v2.2.4:

- EDPS workflow and QCReport plots have been updated

v2.2.3:

- EDPS workflows have been updated

v2.2.1:

- EDPS workflows have been added

v2.2.0:

- EDPS workflows and adari reports have been added

v2.0.2:

- An error in the calculation of the Fourier transform of the fringe
  patterns has been fixed. Specifically, the algorithm that rolls the
  fringe pattern in the spatial direction by half the image width was
  off by 1 pixels for half the image when the image size is odd. This
  has been corrected.

v2.0.1:

- In Paranal the processing on the pipeline machine gives different
  results than the QC1 system.  These two needs to be aligned to be
  able to use the same tools and threshold quantities to assess
  automatically the quality of the data. Data have been reprocessed
  and for the Paranal rules the pipeline parameter spectralBinning
  should be 7 for Nband data and should be 5 for LM band data. This
  has been changed in the oca rule file used by the Paranal DFS
  
v2.0.0:

- As of Matisse pipeline version 1.5.2, the convention for phases and
  closure phases for the N band was inconsistent with the convention
  used for the L and M bands. This caused a 180 degree rotation of the
  N-band images and models relative to the North-up and
  East-to-the-left standard. This has now been corrected beginning
  with the current pipeline version 2.0.0.
  
v1.7.8:

- A problem/segmentation-violation in the computation of the errors
  has been fixed. If only one frame is left after flagging, the error
  is set to be FLT_MAX, as the bootstrap routine can not be used in
  this case to properly give an error estimation
  
v1.7.7:

- The pipeline and workflows have been updated to work with the new
  cpl version 7.3 and newer python versions, respectively
  
- Additional Quality control parameters were added to the pipeline.

- The pipeline code and the workflow plotting routines have been
  updated to avoid problems with NaN pixels.
  

v1.7.6:

- The software to download the latest version of the Earth Orientation
  Parameter (see
  https://www.iers.org/IERS/EN/Science/EarthRotation/EOP.html) from
  IERS (see https://www.iers.org/IERS/EN/Home/home_node.html) has been
  changed. This was necessary as not only the link but also the
  download protocol changed from ftp to https

- The BCD calibration has been added to the mat_cal_oifits recipe. It
  is activated if the recipe parameter cumulBlock is set to TRUE
  (current default value)

- The matisse_viscal workflow and the launched mat_cal_oifits recipe
  has been changed:
  + A bug in the file naming scheme of the recipe has been fixed. The
    bug caused an overwriting of files in some circumstances.
  + The maximum associated night time calibration files have been
    increased from 10 to 12 in the workflow.
  + The recipe parameter --cumulBlock has been set to TRUE in the
    recipe as well as in the workflow. If several BCD combinations are
    included in the input files, the --cumulBlock=TRUE triggers
    additional outputs where the BCD signature in the files has been
    removed. These files can be identified by the header keyword
    DRS.BCD.COMBINED set to T (TRUE).

v1.7.5:

- Small change for SHITF_MAP in LOW L
- Adding a new processing dedicated to GRA4MAT Observations (K4N
  coherent processing).
- Adding KAPPA_MATRIX check in mat_ext_beams
- Adding QC parameter in various recipes

v1.6.0 (2021-05-20):

- Optimization of the Distorsion Map for the MED and HIGH resolution in L band.

v1.5.8 (2021-03-01):

    - Correct handling of the keyword PRO.DATANCOM
    - The recipes mat_wave_cal is now used to update the SHIFT_MAP in the VERY HIGH resolution mode.
    - The recipes mat_cal_oifits includes now the calibration of the BCD exposures
    - Flagging of the MATISSE data with the GRAVITY data in the case of observation with GRA4MAT. 

v1.5.6 (2020-11-30):

- Fixing a header inheritance problem
- Fixing a bug in mat_merge_results affecting the esoreflex workflow

v1.5.5 (2020-11-23):

- Fixing a problem related to a wrong keyword PRO.DATANCOM

v1.5.4 (2020-11-16):

- Adding parallel section by using OpenMP on linux platforms

v1.5.3 (2020-11-06):

- Adding parallel section by using OpenMP on linux platforms
- Fix of an out of boundary access

v1.5.2 (2020-06-04):

- A wrong wavelenth range when computing the following QC parameters
  on the dispersed fringes was fixed:
  
    QC.DETi.TFSQRj.MEAN
    QC.DETi.VISSQRj
    QC.DETi.VISSQRj.SNR
    QC.DETi.CPHASEj
    QC.DETi.CPHASEj.STDEV
    QC.DETi.DPHASEj
    QC.DETi.DPHASEj.STDEV

v1.5.1 (2020-05-06):

- File for the public release updated
- Default value for --excess_count_n and --excess_count_lm (plugin mat_raw_estimates) corrected, option --taryp works now for this plugin too.
- Estimating a possible TARTYP shift uses now the whole interferometric channel. It is possible to change TARTYP U with S/T even if the estimated or specified shift is 0.
- TARTYP=57 is now the default value whatever the band (L/M or N)
- FLAG column of OIVIS2, OIVIS, OIT3 and IOFLUX is now filled wrt the wavelength (L/M band : If (2.8um < lambda < 4.2um or 4.48um < lambda < 5.2um): flag=FALSE else flag=TRUE and 
N band : If (8um < lambda < 9.18um or 9.56um < lambda < 13um): flag=FALSE else flag=TRUE)
- QC Parameters are now computed on the same wavelength range

v1.5.0 (2020-02-11):

- Bugs corrected in the UV coordinates

- New workflows activated: matisse_uncalvis_LM.xml and matisse_uncalvis_N.xml

- Output of mat_cal_det optimized:
  + standard output lists major steps and important results
  + flags allows to get more detailed descriptions and output (see man-page)

v1.4.9 (2019-11-29):

- New disperion laws in L/M band have been computed. We used observations of Be stars
  with GRA4MAT to improve the spectrum of plastic foils. Now the laws are modelized
  with a polynome of order 4. A new parameter (--obsCorrection) has been added to
  mat_est_shift. You have to select --obsCorrection=TRUE in order to use the new laws.

- Correction of the bad tagging (S,T,U) of the raw frames. The correction would be
  specified with the parameter --tartyp of mat_raw_estimates. See man-page for a full
  description.

- We implement spectral binning in LM and N bands. The width in pixel (odd number) of
  the spectral binning could be specified with the parameter --spectralBinning of
  mat_raw_estimates.

- TARTYP correction method improved, uses up to three methods for determining a possible
  shift (more flags available for --tartyp) and allows to specify the shift for a subset
  of exposures in a SOF (--excess_count_list).

v1.4.4 (2019-07-03):

- Unused recipe parameters were removed:
   + mat_raw_estimates and mat_est_tf    : diamStar, diamErr, Catalog
   + mat_raw_estimates and mat_ext_beams : useKappaMatrix, useAvgShape, useAvgSky
   + mat_raw_estimates and mat_est_opd   : chromaticOpdFit
   + mat_raw_estimates and mat_cal_image : binning
   
- A temporal hampel filter was added to the recipes mat_raw_estimates
  and mat_ext_beams:
  The temporal hampel filter was added in order to reduce some extra
  detector noise (HAWAII-2RG detector) which affects the photometry
  estimation for low flux stars (< 5JY with ATs). The name of the new
  parameter (for mat_raw_estimates and mat_ext_beams) is
  "hampelFilterKernel".
  --hampelFilterKernel: Only for L/M band. Apply a temporal Hampel
        filter to all pixels before deriving the photometry. This
        filter improves the photometric estimation in case of faint
        stars (< 5Jy in L band with AT). The parameter fixes the size
        of the kernel of the filter (kernel=0 is recommended).

v1.4.3 (2019-06-12):

- Check for missing TIMDL.START and TIMDN.START header keywords
  added. If these keywords were missing, the pipeline aborted.

v1.4.2 (2019-06-06):

- Some problems in the N band photometry estimation of the
  mat_ext_beams recipe have been fixed: In N band the pipeline applies
  a Gaussian fit of the photometry in each spectral channel. In some
  (rare) cases the fit did not converge. If the fit is not converging,
  the photometry is equal to zero.

- Estimate and use an average non-linearity response for the LM-Band
  detector (mat_cal_det recipe) if requested.

- All combinations of cost function and algorithm mode in
  mat_cal_imarec recipe are now supported. Mi- nor errors fixed for
  mat_cal_imarec. Flux rescaling for calibrated input is implemented.

v1.4.1 (2019-05-22):

- First public release 
