* setup.py is again effective on un*x (takes into account the

[gdcm.git] / ChangeLog

2003-07-02  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * setup.py is again effective on un*x (takes into account the
        addon of the jpeg library)
      * Doc/DoxyDevelInstal.txt added (developper's raw documentation)
      * Doc/Doxy* updated (includes above enhancements)
      * INSTALL now points to web documentation

2003-07-01  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.h doesn't make unnecessary reference to
        gdcmUtil.h anymore.
      * gdcmPython/Makefile.am now avoids calling the wrappers for the
        clean target.
      * src/gdcm*.[cxx|h] :
        - to allow compilation with gcc 3.x
          -- clean up of STL usage (added explicit usage of std::)
          -- small fixes on exceptions.
          Note: linking not tested !
        - clean of doxygen comments to avoid warnings at documentation
          building stage.
      * vtk/vtkGdcmReader.cxx: g++ warning message clean up (platform
        dependent convertion).

2003-06-20  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Clean up of previous JPR garbage commit (not even commented in the
        Changelog):
        - Test/Makefile.am removed the reference to unexisting testTS.cxx
        - moved all jpeg related files from src to src/jpeg/libijg8, with
          addition/modification of Makefile.am and configure.in.
        - the testSuite is still brain damage (not guilty).

2003-06-02  Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * WrapVTK.py : modifications to get correctly vtk source files to 
        compile
      * vtk/vtkGdcmReader.cxx : modification for an already existing name
        (FileName -> Filename)

2003-06-11  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * vtk/vtkGdcmReader.[cxx/h]:
        - UpdateProgress method should now be effective
        - removed the limitation on having only images loaded (as opposed
          to volumes),
        - when building a stack of images/volume, a lesser number of
          gdcmHeader and gdcmFile instances should be used (hopefully
          accelerating things).
        - when building a stack of images/volumes, files which are not
          loadable (wrong path-filename, wrong permissions on existing file,
          or file not parsable by gdcm) are replaced with a black image
          in the stack (for the caller to notice the problem).

2003-06-03  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * vtk/vtkGdcmReader.cxx: hopefully corrected Z extent.

2003-06-02  Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * vtk/vtkGdcmReader[cxx|h] : makes correct code to remove conflicts 
        between standard library (std) and vtk library (problems are found
        under Windows... thanks Windows !)

2003-05-30  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * vtk/vtkGdcmReader[cxx|h] should now be volume aware (read ready for
        debug stage).
      * gdcmPython/demo/vtkGdcmReader.py: commented lines for volume test

2003-05-29  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * vtk/vtkGdcmReader[cxx|h] preparation addons for loading volumes.

2003-05-28  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * When compiling with distutils (see setup.py) the C++ compiler is
        called with different flags than the default ones (with automake).
        In order to fix all those warning, I compiled gdcm with the command
           make 'CXXFLAGS=-Wall -Wstrict-prototypes -D__STDC_LIMIT_MACROS'
        and went for warning fixes:
        - src/gdcmHeader.cxx, Test/dcm2acr.cxx, pourFindTaggs.cxx cleaned
          up from unused variables.
        - vtk/vtkGdcmReader.cxx: potential bug fix.
      * src/gdcmHeader.h: Doxygen warning cleanup

2003-05-27  Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * gdcmPython/gdcm.i:
        - typemaps correction to take into account the systematic replacement
          of "using namespace std" with std:: prefix e.g. std::list,
          std::string... (cf changes of 2003-05-21).
        - gdcmGlob global variable of type gdcmGlobal (defined in
          src/gdcmUtil.cxx and declared in gdcmPython/gdcm.i) is now
          exported to Python a as cvar.
      * gdcmPython/__init__.py now defines two functions GetPubDictTagNames()
        and GetPubDictTagNamesByCategory() as a replacement for deprecated
        gdcmDictSet.GetPubDictTagNames() and
        gdcmDictSet.GetPubDictTagNamesByCategory() class functions (i.e.
        C++ static methods).
      * gdcmPython/demo/printGroupedPublicDict.py is operational again,
        with the above changes.

2003-05-22  Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * setup.py, manifest.in : bug fix under linux

2003-05-22  Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * Modification of setup.py to compile vtk part too. Then, we have 2
        wrappers which must work with same distutils. For that, we have a
        generic distutils in distusiltsWrapping.py ; with :
         - build_extWrap class to wrap generically all extensions,
         - ExtensionWrap base class for all wrapping extension that contains
           a wrapper
         - Wrapper interface which wrap sources
        + In WrapSwig.py  we have extension and wrapper for Swig
        + In WrapVTK.py  we have extension and wrapper for VTK
      * MANIFEST.in : modifications to consider vtk directory and new python
        files for compilation

2003-05-21  Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with Benoit Regrain
      * Added python wrappers of vtkGdcmReader vtk class (see the
        source in vtk/vtkGdcmReader.cxx) :
        - vtk/Makefile.am now builds a library
        - gdcmPython/Makefile.am now builds vtkgdcmPython a second import
          python library (as opposed to _gdcm.so wich are the python
          wrappers of gdcm). vtkgdcmPython.so uses the vtk python wrappers
          (vtkWrapPython and vtkWrapPythonInit utility) to build vtkgdcmPython
          import library.
        - configure.in has an additional flag --enable-vtk that needs
          to be set to enable compilation of vtk related code, that is:
          + vtk/vtkGdcmReader.so and
          + vtk/testvtkGdcmReader C++ demo of vtk wrappers of gdcm)
          + gdcmPython/vtkgdcmPython.so (see above)
        - gdcmPython/demo/vtkGdcmDemo.py corrected (some images cannot
          be read when compressed or when HighBit + 1 != BitsStored),
        - gdcmPython/demo/vtkGdcmReader.py added. This demo illustrates
          the usage of the python wrapper of vtkGdcmReader vtk class.
      * vtk/vtkGdcmReader.cxx: bug fixed (thanks to Benoit Regrain).
      * src/*.[h] all occurences of stl classes are now prefixed with
        std::, and all occurences of "using namespace std;" where removed.
        This is to avoid pollution of global namespace in included files.
        Apparently vtk does not avoid this pitfall: when using both
        gdcm and vtk (as in vtk/vtkGdcmReader.cxx) this ended up in a
        collision of various stl members (principally cout...).
      * gdcmPython/testSuite.py now cleanly removes temprory generated
        file.
      * gdcmPython/gdcm.i now declares a typemap for std::string (to
         comply with above changes)

2003-05-12  Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with JPR
      * src/gdcmHeader>[h/cxx] added gdcmHeader::GetPixelSize()
      * vtk/vtkGdcmReader.cxx now properly inports the image in the
        vtk data structure (an image Flip was required).
      * vtk/testvtkGdcmReader.cxx refers to gdcmData subdir instead of Data.
      * cosmetic changes in documentation.

2003-05-7  Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with JPR
      * src/gdcmHeader.cxx: the constructor no longer exits when an
        unexisting file is given as argument.
      * The subdirectory Data (containing all the images used for the
        test suite) is not part of this repository anymore. A new module
        containing those images is now available at 
          :pserver:xxx@cvs.creatis.insa-lyon.fr:2402/cvs/public
        with the name gdcmData.
        All the python scripts (including the package initialisation file
        gdcmPython/__init__.py) were adapated to take this change into
        account (basically GDCM_DATA_PATH is now GDCM_TEST_DATA_PATH).

2003-05-5  Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with JPR
      * vtk subdir added. Contains vtkGdcmReader.[cxx|h] a vtk class
        inherinting from vtkImageReader and testvtkGdcmReader.cxx a small
        demo of the usage of this class.
        Compilation of this vtk part is only done when using the --enable-vtk
        at configure (or autogen.sh) stage.

2003-04-16  Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with JPR
      * More memmory link related corrections and documentation fixes.
        Notes on valgrind:
          - maximum info is obtained with a command of the form:
            valgrind --leak-check=yes --leak-resolution=high --num-callers=40
                   --show-reachable=yes PrintHeader
          - the remaining reachable blocks seem to come from the STL
            allocation scheme through the usage of map and list. It looks 
            like this memory cannot be freed but it is not a memory leak
            (in fact further invocation to the STL would recollect the
             unused memory allthough it cannot explicitely be freed).
      * gdcmPython/demo/vtkGdcmDemo.py added: this is a small demo
        of displaying an image parsed with gdcm and displayed with VTK.
        Note: some images don't seem to work e.g.
            python vtkGdcmDemo.py  ../../Data/US-RGB-8-esopecho.dcm
      * src/gdcmHeader.x: dicom_vr and Dicts are not class members anymore.
        Allthough this weakens the semantics, it is a ditch attempt to
        make gdcm more thread friendly.

2003-04-15  Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with JPR
      * Memory link hunt (by using valgrind through the command
          valgrind --show-reachable=yes --leak-check=yes PrintHeader).
        - added src/gdcmVR.cxx gdcmVR.h that objectify the previous
          gdcmHeader::_dicom_vr.
        - gdcmHeader::InitVRDict transfered as gdcmVR::gdcmVR().
        - gdcmHeader::dicom_vr is now of type gdcmVR* as opposed to
          VRHT*.
        - gdcmGlobal global object contained class added (see src/gdcmUtil.x)

2003-04-9  Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with JPR
      * src/Makefile.am now exports all the necessary include files at
        make install stage.

2003-04-9 JPR
      * UpdateGroupLength replaced by new one

2003-04-7 JPR
      * UpdateGroupLength re-written using H-Table
        (named UpdateGroupLengthNew untill checks are over)
      
2003-04-7 Eric Boix <Eric.Boix@creatis.insa-lyon.fr> with JPR
      * Data/im_00001 renamed to gdcm-MR-PHILIPS-16-Multi-Seq.dcm
      * gdcmPython/testSuite.py now include a test of gdcmFile::Write.
      * src:
        - gdcmHeader::GetPubElValSet removed.
        - gdcmElValSet::WriteDcm, WriteAcr, WriteExplVR, revamped to
          UpdateGroupLength, WriteElements, Write.
        - gdcmHeader::FileType moved to gdcmCommon.h. The enum FileType
          doesn't contain TrueDicom anymore since ExplicitVR and ImplicitVR
          carried the same semantics.
          - src/gdcmHeaderIdo.cxx changed accordingly.
        - gdcmFile::WriteBase now regroups all the codes from previous
          versions of WriteDcm, WriteDcmImplVR, and WriteACR.
        - enum FileType moved to gdcmCommon.h
      * src/gdcmHeader.cxx AddDefault trashed
      * gdcmGetXSize added
      * getimageDataSize now calls gdcmGetXSize
      * Test/*.cxx changed to agree with above changes

2003-03-31 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.h: LoadElements() is not a public method anymore
        (to avoid double call by end user).
      * Test/*.cxx LoadElements() doesn't need to be called explicitely
        by end user, since gdcmHeader::gdcmHeader now handles it.

2003-03-29 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * gdcm/Doc many doxygen changes:
        - Doxyfile now has two different versions, DoxyfileUsers that generates
          the html.users end users oriented documentation and
          DoxyfileDeveloppers that generates the html.developper developper
          oriented documentation.
        - README changed accordingly
        - requirement.txt removed (old design info moved to TODO).
        - Additional files DoxyInstallation.txt, DoxyIntroduction.txt,
          DoxyMainPage.txt, DoxyPython.txt provide more information.
        - Note: the website http://www.creatis.insa-lyon.fr/Public/Gdcm/ can now
          be updated by simply regenerating the docs with doxygen and copying 
          html.developper and html.users to tux:/home/httpd/html/Public/Gdcm
      * src/gdcmElValSet.cxx: stl <map>.count() can only return 0 or 1. Hence
        all the tests in the form "if (<map>.count() >1)" were removed.
      * src/gdcmFile.cxx: cosmetic changes to avoid messages of doxygen about
        ill-formed documentation sections.
      * src/gdcmHeader.cxx: ditto

2003-03-27 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * setup.py fixed for un*x (needed macro for stdint.h)
      * Test/test.cxx renamed to PrintHeader.cxx to avoid collision on
        un*ces with sh-utils test utility.
      * gdcmPython/demo/test.py renamed to PrintHeader.py
      * gdcmPython/demo/ReorganiseFiles.py new demo script added.
      * gdcmPython/win32/gdcmpycomplet.dsp renamed to gdcmpyembedded.dsp.
      * gdcmPython/win32/gdcmpy.dsp trashed away (brain damaged).
      * src/gdcm*.cxx JPR's bad inclusion fixed.

2003-03-26 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader[h, cxx] gdcmHeader now has an IsReadable predicate.
      * gdcmPython/demo/test.py changed accordingly.

2003-03-25 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * gdcmPython/Makefile.am fixed linking against libstdc++
      * gdcmPython/_gdcm.so symbolic link to gdcmPython/.libs/pygdcm.so added.
      * Swig subdir (historical tests of swig version 1.3.17++) removed
      * Test/testChangeEntete.cxx some lines commented out in order to make
        compilation work.
      * src/gdcmHeader.cxx cleaned up wild JPR's dirty kludge.
      * src/gdcmElValSet.cxx cleaned up frog's forgotten debug message

2003-03-24 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.cxx gdcmHeader::gdcmHeader now calls LoadElements(),
        hence the user doesn't need this explicit call.
        - gdcmPython/testSuite.py, gdcmPython/demo/*.py changed accordingly
          i.e. LoadElements() call removed.
      * src/gdcmCommon.h now defines UINT32_MAX (see stdint.h).
        - src/Makefile.am declares the proper flag for g++.
      * src/gdcmElValSet.h added gdcmElValSet::GenerateFreeTagKeyInGroup
        for adding user defined new tags.
        - src/gdcmHeader.cxx gdcmHeader::NewManualElValToPubDict added.
        - src/gdcmHeader.cxx gdcmHeader::AddAndDefaultElements now
          cleanly adds some gdcm added tags (like gdcmXSize, gdcmYSize...)

2003-03-17 Fabrice Bellet <Fabrice.Bellet@creatis.insa-lyon.fr>
      * Makefile.am, acinclude.m4, configure.in, python.m4, Dicts/Makefile.am,
   Doc/Makefile.am, Test/Makefile.am, gdcmPython/Makefile.am, 
   src/Makefile.am : the project should properly compile and install
   with the autotools, included the python wrappers part. 
      * gdcm.spec.in : added a spec file for RPM packaging.

2003-03-14 Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * src/gdcmUtils.cxx : redifinition of defaults parameters in code

2003-03-12 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * gdcmPython/__init__.py fix of GDCM_DICT_PATH which was not
        to exported to the environement (see gdcmPython/testSuite.py
        for usage).
      * src/gdcmElValSet.[cxx/h], gdcmElValue.[cxx/h], gdcmFile.[cxx/h],
        gdcmHeader.[cxx/h]: ElValSet class renamed to gdcmElValSet, and
        ElValue class renamed to gdcmElValue for obvious consistency reasons.
      * src/gdcmElValSet.cxx, src/gdcmUtil.[cxx-h] Tokenize utility function
        moved away from gdcmElValSet.cxx to gdcmUtil.cxx.

2003-03-10 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * gdcmPython/__init__.py doesn't crash anymore when running in
        in InstallMode or PreInstallMode but with an environement given
        value of GDCM_DICT_PATH.
      * src/gdcmDictSet.[cxx/h] coding style.

2003-03-06 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.h and gdcmHeader.cxx Coding style + doxigenation.
      * src/gdcm.h general comments moved to TODO, and README
      * src/gdcm.h should now be seen as a user commodity (including it
        should suffice to fully use gdcm). It is nowhere including within
        any of the kernel files src/*.cxx and src/*.h.

2003-03-05 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcm.h splitted in gdcmCommon.h, gdcmDict.h, gdcmDictEntry.h,
        gdcmDictSet.h, gdcmElValSet.h, gdcmElValue.h, gdcmFile.h,
        gdcmHeader.h
      * src/gdcm*.cxx only include their corresponding include file
        (as opposed to gdcm.h)
      * gdcmPython/gdcm.i changed accordingly
      * Test/Makefile.am corrected and added NEWS, AUTHORS in order for
        the autogen.sh generated "make snapshot" command to work.
      * autogen.sh removed history related references to crea package.

2003-03-04 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * distutilsSwigCPlusPlus.py work around some swig portability
        bug.

2003-03-03 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Eventually, python/gdcmPython was renamed to gdcmPython. This
        was done in a last ditch attempt to get setup.py _really_
        operationnal. The python/gdcmPython layaout add the advantage
        of hidding away the python related distutils specific files
        (MANIFEST.in, setup.py, distutils*.py...) from the main directory.
        Alas, the src directory was out of scope (i.e. it's relative
        position to setup.py was ../src) which made things impossible to
        handle (see previous version of python/setup.py). Crossing fingers...

2003-02-20 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * python/setup.py now works on Un*x. (harder than I thougth)
      * python/distutilsSwigCPlusPlus.py can now have include files
        among the list of sources.
      
2003-02-20 Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * src/gdcmHeader.cxx : bug fix - opening dicom file in binary !!!
      * modifications in MSVC projects

2003-02-19 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * As stated by the first lines of Test/ExceptionAndPython/README, it
        looks like we can move back to original usage of exception within
        src/gdcm* and hence remove the errno stuff that was introduced
        as a quick and dirty fixture of this problem (added to the TODO).
      * Doc/Doxyfile configuration file for Doxygen and corresponding
        README file for usage added.
      * Test/testWrite.cxx added. This new test comes from a split of
        test.cxx that is now reverted to its original purpous i.e. 
        testing the proper parsing of a Dicom related file.
        testWrite.cxx concentrates on testing the IO part of gdcm.
      * Test/bug1.cxx (containing a bug on WIn32) added.
      * src/gdcm.h, gdcmHeader.cxx, gdcmDictSet.cxx: 
        - gdcmHeader::GetPubTagNames and  gdcmHeader::GetPubTagNamesByCategory
          whose purpose is to publish the content of the TagNames of the
          Dicom public dictionnary were not accessible without an
          instance of class gdcmHeader.
        - those methods are now static methods of gdcmDictSet and hence
          require no instances at all to be invocated.
        - within gdcmDictSet this change required to change some method
          to class methods (i.e. are now static) among which SetDictPath
          (renamed to BuildDictPath) and LoadDefaultPubDict.
      * python/gdcmPython/demo/printGroupedPublicDict.py changed to illustrate
        the above changes by calling the new method classes of gdcmDictSet.
      * python/gdcmPython/__init__.py now exposes gdcm.gdcmDictSet for
        the above to be effective.
      * python/gdcmPython: in order to wrap properly the above changes
        for Python, swig version now needs to be > 1.3.17.
      * python/gdcmPython/Makefile fixed to adapt itself to new 
        configure/make shema introduced by Johan Montagnat (thanks for the
        contribution). Alas this Makefile cannot be turned into a proper
        Makefile.am without some heavy changes in the configure.in
        (for python dectection).
      * python/gdcmPython/gdcm.i: the out typemap map<string, list<string>>*
        now avoids publishing the empty entries.

2003-02-13 Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * python/setup.py : bug fix concerning install under windows
        ExtraPath reused :-P
      * gdcm.dsw : bug fix concerning path for libraries for python !!!

2003-02-13 Benoit Regrain <Benoit.Regrain@creatis.insa-lyon.fr>
      * Dicts/dicomV3.dic : remove double spaces !!!
      * src/gdcmElValSet.cxx : bug fix concerning windows compiler
      * python/gdcmPython/win32/gdcmpycomplet.dsp : bug fix concerning paths
        and links

2003-01-28 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.cxx added a post header parsing AddAndDefaultElements
        method. Doxygenation.
      * src/gdcm.h clean up of JPR spurious comments.

2003-01-17 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * python/distutilsSwigCPlusPlus.py now properly collects the
        shadow classes generated by "swig -c++" (gdcm.py in our case)
        when using "python setup.py install".
      * python/gdcmPython/__init__.py imports gdcm.py and only manually
        reexports the working classes.
      * src/gdcmHeader.cxx all the try/catch/throw calls were replaced
        by the classical C errno scheme. This is to prevent an obscure
        behavior of the python wrappers when importing wxPython.wx prior
        to gdcmPython (which ended up in an abort call). An illustration
        of this oddity can be found in the Test/ExceptionAndPython
        subdir (see the README file). This problem probably due to
        an combination of g++ and dynamic loading.
      * added Test/ExceptionAndPython (see above)

2003-01-17 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Changed the layout of the python part to avoid bloating main
        directory with setup.py, distutilsSwigCPlusPlus.py, MANIFEST.in
        i.e. the distutils arsenal. All the python related stuff is
        again in a python subdir, but the package itself is now in
        python/gdcmPython.
      * setup.py was cleaned up:
        - pythonIncludePath removed
        - python setup.py bdist target is now functional.
      * gdcmPython/__init__.py doesn't export FileName any more (to avoid
        collision with other packages). gdcmPython/demo/*.py changed
        accordingly.
2003-01-15 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * python subdir moved to gdcmPython (preparation of distutils packaging).
      * gdcmPython/setup.py and distutilsSwigCPlusPlus.py added. The
        distutils installer is operational.
      * - gdcmPython/__init__.py now properly loads the swig generated
          shadow classes (from gdcm.py).
        - gdcmPython/demo/*.py changed to import the package gdcmPython
          instead of gdcmPython/demo/load.py.
        - gdcmPython/testSuite.py changed to import the package gdcmPython.
        

2002-12-16 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcm.h, src/gdcmHeader.cxx: added GetPubTagNames() resp.
        GetPubTagNamesByCategory() to gdcmHeader that return a list of the
        entries within the associated public Dicom dictionary resp. the same
        information but sorted by the fourth field (PAT, IMG, DIR) of the
        dictionary.
        - Dicts/dicomV3.dic Entries which had and unspecified fourth field
          are now in the "???" group.
        - python/gdcm.i changed accordingly,
        - python/demo/printGroupedPublicDict.py added, that gives an example
          of the above new functionalities in Python.

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

2002-11-27 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * python/demo/*.py load.py extracted from test.py. Added explore.py
        that only displays required tags and testAll.py that parses all
        the files of the testsuite without using unittest.
      * python/testSuite.py other additional test on new files added.
      * Data/* new test files added accordingly to the testSuite.
      * src/gdcmHeader.cxx avoid overwriting of the dictionary when
        the vr in the file differs from the one of the dictionary.

2002-11-20 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcm.h and gdcmHeader.cxx gdcm no longer loads all the elements
        values, but limits itself to values which are below a user specified
        size (by default 1024 bytes).
      * python/gdcm.i : elements not loaded (because their size is above the
        user specified treshold) are exported with their TagKey as key of the
        python dictionary. This enables the testsuite to make some checks
        on the unloaded elements (since we have their offset and size).
      * python/testSuite.py
        - now tests for the existence and proper value of the pixel data
        - two new examples added.
      * Data/gdcm-MR-SIEMENS-16.acr[12] added.

2002-11-18 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcm.h and gdcmHeader.cxx are now "Big Endian transfer syntax"
        aware. See the strategy comments in gdcmHeader::FindeLength().
      * Test/test.cxx now accepts a filename as first argument.
      * Data/US-PAL-8-10x-echo.dcm and US-RGB-8-epicard.dcm added.
      * python/testSuite.py changed to integrate test on above files.

2002-11-15 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Dicts/dicomV3.dic removed all the trailing (RET) when the
        the vr was also RET.
      * python/testSuite.py more tests added.

2002-11-14 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * python/testSuite.py unittest test suite added (uses Data)
      * Data/* dcm and acr files for the test suite.
      * glib dependance removed from un*x version. We now use ISO C99
        7.18 Integer types (see stdint.h)
        - python/Makefile, src/Makefile src/gdcm.h
      * src/ when an explicit vr (like dicom files) suddenly poped an
        implicit element we use to mark the underlying DictEntry of the
        concerned ElValue with "Implicit". This strategy was damageable
        too the public or provite dictionaries, since some of their vr
        would be changed. Since Header::FindLength heavily relies on the
        detection by Header::FindVR of falsely explicit elements, I
        added an extra ImplicitVr member to ElValue (boolean).
        The rest of the changes we the necessary adaptations.

2002-11-12 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader::FindLength bug fix when trapping falsely explicit
        VR files.
      * src/gdcmHeader::FindVR: hard way fix of falsely explicit vr files.

2002-11-8 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Adaptation for porting to VC++:
        - src/gdcm.h
          o forced to use std namespace (for string to be known)
          o all class use __declspec export style on WIN32
        - src/gdcmUtil.cxx new Exit method that wraps the exit call
          (in stdlib.h on Win32 but in std:: for gcc)
      * src/gdcmDictSet::SetDictPath adds a trailing / to environement
        variable GDCM_DICT_PATH.
      * src/gdcmHeader.cxx verbose comments made discrete.

2002-11-7 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * python/gdcm.i: added a typemap that converts a C++ hashing table
        to native Python dictionary.
      * python/demo/test.py:
        - now uses the native dictionary for exploration of gdcmHeader.
        - takes an optional filename argument (the file to parse).
      * src/gdcm.h and gdcmHeader.cxx: gdcmHeader now has an accessor on
        PubElVals hashing table. 
      * Dicts/dicomV3.dic removed error prone trailing spaces.

2002-11-6 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Python wrapping process moved away from src/Makefile to
        newly created python/Makefile (as well as gdcm.i)
      * python/demo/test.py (that mirrors Test/test.cxx) is effective.
      * src/gdcmHeader::FindLength only looks for current vr when necessary.
      * src/gdcmDictSet.cxx: the dictionnaries directory path is now imported
        from the environement variable GDCM_DICT_PATH (when existing).
      * src/gdcmDict::GetTag bug fix.

2002-10-31 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Straightforward temporary fixes for swig to build the python wrappers.
        src/Makefile now has a python working entry [by working we mean
        that we can import de shadow classes without errors].

2002-10-29 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * hashtest.cxx removed (since allready in Test)
      * src/gdcmlib.h renamed to src/gdcm.h
      * src/dcm.i renamed to src/gdcm.i
      * src/Makefile prepared for python wrapping

2002-10-29 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.cxx :
        - Parsing of header is now separated from loading of element values
          (see gdcmHeader::LoadElements).
        - general clean up of hopefully all file access (creation of
          ReadInt16 and ReadInt32)

2002-10-22 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.cxx :
        - RecupLgr renamed to FindLength and cut off with new IsAnInteger
          method.
        - SetLgrElem renamed to SetLength
        - GetLgrElem renamed to GetLength
        - ~gdcmHeader() made virtual to pesky warning messages at compile.
      * src/gdcmElValSet.cxx fixed both GetElValue methods
      * Dicts/dicomV3.dic falacious entry [7fe0 0010 OT PXL Pixel Data]
        due to IdoDude was cleaned out.

2002-10-21 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.cxx RecupLgr split in FindVR and RecupLgr. FindVR
        concentrates on finding the value representation (when it exists).
        RecupLgr now only finds the element value length, and it avoids
        returning the unecessary skL (skipped length).
        Also, offsetCourant was simply removed (with taille_fich) since
        it was redundant with the File position internals (by using
        ftell and fseek). This also made skL (see above in RecupLgr) usage
        jajun.
      * src/gdcmHeaderIdo.cxx src/gdcmIdo.h added. Those files concentrate
        on the libido specificities which are handled through class 
        inheritance.
      * Test/implicit_vr.dcm dicom v3 in implicit value representation
        added. 

2002-10-17 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * Truckload of changes. Parsing of header is barely functional
        (see Test/test) with the standard dictionary (see Dicts/dicomV3.dic).

2002-10-10 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader.cxx many addons of low level methods

2002-10-07 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * src/gdcmHeader now contains_IdDcmCheckSwap, _IdDcmRecupLgr,
        and _IdDcmSWAP_LONG.

2002-09-12 Eric Boix <Eric.Boix@creatis.insa-lyon.fr>
      * This corresponds to all the changes decided at the meeting of
        May 23 2002. Needless to say we are running late...