ be exported. [...]
+ This is caused by a designed limitation that once a template
+ class is instantiated, it can not be re-instantiated and
+ exported."
+ Since our usage of map<> is ubiquitous in gdcm, this "designed
+ limitation" of VC++ is a pitfall.
+ Hence the Python wrappers of gdcm cannot be incrementally linked
+ against the c++ dynamic library. The dirty but only workaround is
+ to forget about incremental link of dynamic libraries and to generate
+ the Python wrappers library with the inclusions of the underlying C++
+ library.
+ The following modifications concern this matter on Win32/VC++:
+ - wrapping python correct with standalone wrapped dll (don't use separate
+ dll under windows !!!!)
+ - python21_d debug mode enabled (ask Frog how to use it :-)
+ - NO problem with having an STL member of class for example string in C++
+ WITH THE RESTRICTION OF FORGETING ABOUT INCREMENTAL LINK.
+ - Python test of dcmlib in Python is ok under windows on a large set
+ (one) of image(s).
+ * removed glib references
+ * typedef's inserted in gdcm.i for correct swig type management
+