NEWS

- Minor optimizations to the core mapping code, as well as the Simple
  Cylindrical and Orthographic map projections were made.

- MaRC now attempts to automatically populate additional standard FITS
  keywords in the generated map FITS file based on those provided by
  source images, as well as some generated by MaRC (e.g. DATAMIN and
  DATAMAX) during map creation.

- MaRC now requires a compiler that supports the C++17 standard.

1 February 2019 - MaRC 0.10

- Fixed a regression in quadratic roots calculations that caused some
  areas of an Orthographic projection to incorrectly have data of
  value zero, as well as other coordinate calculations.

- Blank areas of integer typed maps now contain the user-supplied
  DATA_BLANK value, if specified in the MaRC input file.  This allows
  FITS image readers to easily identify areas of integer typed maps
  that lack data.

- Corrected a problem where MaRC's Simple Cylindrical projection could
  not handle the case where lower and upper longitudes in the map were
  the same, signifying a 360 degree range, resulting in only single
  longitude being mapped across all map pixels.  MaRC now assumes the
  full 360 degree longitude range is being mapped if both the lower
  and upper longitude are the same, and issues a diagnostic message
  accordingly.

- The MU plane no longer needs to have the sub-solar point specified.
  It was never needed for emission angle calculations.

- The MU0 plane no longer needs to have the sub-observer point and
  range specified.  They were never needed for incidence angle
  computations.

- Improved MaRC input file parser error reporting.  The filename, line
  number and column containing the syntax error are now reported to
  the user.

- Corrected handling of position angles explicitly marked as clockwise,
  i.e. "CW", in MaRC input files.  They were erroneously treated as
  counter-clockwise position angles.

- The user-specific MaRC configuration file is now expected to be
  `$XDG_CONFIG_HOME/marc', which is `~/.config/marc' by default, to
  conform to the XDG Base Directory Specification.  The old MaRC
  configuration filename `~/.marc' is no longer supported.

- Added command line option support.  Try "marc --help".

- Improved logging by leveraging the fmt library as the underlying
  logging framework.  This allows for faster logging, and log level
  differentiated output.

- Corrected progress output for the Orthographic map projection so
  that completion percentage is printed up to 100%.

- MERCATOR map entries in MaRC input files may now use options common
  to all MaRC map projections.

- Fixed a mosaic averaging regression that caused weighted and
  unweighted averaging to be disabled if either was selected.

- Added support for generation of code coverage results for g++
  builds.  Try the following to generate the results:
        ./configure --enable-code-coverage
        make check-code-coverage

- Replaced all uses of the MaRC-specific "bodycentric" and
  "bodygraphic" latitude terminology with the NASA NAIF recommended
  "planetocentric" and "planetographic", respectively.

31 January 2018 - MaRC 0.9.10

- Reduced footprint of the MaRC program (`marc') and library
  (`libMaRC').  For example, including both the MaRC library and
  program, the static footprint was reduced by 28% for a 64-bit Linux
  build with g++ 7.2 when comparing total sizes returned by the `size'
  command.

- Data scaling values, i.e. BSCALE and BZERO, as well as the physical
  data unit (BUNIT) are now automatically written to the map FITS
  header when possible.  This allows FITS reading and viewing software
  to automatically provide the actual physical value to the user,
  instead of forcing the user to manually scale the map data.
  Currently this only applies to single plane maps containing one of
  the MaRC virtual images (MU, MU0, PHASE, LATITUDE or LONGITUDE).

  Maps with multiple planes currently write the scaling values as
  free-form text across several HISTORY cards in the FITS header
  instead of standalone BSCALE, BZERO and BUNIT cards.  In this case,
  data scaling of map values to physical values must generally be
  performed manually.

- Corrected automatic offset calculation for LONGITUDE map planes when
  creating integer typed maps.  Longitude values closer to 360 degrees
  are no longer cutoff.  Be aware that an offset must be applied to
  convert from the map longitude value to the physical longitude
  value.  See the generated map FITS header to obtain the offset
  (BZERO) value.

- Further expanded the MaRC test suite (try "make check").

- The MU, MU0 and PHASE virtual images may now contain all cosine
  values i.e. the interval [-1, 1], and are no longer limited to
  positive values alone.  There was no real need to drop the negative
  values.  To obtain the old behavior where only cosine values [0, 1]
  were mapped set the DATA_MIN keyword for the MU, MU0 and PHASE map
  plane entries in your MaRC input file to zero, e.g.:

      PLANE:
              DATA_MIN: 0
              MU: ...

- The PHASE virtual image no longer ties the existence of a datum to a
  positive cosine of the emission of the angle (mu), meaning cosine of
  phase angle data will also be plotted in areas of the map where
  mu < 0.

- Improved speed of MU, MU0 and PHASE virtual image generation by up
  to 100% by dropping a redundant data "visibility" check.

- Slightly improved overall MaRC performance by leveraging link-time
  optimization on some platforms.

- Corrected problem in conformal (Mercator and Polar Stereographic)
  maps that prevented calculation of latitudes for the entire map from
  completing in some cases.

- Dropped references to "transverse" in documentation related to the
  Mercator projection.  MaRC's Mercator projection is the regular
  variant not the transverse one.

- Improved accuracy of conformal maps through root finding
  improvements.

- Fixed memory violations in conformal maps.

20 August 2017 - MaRC 0.9.9

- Data generated by the MU, MU0, PHASE, LATITUDE and LONGITUDE virtual
  images are now automatically offset and scaled by the largest power
  of 10 suitable for maximizing the number of significant digits retained
  when storing their floating point values in integer typed maps.  For
  example, MU image data, which has a range of [-1, 1], will be scaled
  and offset to a range of [0, 200] for a map with BYTE (unsigned 8
  bit integer) data, [-10000, 10000] for a SHORT (signed 16 bit
  integer) data type map, and [-1e+09, 1e+09] for LONG (signed 32 bit
  integer) typed maps.

  The virtual image data is only automatically scaled and/or offset
  for integer typed maps.  They are not scaled or offset when creating
  maps with a floating point data type (DATA_TYPE of FLOAT or
  DOUBLE).

  The necessary scale and offset needed to obtain the true values of
  these virtual images in integer typed maps is now automatically
  written to the map FITS header as well.

  NOTE: The hardcoded scale value of 10000 for the MU, MU0 and PHASE
        virtual images that existed in previous versions of MaRC has
        been removed for all map data types.  Due to the automatic
        scaling that now occurs, the results will be the same for
        SHORT map types, but not the other types.  As described above,
        a better scaling factor and offset will be chosen.

- Corrected problem where flat-field correction would fail.

- Improved accuracy of vector magnitude calculations by leveraging an
  implementation that avoids floating point underflow and overflow, as
  well as having extremely low floating point error.

- Improved accuracy of quadratic equation root calculations by using
  an implementation that avoids loss of significant digits
  (i.e. catastrophic cancellation).  This impacts the sky removal
  feature and the Orthographic map projection.

- LONG map types are now correctly written to the map FITS file with
  32 bit signed integer data on all platforms, including those where
  the fundamental C language long integer type is 8 bytes wide.

- Improved memory management.
  - Better C++ exception safety.
  - Reduced run-time memory requirements.
  - Slightly improved performance due to less deep copies of large
    arrays.

- Fixed inversion of source photo images from left to right and top to
  bottom.

- Support for maps containing 64 bit signed integer data has been
  added.  Set DATA_TYPE in your MaRC input file to LONGLONG to use
  this feature.

- Fixed problem where negative data was not plotted in maps with a
  floating point data type.

- Initial support for Doxygen based MaRC source code documentation
  generation is now available.  Assuming Doxygen is installed, run
  `make doxygen-doc' from the top-level MaRC source directory to
  generate MaRC library and program source code documentation.
  Generated documentation will reside in the doc/doxygen directory.

24 July 2017 - MaRC 0.9.8

- MaRC now requires C++14 conforming compilers.  The 'configure'
  script will attempt to detect and enable C++14 support in the C++
  compiler in use. Verified to build with g++ 6.3 and Clang++ 3.9.
- Modernized the build infrastructure to leverage the latest versions
  of the GNU autotools (Autoconf, Automake and Libtool).  This should
  provide better build support on more platforms, particularly recent
  ones.
- MaRC now passes the static analysis scan run through Coverity (see
  https://scan.coverity.com/projects/ossama-othman-marc).
- Added better support for automated detection of CPU tuning compiler
  flags.
- Updated LPGL license terms from 2.0 to 2.1.
- Updated GFDL license terms from 1.2 to 1.3.
- Minor optimizations.
- Slight reduction in static footprint.
- Fixed a MaRC::Vector iteration problem that prevented properly
  mapping images where the latitude and longitude at the optical axis
  ({LAT,LON}_AT_CENTER) instead of the of body center
  ({SAMPLE,LINE}_CENTER were given.  Support for the LAT_AT_CENTER and
  LON_AT_CENTER MaRC input file keywords is still disabled, however,
  since the fix is still somewhat experimental.

11 November 2005 - MaRC 0.9.7

- Fixed latitude/longitude visibility calculation so that it correctly
  takes into account an optical axis that is not in line with the
  sub-observation point.

9 November 2005 - MaRC 0.9.6

- Reduced floating point computation error when image focal length and
  scale are available.
- Fixed logic error that prevented images from being mapped
  correctly.

8 February 2005 - MaRC 0.9.5

- Fixed map corruption and redundant mapping problems that occurred
  when specifying more than one MaRC input file on the command line to
  parse.
- MaRC input file PDF documentation now contains a section menu and
  hyperlinks with the document text.

8 November 2004 - MaRC 0.9.4

- Removed dependency on FlexLexer.h when building MaRC.
- Added several regression tests.  Try "make check".

3 November 2004 - MaRC 0.9.3

- Enhanced C library NaN constant detection to support additional ways
  of retrieving the NaN constant.  Addresses build errors on Solaris.

2 November 2004 - MaRC 0.9.2

- Added support for Transverse Mercator projection.  It is a conformal
  projection, i.e. it preserves shape.
- Added support for Polar Stereographic projects.  It too is a
  conformal projection.
- Fixed configuration problem on Solaris.

13 October 2004 - MaRC 0.9.1

- Fixed mosaic averaging code regression so that seams in image
  overlap regions are once again barely visible when averaging images
  taken by the same instrument at the same settings (e.g. wavelength).
- Slight speed improvement due to replacement of old TNT matrix code
  with MaRC-specific simple and fast matrix/vector implementation.
- Fixed a memory leak and a few potential ones.
- Addressed problem where longitudes in a "longitude map" were not
  wrapped into a 360 degree range.
- Added support for GNU g++ 3.4.2.

6 July 2004 - MaRC 0.9

- Major code rewrite.  Much cleaner and better designed and organized.
- MaRC now requires modern compilers that support most ANSI/ISO C++
  features.  Currently tested with GNU g++ 2.95, 3.2.2 and 3.4.0 on
  Linux.
- Added two new types of map planes "LATITUDE" and "LONGITUDE", which
  contain maps of the latitudes and longitudes in degrees,
  respectively.
- Many areas of MaRC have been further optimized.
- Geometric and photometric correction techniques are now "pluggable."
- CFITSIO is the underlying FITS library in MaRC, replacing AIPS++.
  This provides MaRC with all of the FITS file benefits provided by
  CFITSIO, such as on-the-fly FITS file compression/decompression.
- Added support for adding "AUTHOR", "ORIGIN" and "BLANK" FITS
  keywords to map FITS file.
- Sky removal feature no longer modifies source image, reducing the
  chances of losing valid data when mapping.
- Disabled all map projections except for Orthographic and Simple
  Cylindrical.  The others have not yet been ported to the new MaRC
  library API.
- MaRC is now distributed under the terms set forth in the GNU General
  Public License (GPL) and the GNU Lesser General Public License (LGPL).

18 November 1997

- Updated coordinate transformation matrix computations.
  Transformation should now be correct.  The update allows MaRC to
  take into account the fact the camera may not be looking directly at
  the sub-observation point.
- Added a perspective projection.  (STILL EXPERIMENTAL)
- Added code to use weighted averages when mapping in overlap regions.
  Seams are virtually invisible when mapping source images that were
  taken with the same instrument.
- Many code fixes and clean ups.
- Added some support for reading VICAR files.

30 May 1997

- Memory is now handled very differently.  The new memory scheme uses
  MUCH less memory for multi-plane maps than previously.  Memory
  requirements depend ONLY on the number of samples and lines in the
  desired map.  This means that a map with 20 planes would take up the
  same amount of memory as a 1 plane map during mapping!  Of course,
  additional memory is still required to handle the additional images
  in the image queue, but this memory requirement is insignificant
  compared to the memory savings of the new memory usage scheme.
- Negative values may now be mapped on "SHORT" (16-bit) integer maps.