GNU Astronomy Utilities NEWS                          -*- outline -*-

Copyright (C) 2015-2019 Free Software Foundation, Inc.
See the end of the file for license conditions.

* Noteworthy changes in release 0.11 (library 9.0.0) (2019-11-25) [stable]

** New features

  Book:
   - The "General program usage tutorial" now has a section on how to write
     scripts effectively to automate your analysis.

  Arithmetic:
   - The new `add-dimension' operator will stack the popped operands in a
     higher-dimensional dataset. For example to build a 3D cube from
     individual 2D images/slices.
   --onedonstdout: when the output is one-dimensional, print the values on
     the standard output, not into a file.

  BuildProgram:
   - Will use common environment variables like LDFLAGS, CPPFLAGS and CC to
     help in customizing the build of your program.
   --cc: custom C compiler to use. Until now, `gcc' was hard-coded into the
     source and there was no way to choose a custom C compiler.
   --noenv: With this option, no environment variables will be read.

  ConvertType:
   - New `viridis' colormap (value for the `--colormap' option). This is
     the default colormap of Python's Matplotlib, and is available in many
     other plotting tools like LaTeX's PGFPlots.

  Convolve:
   - Spatial domain convolution now possible on 3D data cubes (with a 3D
     kernel).

  CosmicCalculator:
   --lineatz: return the observed wavelength of a line if it was emitted at
     the redshift given to CosmicCalculator. You can either use known line
     names, or directly give a number as any emitted line's wavelength.

  MakeCatalog:
   - Catalogs from 3D inputs now available, with the following new options,
     see book for more.
     --spectrum: label's spectrum (across the third dimension).
     --z: Flux-weighted position in 3rd dimension.
     --geoz: Geometric center in third FITS axis.
     --minz: Minimum third FITS axis position.
     --maxz: Maximum third FITS axis position.
     --clumpsz: Flux weighted center of all clumps in object in 3rd dim.
     --clumpsgeoz: Geometric center of all clumps in object in 3rd dim.
     --w3: Flux weighted center in third WCS axis.
     --geow3: Geometric center in third WCS axis.
     --clumpsw3: Flux wheighted center of all clumps in 3rd dim.
     --clumpsgeow3: Geometric center of all clumps in 3rd dim.
     --areaxy: Projected area in first two dimentions.
     --geoareaxy: Projected geoarea in first two dimensions.
   --inbetweenints: output will contain one row for all integers between 1
     and the largest label in the input (irrespective of their existance in
     the input image). This was the default/only behavior of MakeCatalog
     until now. However, there are situations where the labeled input image
     integers may not be contiguous. For example if the input's only
     labeled pixel values are 11 and 13 from this release MakeCatalog's
     output will only have two rows. If you want the old behavior (of one
     row per integer, even if its not in the image), you can use this
     option.

  MakeProfiles:
   - Can produce mock ellipsoids in a datacube (using X-Z-X Euler angles
     for 3D orientation), the following options have been added, see the
     book for more details.
     --p2col: Second Euler angle (X-Z-X order).
     --p3col: Third Euler angle (X-Z-X order).
     --q2col: Axis ratio (major/dim3 in 3D).
   - The `--kernel' option can build 3D kernels, see the description of
     this option in the book for examples and details on how to run it.

  Match:
   - Matching of catalogs now possible using 3 coordinates (on catalogs
     generated from 3D data cubes), see book for more.

  NoiseChisel:
   - arXiv:1909.11230 added in papers to cite (with the `--cite' option):
     this paper describes the major changes made to NoiseChisel in the last
     10 stable releases since the 2015 paper, most importantly how Segment
     has been separated and the new growth strategy. It is therefore
     necessary to cite it along with the initial 2015 paper when using
     NoiseChisel.

  Segment:
   - arXiv:1909.11230 added in papers to cite (with the `--cite' option):
     this paper describes why Segment has been separated from NoiseChisel
     and some important updates to it compared to the 2015 paper, it is
     therefore necessary to cite it along with that paper when using
     Segment.

  Statistics:
   --contour: compute a contour plot which can be directly fed into the
     PGFPlots package of LaTeX for plotting the contours. Support for more
     formats will be added based on the need/request.

  Table:
   --equal: Output only rows that have a value equal to the given value in
     the given column. For example `--equal=ID,2,4,5' will select only rows
     that have a value of 2, 4 and 5 in the `ID' column.
   --notequal: Output only rows that have a different value compared to the
     values given to this option in the given column.
   - Column Arithmetic operators:
     - `angular-distance': a new operator to easily find the angular
       distance (along a great circle) between points in various table
       columns, or the distances of all the points in the table rows with a
       fixed point. See the book for examples and better explanation.

  Library:
   - gal_binary_connected_indexs: store indexs of connected components.
   - gal_blank_remove_realloc: Remove blanks and shrink allocated space.
   - gal_box_bound_ellipsoid_extent: Extent of 3D ellipsoid.
   - gal_box_bound_ellipsoid: Bounding box for a 3D ellipsoid.
   - gal_statistics_unique: Return unique (non-blank) elements of the input.

** Removed features

** Changed features

** Bugs fixed
  bug #56736: CosmicCalculator crash when a single value given to --obsline.
  bug #56747: ConvertType's SLS colormap has black pixels which should be orange.
  bug #56754: Wrong sigma clipping output when many values are equal.
  bug #56999: Compilation error on macOS 10.9 not recognizing AT_FDCWD.
  bug #57057: BuildProgram not using environment LDFLAGS or CPPFLAGS.
  bug #57101: Crop segmentation fault when no overlap exists in image-mode.
  bug #57164: MakeCatalog crashes when a label isn't in the dataset.
  bug #57180: MakeCatalog reporting infinity S/N when --instd isn't an image.
  bug #57200: Generated pkgconfig must request wcslib, not wcs.
  bug #57293: NaN value for brightness-related columns when values have NaN.





* Noteworthy changes in release 0.10 (library 8.0.0) (2019-08-03) [stable]

** New features

  Installation:
   - With the the following options at configure time, its possible to
     build Gnuastro without the optional libraries (even if they are
     present on the host system): `--without-libjpeg', `--without-libtiff',
     `--without-libgit2'.

  All programs:
   - When an array is memory-mapped to non-volatile space (like the
     HDD/SSD), a warning/message is printed that shows the file name and
     its size. Later, when its deleted, a warning/message is also printed,
     informing you that it has been deleted. These warnings can be very
     useful when you actually have enough RAM, but forget to increase the
     `--minmapsize' value (therefore significantly slowing down the
     program). When you don't have enough RAM, but don't want to be annoyed
     with the warnings, you can use the new `--quietmmap' option to disable
     them.

  Arithmetic:
   - `unique' operator removes all duplicate (and blank) elements from the
     dataset and returns a single-dimension output, containing only the
     unique values in the dataset.

  Crop:
   - Can also crop 3D datasets (data cubes). A 3D crop has the same syntax
     as the old 2D mode, only when the dataset is 3D, three coordinates
     (values, ranges or catalog-columns) should be given to the relevant
     option. Just note that `--polygon' crops are still not supported in
     3D.

  CosmicCalculator:
   --obsline: alternative way to set the used redshift. With this option
     instead of explicitly giving the redshift, you can give a rest-frame
     and observed wavelength and the redshift to use will be calculated
     internally. For many lines, it is possible to give the line name
     instead of its rest-frame wavelength. For example
     `--obsline=lyalpha,6000' will use the redshift where the Lyman-alpha
     line has been shifted to 6000 Angstroms.
   --usedredshift: Print the used redshift as a "Specific calculation" (in
     line with other single-valued calculations).

  Fits:
   --primaryimghdu: Copy/cut the given image HDU to the zero-th/first
     HDU of the output file that doesn't yet exist.

  Statistics:
   --sigclip-number, --sigclip-median, --sigclip-mean, --sigclip-std: Do
     sigma-clipping and only print the desired value as a single-value
     measurement. Until now sigma-clipping results included a lot of
     visually useful information, which also made automatic usage of
     results hard. These options fix this issue. Please see the example in
     the book under `--sigclip-median' for a nice use case.

  Table:
   - Column arithmetic. It is now possible to apply many operations on the
     input table columns before printing them in the output. Similar to
     Arithmetic, but on table columns. The operators and notation is just
     like the Arithmetic program. See the "Column Arithmetic" section of
     the book for a detailed discussion and several examples.
   - WCS to Image coordinate conversion with `wcstoimg' and `imgtowcs'. For
     example if the input catalog has at least an `ID' column and two `RA'
     and `DEC' columns, the set of options below will produce 5 columns
     where the last two columns are the image coordinates for each row
     based on the WCS in `a.fits':
           `-cID,RA,DEC -c"arith RA DEC wcstoimg" --wcsfile=a.fits'
   --head: Only output the given number of rows from the top of columns.
   --tail: Only output the given number of rows from the bottom of columns.

  Library:
   - New `speclines.h' library functions and macros related to spectral
     lines. It has many macros with line wavelengths, and several functions
     for using them in combination with their names.
   - list.h: Functions to return the last element in linked lists. For
     example `gal_list_sizet_last' or `gal_list_data_last'.
   - gal_arithmetic_operator_string: Return operator string from code.
   - gal_arithmetic_set_operator: Return operator code from string.
   - gal_blank_initialize_array: Initialize an array with blank values.
   - gal_dimension_remove_extra: Remove extra (length 1) dimensions.
   - gal_list_data_to_array_ptr: Make an array of pointers from the list.
   - gal_fits_img_info_dim: Only return the size information of a dataset.
   - GAL_TYPE_INT: Corresponding to respective width based on system.
   - GAL_TYPE_UINT: Corresponding to respective width based on system.
   - GAL_BLANK_INT: Blank value for `int' (can be 16-bit or 32-bit).
   - GAL_BLANK_UINT: Blank value for unsigned `int' (can be 16-bit or 32-bit).

** Removed features

** Changed features

  Installation:
   - Better `./configure' tests (using Gnulib's `AC_LIB_HAVE_LINKFLAGS') to
     avoid some crashes during `make' when the host had multiple
     conflicting versions of some dependencies (GSL in particular).

  Arithmetic:
   - The output of co-adding operators is no longer the same type as the
     input in general. The output of the `min' and `max' operators are
     still the same type as the input. However the `number' and
     `sigclip-number' operators will output an unsigned 32-bit integer type
     and the rest (`sum', `mean', `std', `median', `sigclip-median',
     `sigclip-mean' and `sigclip-std') return 32-bit floating point
     datasets

  MakeCatalog:
   - When a clumps catalog is requested, MakeCatalog will automatically
     deduce the total number of clumps (at a small cost in
     performance). Until now, it was mandatory for the clumps label dataset
     to contain the total number of clumps in the `NUMLABS' keyword.

  Library:
   - gal_statistics_outlier_flat_cfp: Improved implementation with new API.
   - New `quietmmap' argument added to the following functions (as the
     argument following `minmapsize'). For more, see the description above
     of the new similarly named option to all programs: `gal_array_read'
     `gal_array_read_to_type', `gal_array_read_one_ch',
     `gal_array_read_one_ch_to_type', `gal_data_alloc',
     `gal_data_initialize', `gal_fits_img_read',
     `gal_fits_img_read_to_type', `gal_fits_img_read_kernel',
     `gal_fits_tab_read', `gal_jpeg_read', `gal_label_indexs',
     `gal_list_data_add_alloc', `gal_match_coordinates',
     `gal_pointer_allocate_mmap', `gal_table_read', `gal_tiff_read' and
     `gal_txt_image_read'

  Book:
   - The two larger tutorials ("General program usage tutorial", and
     "Detecting large extended targets") have been broken into subsections
     for easier readability.
   - The "Hubble visually checks and classifies his catalog" tutorial has
     been removed because it didn't come with a dataset, so it was hard for
     people to use. Also, all its operations were already present in the
     general tutorial.

** Bugs fixed
  bug #56195: astscript-sort-by-night crashing because of AWK.
  bug #56246: Single-valued measurement must abort with no value in Statistics.
  bug #56256: Segmentation fault when reading plain text array/image.
  bug #56257: ConvertType: Values not preserved when converting text to FITS.
  bug #56299: CosmicCalculator fails at z=0.
  bug #56324: Column metadata not usable when input is from pipe/stdin.
  bug #56424: Warp crashes with empty string given to options.
  bug #56480: Segfault in statistics library's histogram function.
  bug #56641: MakeProfile's center position changes based on precision.
  bug #56635: Update tutorial 3 with bug-fixed NoiseChisel.
  bug #56662: Converting -R to -Wl,-R causes a crash in configure on macOS.
  bug #56671: Bad sorting with asttable if nan is present.
  bug #56709: Segment crash when input has blanks, but labels don't.
  bug #56710: NoiseChisel sometimes not including blank values in output.





* Noteworthy changes in release 0.9 (library 7.0.0) (2019-04-17) [stable]

** New features

  All programs:
   --checkconfig: print the names and values given to options as they are
     parsed on the command-line or in various configuration files (the
     configuration file name is also printed). This can be very useful in
     debugging (finding which configuration file is responsible for a given
     option's value).

  Arithmetic:
   - The new `tofile-' and `tofilefree-' operators will save the top
     operand into a file. They can be very handy in debugging/understanding
     an Arithmetic command (especially as it gets complicated), or to
     produce multiple files/extensions with a single call to Arithmetic.
   - Four new operators have been added to allow coadding multiple datasets
     into one using sigma-clipping: `sigclip-number', `sigclip-mean',
     `sigclip-median', and `sigclip-std'. These are very useful when
     several inputs have strong outliers that affect the median, or the
     mean is required.
   - Multithreaded operation for the following operators that
     combine/co-add multiple inputs into one output with same size: `min',
     `max', `number', `sum', `mean', `std', `median', `sigclip-number',
     `sigclip-median', `sigclip-mean', `sigclip-std'.
   --wcsfile and --wcshdu: these two options can be used to specify a
     different file for reading the WCS of the output. This is useful when
     the default (the WCS of the first dataset that is read) is not the
     required one.
   --interpmetric: new option that is necessary with the
     `interpolate-medianngb' operator. For more, see the description of
     this option in NoiseChisel.

  Fits:
   - Add "title" to group FITS keywords with `--write=/,"title name". This
     "title" is composed of two keyword records/lines: a blank one (all
     whitespace), followed by another starting with `/' and ending in any
     string given to this option. This visually separates the keywords and
     acts as a title. Classifying the keywords into contextually similar
     groups greatly helps in visual inspection and is encouraged.
   - Calculate and write `CHECKSUM' and `DATASUM' integrity keywords into
     the specified header using `--write=checksum' (for both) or
     `--write=datasum' (only for `DATASUM').
   --datetosec: Convert the FITS date format (old or new) to number of
     seconds since since the Unix epoch time (1970-01-01,00:00:00). The
     FITS date format (for example `YYYY-MM-DDThh:mm:ss') is hard to use
     for automatic processing (requires calendar peculiarities like number
     of days in each month, or leap years and etc). With this option a
     single integer is returned that can be used for example to sort FITS
     files by date keywords without worrying about calendar peculiarities.
   --verify: confirm if the `DATASUM' and `CHECKSUM' keyword values agree
     with the specified HDU's content and/or data.
   --copykeys: Copy several keyword records (in a given range) from one
     FITS HDU/extension into another (possibly in another file).
   --outhdu: The name/number of the output HDU (for `--copykeys').

  Match:
   - All the columns from one of the input catalogs can now be merged with
     any of the columns of the second using the special `_all' name of
     `--outcols'. For example the output of `--outcols=a_all,b5' will
     contain all the columns from the first input and the 5th column of the
     second input. This greatly simplifies the merging of different table
     columns into one.
   --coord: manually specify coordinates to match on the
     command-line. Until now, if you only wanted to make check a specific
     coordinate's matching with a catalog. It was necessary to make a
     single-row catalog as a file and feed that into Match. With this
     option, you can now specify the coordinates to match against another
     catalog with the command-line.

  NoiseChisel:
   --interpmetric: Set the metric to use to identify the nearest neighbors
     for tile interpolation (quantile threshold, initial Sky, and final
     Sky). Until now only the manhattan/taxicab metric was available, which
     is fast, but could cause 45-degree lines in the interpolation. From
     this version, with this option, its also possible to use the radial
     distance (which is now the default). Just note that if there are many
     tiles over the image, the radial distance will be slower.
   --snthresh: Manually set the signal-to-noise ratio of true
     pseudo-detections. With this option, NoiseChisel will not attempt to
     find pseudo-detections over the noisy regions of the dataset, but will
     directly go onto applying the manually input value.
   - Several new options have been added to remove NoiseChisel's dependence
     on values that were hard-coded in its source and thus not modifiable
     at run-time by the user. To allow full configurability, these steps
     can also be configured by the user.
     --pseudoconcomp: allows setting the connectivity (4 or 8, in a 2D
       image) to define separate pseudo-detections. If its stronger,
       pseudo-detections that are touching on the corner will be identified
       as one.
     --dopening: The number of openings to do after applying `--dthresh'.
     --dopeningngb: The connectivity (4 or 8) to use for `--dopening'.

  Statistics:
   --interpmetric: Similar to NoiseChisel.

  Table:
   --range: Limit the output rows to those with a value within the given
     numeric range with this format: `--range=COL,low,high'. This is very
     useful when only certain rows of the input are required not the
     output. The advantage over piping to AWK is that you can save the
     output directly to FITS (preserving the metadata). See the book for
     more.
   --sort: Sort the output rows based on the value of the given column in
     ascending order.
   --descending: When called with `--sort', will arrange the output rows in
     descending order.

  Installed scripts:
   With this release, Gnuastro also installs Bash scripts for common
   higher-level usage of (possibly multiple) programs. These scripts have a
   `astscript-*' name, to easily show up on the command-line as Gnuastro
   executables with the other Gnuastro programs, but are identifiable from
   them. They support options just like the programs (which can be listed
   with `--help'). Please see the new "Installed scripts" chapter of the
   book for more.
   - astscript-sort-by-night: New Gnuastro executable, using Gnuastro's
     Fits program to identify files with dates in the same night (possibly
     spanning two calendar dates).

  Library:
    GAL_BLANK_LONG: new macro for the `long' type blank value.
    GAL_BLANK_ULONG: new macro for the `unsigned long' type blank value.
    gal_blank_number: Return the number of blank elements in a dataset.
    gal_dimension_dist_radial: Radial distance between two coordinates.
    gal_fits_key_date_to_struct_tm: FITS date format to C broken-down time.
    gal_fits_key_date_to_seconds: FITS date format to Unix epoch time.
    gal_qsort_index_single_TYPE_i: Set of functions to sort indexs ascending.
    gal_qsort_index_single_TYPE_d: Set of functions to sort indexs descending.
    gal_statistics_outlier_cumulative: Uses flatness of the cumulative
       distribution to find outliers.
    gal_table_list_of_indexs: returns the list of indexs matching columns.
    gal_type_is_int: to see if we have an integer (any width, any sign).

** Removed features

** Changed features

  Arithmetic:
   - `num' operator is renamed to `number'.
   - `numvalue' operator is renamed to `numbervalue'.
   - `--dontdelete' will append the output to any existing file. Note that
     this change is only in Arithmetic, other programs will still just
     complain and abort.

  ConvertType:
   --forcemin & --forcemax: until now, `--flminbyte' and `--flmaxbyte' were
     used to force the range of conversion to color channels (even if the
     range is beyond the limits of the dataset). With the introduction of
     color maps in Gnuastro 0.8, it is also necessary to force a range on
     non-byte datasets. It is thus necessary to use a more generic name.

  MakeCatalog:
   --std: Until now, this option would measure the mean standard deviation
     under the label. But this is not a statistically meaningful measure
     for the Sky standard deviation and could be incorrectly used. From now
     on, this option measures the square root of the mean variance, or root
     mean square (the correct definition of the Sky standard deviation).

  NoiseChisel:
   --ignoreblankintiles: Until now `--ignoreblankinsky', would specify if
     blank values should also be written into the tiled Sky and Sky
     standard deviation outputs. But NoiseChisel can optionally produce
     many more tiled outputs (for example with `--checkqthresh'). So the
     option was renamed to `--ignoreblankintiles' to highlight that the
     status of blank elements can be set in all tiled outputs.

  Statistics:
   --ignoreblankintiles: similar to same option in NoiseChisel.

  Table:
   --colinfoinstdout: now corresponds to the `-O' short option. Until this
     version, the `-s' short option was used for it. But with the new
     `--sort' option, `-s' may cause confusion.

  Library
   gal_arithmetic: new argument: number of threads to use (when relevant).
   gal_eps_write: new argument: optional bit-optimization with `dontoptimize'.
   gal_pdf_write: new argument: optional bit-optimization with `dontoptimize'.

** Bugs fixed
  bug #55313: Fits program writing --write values in reverse order
  bug #55333: Fits program crash when --write or --update have no value.
  bug #55439: Arithmetic segmentation fault when reusing dataset name.
  bug #55478: Memory mapping crashes when .gnuastro is not writable.
  bug #55491: NoiseChisel crash when no tiles good for quantile thresholding.
  bug #55544: Arithmetic's output WCS with where operator is not as expected.
  bug #55740: Diamond shapes in nearst-ngb interpolation affecting NoiseChisel.
  bug #55763: Crop not keeping Blank pixels on unsigned types.
  bug #55844: WCS library (and thus all programs) can't deal with CROTAi values.
  bug #55845: Crash when necessary column not found from standard input.
  bug #55988: MakeProfiles segfault when input catalog has no rows.
  bug #56001: Bad tile sizes with only one tile and small remainder.
  bug #56048: Crash when WCS cannot be written to FITS file.





* Noteworthy changes in release 0.8 (library 6.0.0) (2018-12-28) [stable]

** New features

  All programs:
   - Input files and parameters written as keywords in the first extension
     of output when it is FITS. This is only relevant for some programs,
     (for example not the Fits or Table programs).
   - Standard input (for example from pipes) is now available to feed input
     to all programs that accept plain text input (ConvertType, Convolve,
     Match, MakeProfiles, Statistics, Table).
   - Updated acknowledgment statement (output of `--cite' option).

  Arithmetic:
    --onedasimage: write output as an image if it has one dimension, not table.

  ConvertType:
    --colormap: color map to display a single-channel dataset (for example
      FITS image) with a range of colors in formats that support color
      (like JPEG, or PDF). Until now, the only available mapping was
      grayscale. Now "Hue, Saturation, Value" (HSV) and "SLS" (from SAO
      DS9) colormaps are also supported.
    --rgbtohsv: Convert the RGB input channels to HSV (when the output is
      in FITS format).

  Convolve:
    - Convolves 1D arrays (table columns, for example spectra)
      also. Therefore two new options have been added to it: `--column'
      (`-c', similar to other programs that read table columns), and
      `--kernelcolumn' (to specify the column of the kernel in its own
      file/extension).

  Fits:
    --numhdus: prints the number of HDUs in the given FITS file.

  NoiseChisel:
    - New outlier identification algorithm for quantile thresholds. This is
      very useful when there are extended and bright sources in the
      dataset: the tiles containing very faint signal that pass the general
      pixel-value distribution test due to the flatness of the extended
      profiles, can be identified and removed as outliers in comparison
      with the other passed tiles. The outlier finding algorithm
      (`gal_statistics_outlier_positive': a new library function) uses the
      distribution of distances between the sorted elements and is
      configured with these options.
       --outliersclip: Sigma-clipping parameters for the process.
       --outliersigma: Multiple of sigma to define an outlier.
    --blankasforeground: Treat blank elements as foreground (regions above
          the threshold) in the binary processing steps: initial erosion
          and opening as well as the filling holes and opening of
          pseudo-detections. From this version, by default, blank elements
          in the dataset are considered to be background, so if a
          foreground pixel is touching it, it will be eroded. This option
          is irrelevant if the datasets contains no blank elements, but can
          help remove false positives that are touching blank elements.
    --holengb: allows defining the connectivity of the holes that are
          filled when defining pseudo-detections. Until now, this was
          hard-wired into the code (=8) and not modifiable at run-time.
    --skyfracnoblank: Ignore blank pixels when estimating the fraction of
          undetected pixels for Sky estimation. NoiseChisel only measures
          the Sky over the tiles that have a sufficiently large fraction of
          undetected pixels. This is done to decrease the bias caused by
          faint un-detected wings of bright galaxies or stars, see
          description of `--minskyfrac' for more. Until now the reference
          for this fraction was the whole tile size (irrespective of how
          many blank elements it contains). With this option, it is now
          possible to ask for ignoring blank pixels when calculating the
          fraction. This is useful when blank/masked pixels are distributed
          across the image.

  Statistics:
    - If an input table has only one column, Statistics won't complain and
      abort when no `--column' (`-c') is given: there is only one column to
      use anyway, so it will be used. In the absence of which column to
      use, it will still complain and abort if the input has more than one
      column.
    - Sky estimation: new outlier estimation algorithm similar to
      NoiseChisel.
    - Input can be given using the standard input (for example a pipe).

  Library:
    - gal_blank_flag_apply: Set all flagged/masked elements to blank.
    - gal_fits_key_list_reverse: Reverse the given list of FITS keywords.
    - gal_fits_key_write_title_in_ptr: Write a two-line title FITS keyword.
    - gal_fits_key_write_in_ptr: New name of gal_fits_key_write.
    - gal_fits_key_write_version_in_ptr: old gal_fits_key_write_version.
    - gal_fits_key_write_config: write key list and version as config header.
    - gal_statistics_outlier_positive: Find the first positive outlier.
    - gal_txt_stdin_read: Read lines in standard input into a list of strings.

** Removed features

  NoiseChisel:
    --mirrordist: mean quantile is now used (not mode), see changes below.
    --qthreshtilequant: removed due to new outlier rejection algorithm.

  Statistics:
    --mirrordist: mean quantile is now used (not mode), see changes below.

** Changed features

  Arithmetic:
    - If the output has one dimension, it will be written as a table, not a
      FITS image/array. This can be changed with the new `--onedasimage'
      option.

  Convolve:
    - The short option for `--numchannels' is now `-n'. Until now, it was
      `-c', but that would conflict with the short option used for
      `--column' in all the other programs that also read from a table.

  MakeProfiles:
    --mergedsize: new name for the old `--naxis' option. Since the option
          names and values are now written into the FITS header of the
          output, this option's name would get confused with the mandatory
          FITS keyword `NAXIS'.

  NoiseChisel:
    - Until now, the mode's quantile was used to identify tiles with no
      significant signal. But from this version, the mean's quantile in
      each tile is used instead. The reason is that the mean is more
      sensitive to outliers (signal). Therefore the old `--modmedqdiff' is
      now called `--meanmedqdiff' .

  Statistics:
    --meanmedqdiff: new name for `--modmedqdiff'. Similar to NoiseChisel.

  Library:
    - gal_array_read: list of strings (from standard input) acceptable.
    - gal_array_read_to_type: list of strings (from stin) acceptable.
    - gal_array_read_one_ch: list of strings (from stdin) acceptable.
    - gal_array_read_one_ch_to_type: list of strings (from stdin) acceptable.
    - gal_data_copy_to_allocated: Also copies string metadata (e.g., name).
    - gal_fits_key_write: filename and hdu instead of FITS pointer.
    - gal_fits_key_write_version: filename and hdu instead of FITS pointer.
    - gal_fits_key_write_filename: write at the top or end of the input list.
    - gal_statistics_outlier_positive: Window-size is now adjustable (new argument).
    - gal_table_info: list of strings (from stdin) acceptable.
    - gal_table_read: list of strings (from stdin) acceptable.
    - gal_txt_table_info: list of strings (from stdin) acceptable.
    - gal_txt_image_info: list of strings (from stdin) acceptable.
    - gal_txt_table_read: list of strings (from stdin) acceptable.
    - gal_txt_image_read: list of strings (from stdin) acceptable.

** Bugs fixed
  bug #54493: Warp crashes when type isn't set.
  bug #54526: Invalid r, q and truncation of point profiles in MakeProfiles.
  bug #54579: NoiseChisel pseudo-detection failure when dataset is negative.
  bug #54782: Segment's check image not removing sky clumps some tiles.
  bug #54810: Arithmetic crash when previously named operand renamed.
  bug #55025: MakeCatalog's `--prepforconv' option being ignored.
  bug #55079: Blank EPS or PDF page when width options not given.
  bug #55157: No sanity check on values given to Crop's --section.
  bug #55295: Crash when more than one collapse operator called.
  bug #55298: Arithmetic reading 255 on command-line as blank.





* Noteworthy changes in release 0.7 (library 5.0.0) (2018-08-08) [stable]

** New features

  Installation:
    --enable-debug: debugging flags, no optimization, no shared libraries.
    --enable-check-with-valgrind: Run `make check' tests within Valgrind.

  Arithmetic:
    - `set-A': Set a name (`A' in this case) for the popped dataset. This
               allows only reading the dataset it into memory once and
               possibly using it many times.
    - `fill-holes': Flip background (0) pixels surrounded by foreground (1).
    - `collapse-sum': collapse/remove a dimension by summing over it.
    - `collapse-min': collapse/remove a dimension by using minimum value.
    - `collapse-max': collapse/remove a dimension by using maximum value.
    - `collapse-mean': collapse/remove a dimension by averaging over it.
    - `collapse-number': Number of elements included in the collapse.

  CosmicCalculator:
    - Default cosmology set to Plank 2018 results (Paper VI).

  MakeCatalog:
    --minx: minimum position along first FITS axis.
    --maxx: maximum position along first FITS axis.
    --miny: minimum position along second FITS axis.
    --maxy: maximum position along second FITS axis.

  Table:
    --colinfoinstdout: column information when writing to standard output.

  Library:
    - gal_array_name_recognized_multiext: If format contains multiple datasets.
    - gal_dimension_collapse_sum: collapse/remove a dimension by summing.
    - gal_dimension_collapse_mean: collapse/remove a dimension by averaging.
    - gal_dimension_collapse_number: collapse/remove a dimension by number.
    - gal_dimension_collapse_minmax: collapse/remove a dimension by extremum.
    - gal_wcs_remove_dimension: Remove a dimension in the given WCS structure.

** Removed features

** Changed features

  Crop:
    --checkcenter: the units of value depend on mode (image or WCS).

  MakeCatalog:
    - `--checkuplim': new name for `--checkupperlimit'.
    - `--brightnessnoriver': new name for `--noriverbrightness'.

  Library:
    - gal_txt_write: new `colinfoinstdout' argument.
    - gal_table_write: new `colinfoinstdout' argument.

** Bugs fixed

  bug #54057: Building failure due to not finding gsl_interp_steffen.
  bug #54063: Match tests in make check fail randomly.
  bug #54186: MakeCatalog's --checkupperlimit not keeping output's name.
  bug #54188: MakeCatalog's Upperlimit not being sigma-clipped properly.
  bug #54284: Crop segfault when catalog contains no data.
  bug #54285: make check fails if g++ not present.
  bug #54286: BuildProgram's configuration file, not built by default.
  bug #54297: No Match output when --notmatched called and no match.
  bug #54298: Table not writing array when there are no rows.
  bug #54312: Crash when CFITSIO doesn't have fits_is_reentrant function.
  bug #54346: Non '-I' or non '-L' strings in CPPFLAGS or LDFLAGS cause crash.
  bug #54358: Arithmetic's where, not ignoring blank values in condition.
  bug #54406: Insufficient sanity checks in mode symmetricity calculation.
  bug #54430: BuildProgram fails with older Libtool versions on dash.




* Noteworthy changes in release 0.6 (library 4.0.0) (2018-06-04) [stable]

** New features

  Building:
    - New optional dependency: The TIFF library (libtiff).

  All programs:
    - Input image dataset(s) can be in any of the formats recognized by
      Gnuastro (e.g., FITS, TIFF, JPEG), provided that their libraries
      (optional dependencies) were present at installation time.

  New program:
    - Segment: new program in charge of segmentation over detections. This
      operation was previously done by NoiseChisel. NoiseChisel is now ONLY
      in charge of detection.

  Arithmetic:
    - erode: Erode the foreground of a binary dataset.
    - dilate: Dilate the foreground of a binary dataset.
    - filter-sigclip-mean: sigma-clipped, mean filter operator.
    - filter-sigclip-median: sigma-clipped, median filter operator.
    - connected-components: label the connected elements of the input.
    - invert: subtract the maximum of unsigned types (absorption to emission).
    - interpolate-medianngb: Interpolate (only blanks) with nearest neighbors.

  ConvertType:
    - TIFF images can also be used as input.

  MakeCatalog:
    MakeCatalog will only read the datasets necessary for the requested
    columns. Until now, it would read all the possible datasets and all the
    intermediate measurements. This is thus major improvement in memory and
    CPU usage. As a result, the input argument is no longer assumed to be
    the values file, but the object labels file. Please see the
    "MakeCatalog inputs and basic settings" section of the book for
    more. Here is the summary of the new options:
    --insky: Sky value as a single value or file (dataset).
    --instd: Sky standard deviation as a single value or file (dataset).
    --valuesfile: filename containing the values dataset.
    --valueshdu: HDU/extension containing the values dataset.
    --clumpscat: Make a clumps catalog also.
    --noclumpsort: Don't sort the clumps catalog by host object ID.
    --subtractsky: Subtract the given Sky from the values dataset.
    --variance: input standard deviation image is actually variance.
    --checkupperlimit: make a table for random positions and measurements.
    --geoarea: the number of labeled pixels (irrespective of value).
    --brightnesserr: error in estimating the brightness.
    --mean: calculate the mean pixel value within an object or clump.
    --median: calculate the median pixel value within an object or clump.
    --upperlimitsigma: position in random distribution (in units of sigma).
    --upperlimitquantile: position in random distribution (quantile).
    --upperlimitonesigma: 1sigma value of the random distribution.
    --upperlimitskew: (mean-median)/sigma or skewness of random distribution.

  NoiseChisel:
    - New tutorial on detecting large and extended targets.
    --rawoutput: only output the detection labels and Sky and its STD.
    --ignoreblankinsky: don't set the pixels that are blank in the input to
      blank in the Sky and Sky standard deviation outputs (when
      `--oneelempertile' is not called).
    --label: label the connected detections. Until now this was the default
      behavior. However, from this release, NoiseChisel is only in charge
      of detection. Segmentation is done by a new program (Segment). Since
      detection is ultimately just a binary operator, the default output
      now has an 8-bit unsigned integer type with values of 0 or 1. With
      this option, you can ask for it to label/count the connected
      detections instead of the default binary output.

  Statistics:
    --manualbinrange: histogram or CFP range can be outside of distribution.
    --ignoreblankinsky: similar to same option in NoiseChisel.

  Libraries:
    gal_array_read: read array from any of known formats (FITS, TIFF, JPEG,...).
    gal_array_read_to_type: similar to `gal_array_read', but to given type.
    gal_array_read_one_ch: Read a dataset, abort if it has multiple channels.
    gal_array_read_one_ch_to_type: Make sure input is in one channel and type.
    gal_binary_label_holes: label the holes within the foreground.
    gal_blank_is: check to see if argument is blank in its type or not.
    gal_eps_name_is_eps: Returns 1 if given filename is EPS.
    gal_eps_suffix_is_eps: Returns 1 if given suffix is EPS.
    gal_eps_to_pt: Converts dataset size to PostScript points.
    gal_eps_write: Writes a dataset into an EPS file.
    gal_interpolate_1d_blank: Fill blank elements through interpolation.
    gal_interpolate_1d_make_gsl_spline: Allocate and initalize `gsl_spline'.
    gal_jpeg_name_is_jpeg: Returns 1 if given filename is JPEG.
    gal_jpeg_suffix_is_jpeg: Returns 1 if given suffix is JPEG.
    gal_jpeg_read: Reads input JPEG image into `gal_data_t'.
    gal_jpeg_write: Writes a `gal_data_t' into a JPEG file.
    gal_label_grow_indexs: grow known indexs into desired areas.
    gal_label_watershed: apply watershed algorithm on desired region.
    gal_label_clump_significance: measure significance of all clumps in region.
    gal_pdf_name_is_pdf: Returns 1 if given filename is PDF.
    gal_pdf_suffix_is_pdf: Returns 1 if given suffix is PDF.
    gal_pdf_write: Writes a dataset into an PDF file.
    gal_pointer_allocate_mmap: Allocate space in a file, not in RAM.
    gal_qsort_index_single_d: Sort indexs of single array in decreasing order.
    gal_qsort_index_single_i: Sort indexs of single array in decreasing order.
    gal_qsort_index_multi_d: Sort indexs in multiple arrays (different threads).
    gal_qsort_index_multi_i: Sort indexs in multiple arrays (different threads).
    gal_tiff_name_is_tiff: check if name contains a TIFF suffix.
    gal_tiff_suffix_is_tiff: check if suffix is a TIFF suffix.
    gal_tiff_dir_string_read: convert a string to a TIFF directory number.
    gal_tiff_read: Read the contents of a TIFF "directory" to `gal_data_t'.

** Removed features

  NoiseChisel:
    - Segmentation (and thus the options below) moved to the new Segment
      program: --onlydetection, --segsnminarea, --checkclumpsn, --segquant,
      --keepmaxnearriver, --gthresh, --minriverlength, --objbordersn,
      --grownclumps, --checksegmentation.

    --skysubtracted: no longer necessary, included in noise measuremnts.

  MakeCatalog:
    --skysubtracted: no longer necessary, included in noise measuremnts.

  Library:
    - The macros `GAL_STATISTICS_SORTED_NOT',
      `GAL_STATISTICS_SORTED_INVALID', `GAL_STATISTICS_SORTED_INCREASING',
      `GAL_STATISTICS_SORTED_DECREASING': these macros are removed because
      we already have the `GAL_DATA_FLAG_SORT*'' bit-flags in `gal_data_t'.


** Changed features

  Arithmetic:
    - filter-mean: a blank value in the input can be non-blank in the
         output when non-blank elements present in filter.
    - filter-median: similar to filter-mean.

  Fits:
    --history: can be called/written multiple times in one run.
    --comment: can be called/written multiple times in one run.

  MakeCatalog:
    - The `WCLUMPS' keyword in the objects labeled image is no longer used
         to see if a clumps catalog should also be made. To build a clumps
         catalog, you can now use the `--clumpscat' option.
    - Estimation of noise-level is now done per-pixel over the whole
         label. Until now the average noise level was used.
    --objectsfile has been removed. The main input argument is now assumed
         to be the objects file.

  NoiseChisel:
    From this release, NoiseChisel is only in charge of detection and won't
    do segmentation any more. The new Segment program is in charge of
    segmentation. Many of the changes below are due to this new limited
    scope of NoiseChisel.
    --kernel: value `none' will disable convolution.
    - Renamed options:
      --convolvedhdu ==> --chdu
      --wkhdu        ==> --whdu
      --detsnminarea ==> --snminarea
      --checkdetsn   ==> --checksn
      --detquant     ==> --snquant
    - By default the output detection map is a binary image (values of 0 or 1).
    - With no output name, the output has a `_detected.fits' suffix.

  Segment:
    - [Previously in NoiseChisel]: For finding true clumps, the difference
      in the peak of the clump and the highest valued river pixel, divided
      by the noise standard deviation are used. Until now, the total
      signal-to-noise ratio was used as a criteria. In initial tests, this
      algorithm was much more promising in detecting clumps over strong
      gradients and also on flatter gradients.

  Table:
    --column: multiple columns (comma separated) can be used in one
         instance of this option (multiple instances of this option are
         still acceptable also).

  Libraries:
    gal_binary_holes_fill: new name for `gal_binary_fill_holes'.
    gal_dimension_is_different: new name for `gal_data_dsize_is_different'.
    gal_fits_img_read: now only reads the data not the WCS, therefore it no
        longer needs the last two arguments. A subsequent call to
        `gal_wcs_read' can be used to read the WCS information in the file.
    gal_pointer_increment: new name for `gal_data_ptr_increment'.
    gal_pointer_num_between: new name for `gal_data_ptr_dist'.
    gal_pointer_allocate: replaces `gal_data_malloc_array' and
        `gal_data_calloc_array', through an argument you can ask for the
        allocated memory to be cleared or not.
    gal_qsort_TYPE_i: new name for gal_qsort_TYPE_increasing.
    gal_qsort_TYPE_d: new name for gal_qsort_TYPE_decreasing.
    gal_statistics_is_sorted: can now also update the bit-flags regarding
        the sorted nature of the input (to optimize future calls to the
        function).
    gal_statistics_quantile_function: returns `inf' or `-inf' if the given
        value is smaller than the minimum or larger than the maximum of the
        input dataset's range. Until now, it would return blank in such
        cases.
    gal_statistics_number: the output dataset now has a `size_t' type. Until
        now it was `uint64_t'.

** Bugs fixed

  bug #50957: --version output not possible on Mac OS X
  bug #52979: Many unused result warnings for asprintf in some compilers.
  bug #53122: Configure time CPPFLAGS and LDFLAGS don't pass to BuildProgram.
  bug #53142: Crash when printing values with the `--onlyversion' option.
  bug #53147: NULL value of onlyversion option causing a crash.
  bug #53226: Match output directory ignored when making multiple files.
  bug #53230: Statistics program bad results on integer columns with limits.
  bug #53268: NoiseChisel crash when no growth is possible.
  bug #53295: MakeCatalog parses area larger than clump.
  bug #53304: NoiseChisel crash when there is no detection.
  bug #53312: Fits crash on keyword editing (except --delete).
  bug #53407: Instrumental noise in MakeNoise should be squared.
  bug #53424: Sigma-clipping seg-faults when there are no more elements.
  bug #53580: Warp crash when no WCS present.
  bug #53825: NoiseChisel not accounting for fully zero-valued tiles.






* Noteworthy changes in release 0.5 (library 3.0.0) (2017-12-22) [stable]

** New features

  Manual/Book: An extended tutorial is added showing some general
  applications of almost all the programs. This may be a good place to get
  a feeling of how Gnuastro is intended to be used and some of the
  programs.

  New Program and library: Match is a new program that will match two given
  inputs (currently catalogs). Its output is the re-arranged inputs with
  the same number of rows/records such that all the rows match. The main
  work is also done with the new low-level `gal_match_catalog' library
  function which can also be used in more generic contexts.

  All programs: a value of `0' to the `--numthreads' option will use the
  number of threads available to the system at run time.

  Arithmetic: The new operators `filter-median' and `filter-mean' can be
  used to filter (smooth) the input. The size of the filter can be set as
  the other operands to these operators.

  BuildProgram: The new `--la' option allows the identification of a
  different Libtool `.la' file for Libtool linking information.

  BuildProgram: The new `--deletecompiled' option will delete the compiled
  program after running it.

  CosmicCalculator: all the various cosmological calculations can now be
  requested individually in one line with a specific option added for each
  calculation (for example `--age' or `--luminositydist' for the age of the
  universe at a given redshift or the luminosity distance). Therefore the
  old `--onlyvolume' and `--onlyabsmagconv' options are now removed. To
  effectively use these new features, please review the "Invoking
  CosmicCalculator" section of the book.

  Fits: when an extension/HDU is identified on the command-line with the
  `--hdu' option and no operation is requested, the full list of header
  keywords in that HDU will be printed (as if only `--printallkeys' was
  called).

  MakeCatalog: physical nature agnostic WCS column names. Previously the
  first WCS axis was always assumed to be RA and the second DEC. So for
  example even if you had a spectrum (with X and wavelength as the two WCS
  dimensions), you would have to ask for `--ra' and `--dec'. The new `--w1'
  and `--w2' options are now generic and don't assume any particular type
  only their order in the FITS header. MakeCatalog now also uses the CTYPE
  and CUNIT keywords to set the names and units of its output columns. The
  `--ra' and `--dec' options are now just internal aliases for `--w1' or
  `--w2' which will be determined based on the input's CTYPE keyword. Also
  the new `--geow1', `--geow2', `--clumpsw1', `--clumpsw2',
  `--clumpsgeow1', `--clumpsgeow2' options replace the old options
  `--geora', `--geodec', `--clumpsra', `--clumpsdec', `--clumpsgeora',
  `--clumpsgeodec'. No alias is currently defined for the latter group.

  MakeCatalog: the new `--uprange' option allows you to specify a range for
  the random values around each object. This is useful when the noise
  properties of the dataset vary gradually and sampling from the whole
  dataset might produce biased results.

  NoiseChisel: with the new `--convolved' and `--convolvedhdu' options,
  NoiseChisel will not convolve the input any more and use the given
  dataset instead. In many cases, as the inputs get larger, convolution is
  the most time consuming step of NoiseChisel. With this option, you can
  greatly speed up your tests (to find the best parameters by varying them,
  for a given analysis). See the book for more information and examples.

  NoiseChisel: with the new `--widekernel' option it is now possible to use
  a wider kernel to identify which tiles contain signal. The rest of the
  steps (identifying the quantile threshold on the selected tiles and etc)
  are done on the dataset convolved with `--kernel' as they were
  before. Since it is time consuming, this is an optional feature.

  NoiseChisel: with the new `--qthreshtilequant' option, it is now possible
  to discard high-valued (outlier) tiles before estimating qthresh over the
  whole image. This can be useful in detecting very large diffuse/flat
  regions that would otherwise be detected as background (and effectively
  removed).

  NoiseChisel: the finally selected true detections are now grown based on
  signal contiguity, not by blind dilation. The growth process is the same
  as the growing of clumps to define objects. Only for true detections, the
  growth occurs in the noise. You can configure this growth with the
  `--detgrowquant' and `--detgrowmaxholesize'. With this new feature it is
  now possible to detect signal out to much lower surface brightness limits
  and the detections don't look boxy any more.

  Cosmology library: A new set of cosmology functions are now included in
  the library (declared in `gnuastro/cosmology.h'). These functions are
  also used in the CosmicCalculator program.

  `gal_table_read' can now return the number of columns matched with each
  input column (for example with regular expressions), a new argument has
  been added to allow this feature.

  `gal_fits_key_img_blank': returns the value that must be used in the
  BLANK keyword for the given type as defined by the FITS standard.

  `gal_txt_write' and `gal_fits_tab_write' now accept an extension name as
  argument to allow a name for the FITS extension they write.

  `gal_box_bound_ellipse_extent' will return the maximum extent of an
  ellipse along each axis from the ellipse center in floating point.

** Removed features

  Installation: The `--enable-bin-op-*' configuration options that were
  introduced in Gnuastro 0.3 have been removed. By managing the arithmetic
  functions in a better manner (a separate source file for each operator),
  compilation for all types (when done in parallel) takes about the same
  time as it took with the default (only four) types until now.

  MakeCatalog: `--zeropoint' option doesn't have a short option name any
  more. Previously it was `-z' which was confusing because `-x' and `-y'
  were used to refer to image coordinate positions.

  NoiseChisel: The `--dilate' and `--dilatengb' options have been
  removed. Growing of true detections is no longer done through dilation
  but through the `--detgrowquant' and `--detgrowmaxholesize' options (see
  above).

** Changed features

  CosmicCalculator: The redshift is no longer mandatory. When no redshift
  is given, it will only print the input parameters (cosmology) and abort.

  MakeCatalog: when the output is a FITS file, the two object and clumps
  catalogs will be stored as multiple extensions of a single file. Until
  now, two separate FITS files would be created. Plain text outputs are the
  same as before (two files will be created).

  `gal_binary_fill_holes' now accepts a `connectivity' and `maxsize'
  argument to specify the connectivity of the holes and the maximum size of
  acceptable holes to fill.

  `gal_fits_img_read' and `gal_fits_img_read_to_type' now also read the WCS
  structure of the extension/HDU in a FITS file and have two extra
  arguments: `hstartwcs' and `hendwcs'. With these options it is possible
  to limit the range of header keywords to read the WCS, similar to how
  they are used in `gal_wcs_read'.

  `gal_txt_write', `gal_table_write_log', `gal_fits_tab_write' and
  `gal_txt_write' don't have the `dontdelete' argument any more. The action
  they take if the file already exists depends on the file: for FITS, a new
  extension will be added and for text, they will abort with an error.

  `gal_tile_block_write_const_value' and `gal_tile_full_values_write' now
  accept a new `withblank' option to set all pixels that are blank in the
  tile's block to be blank in the check image.

  `gal_wcs_pixel_area_arcsec2' will return NaN (instead of aborting) when
  input is unreasonable (not two dimensions or not in units of degrees).

  `gal_wcs_world_to_img' and `gal_wcs_img_to_world': Until now, these two
  WCS conversion functions would explicitly assume RA and Dec and work
  based on input arrays (so for example it was also necessary to give the
  number of elements and etc). They now accept `gal_data_t' as input for
  the input coordinates, thus their API has been greatly simplified and
  their functionality increased.

** Bugs fixed

  ConvertType crash when changing values (bug #52010).

  Arithmetic not accounting for integer blank pixels in binary operators
  (bug #52014).

  NoiseChisel segfault when memory mapping to a file (bug #52043).

  CFITSIO 3.42 and libcurl crash at Gnuastro configure time (bug #52152).

  MakeCatalog crash in upper-limit with full size label (bug #52281).

  NoiseChisel leaving unlabeled regions after clump growth (bug #52327).

  Arithmetic crash with no input tokens (bug #52422).

  Libtool checks only in non-current directory (bug #52427).





* Noteworthy changes in release 0.4 (library 2.0.0) (2017-09-13) [stable]

** New features

  All programs: `.fit' is now a recognized FITS file suffix.

  All programs: ASCII text files (tables) created with CRLF line
  terminators (for example text files created in MS Windows) are now also
  readable as input when necessary.

  Arithmetic: now has a new `--globalhdu' (`-g') option which can be used
  once for all the input images.

  MakeNoise: with the new `--sigma' (`-s') option, it is now possible to
  directly request the noise sigma or standard deviation. When this option
  is called, the `--background', `--zeropoint' and other option values will
  be ignored.

  MakeProfiles: the new `--kernel' option can make a kernel image without
  the need to define a catalog. With this option, a catalog (or
  accompanying background image) must not be given.

  MakeProfiles: the new `--pc', `--cunit' and `--ctype' options can be used
  to specify the PC matrix, CUNIT and CTYPE world coordinate system
  keywords of the output FITS file.

  MakeProfiles: the new `distance' profile will save the radial distance of
  each pixel. This may be used to define your own profiles that are not
  currently supported in MakeProfiles.

  MakeProfiles: with the new `--mcolisbrightness' ("mcol-is-brightness")
  option, the `--mcol' values of the catalog will be interpretted as total
  brightness (sum of pixel values), not magnitude.

  NoiseChisel: with the new `--dilatengb' option, it is now possible to
  identify the connectivity of the final dilation.

  Library: Functions that read data from an ASCII text file
  (`gal_txt_table_info', `gal_txt_table_read', `gal_txt_image_read') now
  also operate on files with CRLF line terminators.

** Removed features

** Changed features

  Crop: The new `--center' option is now used to define the center of a
  single crop. Hence the old `--ra', `--dec', `--xc', `--yc' have been
  removed. This new option can take multiple values (one value for each
  dimension). Fractions are also acceptable.

  Crop: The new `--width' option is now used to define the width of a
  single crop. Hence the old `--iwidth', `--wwidth' were removed. The units
  to interpret the value to the option are specified by the `--mode'
  option. With the new `--width' option it is also possible to define a
  non-square crop (different widths along each dimension). In WCS mode, its
  units are no longer arcseconds but are the same units of the WCS (degrees
  for angles). `--width' can also accept fractions. So to set a width of 5
  arcseconds, you can give it a value of `5/3600' for the angular
  dimensions.

  Crop: The new `--coordcol' option is now used to determine the catalog
  columns that define coordinates. Hence the old `--racol', `--deccol',
  `--xcol', and `--ycol' have been removed. This new option can be called
  multiple times and the order of its calling will be used for the column
  containing the center in the respective dimension (in FITS format).

  MakeNoise: the old `--stdadd' (`-s') option has been renamed to
  `--instrumental' (`-i') to be more clear.

  MakeProfiles: The new `--naxis' and `--shift' options can take multiple
  values for each dimension (separated by a comma). This replaces the old
  `--naxis1', `--naxis2' and `--xshift' and `--yshift' options.

  MakeProfiles: The new `--ccol' option can take the center coordinate
  columns of the catalog (in multiple calls) and the new `--mode' option is
  used to identify what standard to interpret them in (image or
  WCS). Together, these replace the old `--xcol', `--ycol', `--racol' and
  `--deccol'.

  MakeProfiles: The new `--crpix', `--crval' and `--cdelt' options now
  accept multiple values separated by a comma. So they replace the old
  `--crpix1', `--crpix2', `--crval1', `--crval2' and `--resolution'
  options.

  `gal_data_free_contents': when the input `gal_data_t' is a tile, its
  `array' element will not be freed. This enables safe usage of this
  function (and thus `gal_data_free') on tiles without worrying about the
  memory block associated with the tile.

  `gal_box_bound_ellipse' is the new name for the old
  `gal_box_ellipse_in_box' (to be more clear and avoid repetition of the
  term `box'). The input position angle is now also in degrees, not
  radians.

  `gal_box_overlap' now works on data of any dimensionality and thus also
  needs the number of dimensions (elements in each input array).

  `gal_box_border_from_center' now accepts an array of coordinates as one
  argument and the number of dimensions as another. This allows it to work
  on any dimensionality.

  `gal_fits_img_info' now also returns the name and units of the dataset
  (if they aren't NULL). So it takes two extra arguments.

  `gal_wcs_pixel_scale' now replaces the old `gal_wcs_pixel_scale_deg',
  since it doesn't only apply to degrees. The pixel scale units are defined
  by the units of the WCS.

  `GAL_TILE_PARSE_OPERATE' (only when `OTHER' is given) can now parse and
  operate on different datasets independent of the size of allocated block
  of memory (the tile sizes of `IN' and `OTHER' have to be identical, but
  not their allocated blocks of memory). Until now, it was necessary for
  the two blocks to have the same size and this is no longer the case.

** Bugs fixed

  MakeProfiles long options on 32bit big endian systems (bug #51341).

  Pure rotation around pixel coordinate (0,0) (bug #51353).

  NoiseChisel segfault when no usable region for sky clumps (bug #51372).

  Pixel scale measurement when dimension scale isn't equal or doesn't
  decrease (bug #51385).

  Improper types for function code in MakeProfiles (bug #51467).

  Crashes on 32-bit and big-endian systems (bug #51476).

  Warp's align matrix when second dimension must be reversed (bug #51536).

  Reading BZERO for unsigned 64-bit integers (bug #51555).

  Arithmetic with one file and no operators (bug #51559).

  NoiseChisel segfault when detection contains no clumps (bug #51906).

  Correct size checking when allocating gal_data_t (bug #52544).



* Noteworthy changes in release 0.3 (library 1.0.0) (2017-06-01) [stable]

  This is a full re-write of Gnuastro. Most importantly, Gnuastro now has a
  new generic data container (`gal_data_t'). This new container can now
  deal natively with all standard numeric data types, work in RAM or
  HDD/SSD, keep data in any dimensions and has enabled many other very
  useful features. Some of the most prominent of the new features are
  discussed below. It is strongly recommended to review the respective
  section of the Gnuastro manual/book for a better feeling of all the new
  features.

  As discussed below, some program names have changed, if you have a
  previous version of Gnuastro installed from source, it is recommended to
  uninstall it first (with `make uninstall' using the corresponding
  tarball), then install this new version. Building Gnuastro can be slow,
  so please build in parallel with Make's `-j8' option (to build on 8
  threads).

** New programs or library features

  Library functions that deal with datasets now use this generic data
  container for inputs and outputs, significantly simplifying their
  API. Nearly all library functions have been re-written with much more
  clear names, argument lists and individual purpose. Some example library
  functions are shown below, also see the "Library demos" section of the
  book for some complete working example:

     -- `gal_table_read' and `gal_table_write' will read and write data to
        plain text, FITS ASCII and FITS Binary formats.

     -- `gal_fits_img_read' and `gal_fits_img_write' can read a FITS image
        to memory or write a FITS image from memory.

  Gnuastro now defines a simple comment line format to keep basic
  information in a plain text table, see the "Gnuastro text table format"
  section of the book. In short for every column, a comment line like below
  can be used to give a name, units, comments, or a type to a column. This
  allows a FITS binary table for example to be written to plain text and
  converted back to binary without loosing any information (except for very
  small floating point errors if not enough decimals are printed).

      # Column N: NAME [UNIT, TYPE, BLANK] COMMENT

  The new Fits program replaces the old Header program. But besides
  reading/checking FITS header keywords, it can now also work on FITS
  extensions/HDUs. For example with no options, it will list all the HDUs
  in a FITS file along with basic information. It can copy a whole HDU to
  another file, or delete a HDU from a FITS file. To get the previous
  behavior of listing all the keywords in a FITS HDU, you can run it with
  the `-p' option.

  All programs now write data into the second HDU of a FITS file to allow a
  clean first HDU. Note that following CFITSIO, HDU counting still starts
  from zero, so FITS images and tables written by Gnuastro in a new file
  can always be accessed with the `--hdu=1' option (which is now the
  default).

  If any program is run within a Git version controlled directory, a
  `COMMIT' header keyword will be added to the created FITS files, see the
  "Output headers" section of the book for a discussion on the usefulness
  of this new feature.

  BuildProgram: a new program to easily compile, link and run a C program
  you have written with Gnuastro's libraries without having to worry about
  which libraries (Gnuastro dependencies) your program needs. Debugging
  (`-g'), optimizations (`-O'), warnings (`-W'), include search path
  (`-I'), link search path (`-L'), and linked libraries (`-l') compiler
  options are also supported. BuildProgram will greatly facilitate the easy
  usage of Gnuastro's libraries.

** Removed programs or library features

  The following program names have been renamed:

    ImageCrop       --> Crop
    ImageWarp       --> Warp
    ImageStatistics --> Statistics
    Header          --> Fits

  SubtractSky has been removed. The Statistics program now has tools to
  estimate the Sky value and it can be subtracted with Arithmetic.

** New features

  All Gnuastro programs that read and write tables can now do so in plain
  text table format or in FITS ASCII or FITS Binary tables. Depending on
  the filename or with the new `--tableformat' common option to all
  programs.

  The option management system in all Gnuastro programs has been completely
  re-written with many new features, some of the most important ones are
  listed below. For developers, you will notice that there is no more usage
  of macros and adding new options has become much more easier.

     -- All programs will now also look for a `gnuastro.conf' configuration
        file to keep common options for all programs in every directory.

     -- The `--lastconfig' option can be used on the command-line or in any
        configuration file to stop parsing any further configuration files.

     -- The `--config' option can now be used to identify any arbitrary
        file to be parsed as a configuration file. Any file that is given
        to this option is parsed immediately.

     -- The `--printparams' option now also prints the short documentation
        of each option (same description in `--help') after its value.

  It is now possible to choose columns in tables based on column names as
  well as column numbers. It is also possible to search for columns based
  on searching in their units or comments. When using column numbers,
  counting now starts from 1 (one), not 0 (as before). See the new
  "Selecting table columns" section for more on these new features.

  Where relevant, all programs now accept a `--type' option that you can
  use to specify the numerical datatype of the output.

  With the new common option `--minmapsize', you can specify a minimum size
  of an array (in bytes) to store data in SSD/HDD and not in RAM. This can
  be instrumental when you are dealing with large datasets, or even smaller
  ones, but when your RAM is getting full.

  Making a log file is now optional and users have to explicitly ask for it
  with the `--log' option.

  Slower building of Gnuastro: Binary operators (e.g., plus or multiply)
  are now done in the native type of the input dataset. Doing so for all
  the different combinations of types, greatly slows down the initial
  compilation of Gnuastro (after running `make'). So for every type there
  is now a `--enable-bin-op-*' configure time option. When the dataset's
  type isn't compiled (only for the binary operators), it will be converted
  to a compiled type and then converted back in the end.

  Arithmetic: all operations are done in the native data type of the
  dataset. Until now, it would convert the data internally to double
  precision floating point, do the requested operation and write the data
  back in the proper type. So this new implementation is much more
  efficient.

  Arithmetic: two new classes of operators: type conversion operators to
  all standard types and integer-only operators (all C bit-wise operators
  and the modulo operator).

  ConvertType: can also print the input dataset to the command-line
  (`stdout'). To use this feature set the output filename to `stdout'.

  Convolve now has the `--minsharpspec' option to specify the minimum
  spectrum value to use in deconvolution (matching PSFs).

  Crop: when in WCS mode it can still only work on aligned images. However,
  very small floating point errors in writing the WCS (for example 10e-13
  degrees) are now acceptable. Until now these would cause Crop to complain
  and abort.

  Crop: the name of the crop can be pre-determined based on the values in a
  given in a table column. This allows your object's IDs to be directly
  used as the crop's file name for example.

  MakeCatalog now also reports the surface brightness in
  mag/arcsec^2. Until now, MakeCatalog would only print the dataset's
  surface brightness in units of magnitudes/pixel. But that is not nicely
  comparable to other datasets. Hence, using a simple calculation (from the
  pixel projected size, fully derived in the book), it now also reports the
  surface brightness in magnitudes/arcsec^2 also.

  MakeProfiles: Profile codes now start from `1' (until now they started
  from `0').

  MakeProfiles: now accepts the radial function of profiles as
  human-readable strings instead of a code for each profile (which was very
  cryptic, although codes are also still acceptable). For example in the
  profile column you can now write `sersic' instead of the code `1'.

  NoiseChisel: the new `--cleandilated' option will remove dilated objects
  that have a low S/N (it is mainly useful on very clean or mock images).
  For non-clean noise, it will result in a decrease of completeness. With
  this new option, NoiseChisel will also print detection S/N values when
  run with the `--checkdetsn' option.

  Statistics: now reads table columns as well as images and does basic
  operations on them. It can also only work on a certain range of the data
  (instead of the whole set). Alternatively, you can define the range on
  another reference column, but use values of the main column.

  Statistics: all its single-valued measurements can now be done on a
  tessellation (tile grid) over the input dataset.

  Statistics: can now estimate the Sky value on the input dataset using the
  mode's quantile similar to what NoiseChisel does to find its initial
  threshold.

  Statistics: has several new single valued calculations: `--quantile',
  `--quantfunc' (quantile function), `--mode', `--modequant', `--modesym',
  and `--modesymvalue'.

  Warp: align the image with the celestial coordinates using the `--align'
  option.

  Warp: standard modular warpings can now be requested without an input
  matrix, using the following options: `--shear', `--flip', `--project',
  `--rotate', `--scale', `--translate'. Any number of these transformations
  (along with the `--align' option) can be called on the command-line and
  they will be applied in the same order to create one warping matrix. By
  default the WCS will also be corrected.

** Changes in behavior

  Mask image options have been removed from all programs. Instead, all
  programs can work directly on data with blank values. So when some pixels
  must be masked, the Arithmetic program's `where' operator can be used to
  select special pixels and set them to blank. In particular bit-wise
  operations are now available in Arithmetic to use bit-mask
  images. Managing all these different choices in every program would only
  confuse the user (with too many options).

  Arithmetic: the `x' letter is now used to represent the multiplication
  operator. Previously it was `*' which needed quotation and was thus very
  inconvenient.

  Convolve: the old `--frequency' and `--spatial' options have been removed
  and are replaced by `--domain' which accepts values of `frequency' and
  `spatial'.

  Convolve: the old `--viewfreqsteps' was changed to `--checkfreqsteps' to
  fit with the general style of such check images in all Gnuastro's
  programs.

  Crop: `--section' syntax is now inclusive in both bounds.

  Crop: only checks if the center of a crop is filled when the crop was
  defined by its center (for example with `--ra' and `--dec'). The verbose
  outputs of Crop are also not cryptic 0s or 1s. The are human readable
  text.

  Crop: doesn't have separate `--imgmode' and `--wcsmode' options any
  more. There is now a single `--mode' option which accepts values of `img'
  or `wcs'.

  MakeProfiles: the old `--inputascanvas' is now called `--clearcanvas'.

  MakeProfiles: until now, it would abort with an error when the input
  columns had blank values. But for masking, it might happen that you set a
  blank magnitude. So this check has now been removed when reading the
  magnitude column.

  NoiseChisel: default value of the `--minskyfrac' option (new name for the
  old `--minbfrac') is now 0.7 as opposed to 0.5. This will allow much
  better estimation of noise properties (by default). It may be slightly
  too high for a crowded field, but the users can change it on the
  command-line (or in a configuration file) for such datasets.

  NoiseChisel: when it is run with any of the `--check' options, it will
  abort after all the check images have been created. This is very useful
  for checking your parameters until each step and not be distracted (or
  have to wait) for later steps to finish.

  Statistics: will not make a histogram and cumulative frequency files, or
  calculate sigma-clipped results by default (with no options). It will
  just print some basic information.

  Table: Previously, if a column was requested, the `-i' option would be
  ignored. But it often happens that the users forget a column name after
  already typing several of their desired columns. So the opposite behavior
  is preferred. Because when more than a couple of columns are needed, you
  will probably forget the column identifiers of the last few and having to
  retype everything is very frustrating. Something like how `--help' takes
  precedence over all other options.

  Table: to select column(s) by regular expression searching, the name now
  has to be put in `/ /' (similar to AWK). If a value isn't in `/ /', the
  programs will only select a column with the exact match.

  Warp: when a 2 by 2 matrix is given, the FITS pixel positions (which
  define the center of a pixel as an integer) are automatically implemented
  internally, see "Invoking Warp" in the manual for more.

  Warp: the old `--nofitscorrect' option has been changed to
  `--centeroncorner' to be more clear. The new option is now more general
  than before and also works on warping with a matrix, not just on modular
  warpings.

  Warp: the old `--nowcscorrection' option has been given a more clear name
  of `--keepwcs'. With this option, Warp will not apply the warp the
  input's WCS structure.

  Warp: the old `--maxblankfrac' option has been changed to
  `--coveredfrac'. Until now, Warp would only look for the fraction of
  input blank/NaN pixel area over the output pixel. But this would be
  useless on the edges of the image. So the new `--coveredfrac' option
  takes the acceptable fraction of output pixel area that must be covered
  by input pixels in order to give that output pixel a value. You can use
  this to set edge pixels that are not fully covered in the new grid to
  blank and have a flat warped image.

** Bugs fixed

  Using `%zu' to print `size_t' variables for clean build on 32-bit
  systems.

  Crash in Table for some operating systems due to memory is now fixed (bug
  #49347).

  Table's man-page is now created and installed (bug #49418).

  Fixes in the documentation (sr #109170, bug #49419).

  Check for malloc returning valid pointer (bug #49459).

  Segfault in mesh interpolation corrected (bug #49588).

  Corrected bad status usage in calls to wcsp2s and wcss2p (bug #49752).

  Stricter checking in Crop's polygon point list (bug #48978).

  Correction in alignment and getting pixel scale (bug #50072).

  Decomposing PCi_j and CDELTi matrices in output WCS (bug #50073).

  Using image naxes[n] when checkcenter is larger (bug #50099).

  Memory leak in MakeCatalog corrected (bug #51118).

  Fix copy-paste error in MakeCatalog flag arrays (bug #51130).





* Noteworthy changes in release 0.2 (library 0.0.0) (2016-10-03) [stable]

** Bugs fixed

  Linker errors on some operating systems have been fixed (bug #48076).

  Several memory allocation, checks or redundancies have been fixed: bugs
  #48453, #48516, #48603, #48611, #48571, #48650, #48657, #48692, #48770,
  #47866, #48899, #49049, #49007.

  ImageCrop no longer crashes with very long output file names (bugs #46241
  and #45380).

** New programs or headers

  Table: a new utility to read and write FITS binary and ASCII tables. It
  can also print the column information or select columns using regular
  expressions (task #13579).

  Shared libraries and headers are now installed. The libraries can be used
  in C and C++ programs. This release includes the following headers:
  `gnuastro.h', `array.h', `box.h', `fits.h', `linkedlist.h', `mesh.h',
  `polygon.h', `qsort.h', `spatialconvolve.h', `statistics.h', `threads.h',
  `wcs.h', `txtarray.h' (task #13765).

  Gnuastro now comes with a script in its top source directory
  (`tmpfs-config-make') to configure and build it in the tmpfs (on the
  RAM), for those systems that have it. See the new "Configure and build in
  RAM" section in the book for more (task #14100).

** New features

  MakeProfiles also accepts WCS positions (task #13566).

  Flat profiles in MakeProfiles can be given a profile specific value. The
  new `--mforflatpix' option MakeProfile will use the value in the
  magnitude column as a fixed value for each pixel. This can be very useful
  in defining a mask, or creating segmentation maps or labeled images (task
  #14115).

  MakeProfiles can now use input image as canvas. Instead of specifying the
  WSC and image size parameters manually. With the new `--inputascanvas'
  option, MakeProfiles will get this information (along with blank pixels)
  from an already existing image (task #14116).

  Type of output in MakeProfiles and Arithmetic can be specified with the
  `--type' option.

  Magnitude error column in MakeCatalog with the `--magnitudeerr' option.

  Arithmetic now has new conditional (task #13870) and logical operators
  (task #14153) along with an operator for actions only when conditions are
  true: `where'. The new `isblank' operator will also enable you to select
  blank, or masked, pixels (task #14146).

  The `--noerodequant' in NoiseChisel enables it to detect small and
  sharper profiles by disabling erosion on pixels above a certain quantile
  (task #14139).

  MakeCatalog can also calculate the upper limit magnitude for each given
  object in the image by randomly positioning the object's footprint over
  undetected regions of the image (task #14167).

  The source tarball is now also distributed with Lzip for a much better
  compression ratio and more robust archival file format.

** Changes in behavior

  The two MakeProfiles options `--setconsttonan', `--setconsttomin' have
  been removed (see `--mforflatpix' above for their alternative).

  MakeCatalog makes clump catalog only when asked (when the `WCLUMPS'
  header exists in the objects HDU). This can be very useful in cases like
  aperture photometry, when the object labels are not generated by
  NoiseChisel and so a clump image doesn't exist (task #14122).

  Default cosmological parameters in CosmicCalculator set to Plank 2015
  results: A&A (2016), 594, A13 (arXiv 1502.01589).

  The `--envseed' option (to read random number generator type and seed
  from the environment) to MakeProfiles and MakeNoise can also be given in
  the configuration files.





* Noteworthy changes in release 0.1 (2016-05-30) [stable]

** Bugs fixed

  MakeCatalog's problem in checking the sizes of all input images is now
  fixed.

  NoiseChisel's problem with reading the `--kernel' option is now
  corrected (bug #46750).

  lib/mesh.c's problem in correctly calculating the mesh sizes was
  corrected (bug #47611).

  `make check' will not look into system utility configuration files. In
  the previous release, if Gnuastro was already installed, the
  configuration files already present on the system would also be read. Now
  only configuration files in the tested package are used (bug #47833).

  Ghostscript's version is now checked at configure time after its
  existence. ConvertType uses the `-dPDFFitPage' option to Ghostscript
  which was introduced in version 9.10, so older versions would pass
  configure but at `make check' time, the PDF test would fail. Now this
  test is skipped (bug #47868).

  Most tests would fail when `make check -jN' was run (to do the checks on
  N threads). A dependency structure has now been defined to fix this
  problem and greatly speed up the testing process (bug #47957).

** New utilities

  Arithmetic: For arithmetic operations on the pixels of input images. With
  this utility, it is now possible to add multiple images with each other,
  or easily calculate a median image. It as a large set of other arithmetic
  operations and some functions which can be done on the input image(s). An
  unlimited number of input images can be given, the images will only be
  loaded when necessary and will be freed as soon as they are no longer
  necessary.

  CosmicCalculator: For doing cosmological calculations at a given
  redshift.

** New features

  All the utilities that would produce a log file now have a `--nolog'
  option to avoid printing a log file.

  The tiled image compression convention (.fits.fz, created with `fpack')
  files can now be used as input in the utilities.

  ImageCrop can now also crop a polygon from the input image. The polygon
  vertices can be given in the world or image coordinates. The simple
  `--polygon' option will keep the insides of the polygon while the
  `--outpolygon' will keep the outside of the polygon.

  ImageCrop and ImageWarp can now read the WCS information of a FITS header
  from a specific region with the `--hstartwcs' and `--hendwcs'. In some
  older FITS images, when the WCS distortions were not as standardized as
  now, there were cases which would confuse WCSLIB.

  NoiseChisel can now save the grown clumps image instead of the original
  clumps image in the output with the `--grownclumps' option.

  Convolve can now do deconvolution with the `--makekernel' option.

  MakeProfiles now has a `--setconsttonan' option which will fill the
  constant profiles with a NaN (blank) value, not a number, allowing the
  creations of elliptical masked regions for example.

  Header can now import a keyword directly from a string with the `--asis'
  option.

  MakeCatalog can now output the geometric (average position independent of
  pixel flux value) positions of the objects too.

  MakeCatalog can now produce the object's elliptical parameters (for
  example semi-major axis, semi-minor axis, and position angle). This can
  also be done both in standard flux weighted and geometric methods too.

  MakeCatalog now has a `--threshold' function to only use pixels above a
  given threshold in each object or clump. This is useful to avoid diffuse
  regions in calculations.

  MakeCatalog now has a `--noriverbrightness' option. With this option it
  is possible to calculate the clump flux without subtracting the river
  pixels on its circumference.

  The number of CPU threads is no longer a configuration option, it is now
  determined at run-time for each program. Therefore it is now easily
  possible to built Gnuastro on one system to use on another (commonly done
  in the GNU/Linux package managers). Therefore ./configure no longer has a
  `--with-numthreads' option.

  Every commit in Gnuastro's history (after implementing this feature) can
  now be given a unique version number. Since the version number is printed
  in possible outputs, this feature can help reproducibility, even when the
  an official/stable release isn't used.

  The AUTHORS file is now automatically generated from the version
  controlled history. Also all the authors that have contributed to
  Gnuastro are included in the second (copyright) page of the PDF book.

  All the bootstrapped directories are now moved with a new `bootstrapped'
  directory in the top source directory. This significantly cleans up this
  directory, allowing users to more easily find the hand-written Gnuastro
  source files they like.

  A `bug-gnuastro' Info page was created so users can easily go to that
  page for information on how to submit bug reports. It is accessible on
  the command line with the command `info bug-gnuastro'.

** Changes in behavior

  The separate utilities no longer have a separate version number. With the
  introduction of unofficial version numbers generated from each commit in
  Gnuastro's history, the utility version numbers would cause confusion.

  MakeProfiles will now add a suffix to the individual images and put them
  in the output directory if specified.

  NoiseChisel and MakeCatalog now use the median mesh standard deviation to
  define the over-all depth of the image. Previously they used the maximum
  value (or the least depth).

  NoiseChisel no longer outputs a sky subtracted image. This job can now be
  done with the new Arithmetic utility.

  NoiseChisel's `--segsnhistnbins' option was renamed to
  `--clumpsnhistnbins'.

** Improvements

  NoiseChisel's default quantiles were changed to 0.95 from 0.99. Since the
  old value was too severe.

  NoiseChisel's S/N thresholds are now found from the distribution of
  pseudo-detections and clumps from the full image, not within each large
  mesh. This was done to increase the accuracy of the S/N threshold. There
  were commonly not enough points in large mesh sizes and this would add
  scatter. With ImageCrop's polygon capabilities, it is now easily possible
  to cut out the region that has uniform noise properties (depth and
  correlated noise). Therefore the old `--checkdetectionsn' and
  `checkclumpsn' options are no longer present.

  When building from the version controlled source, the whole bootstrapping
  process is done with one script. In the previous version, all the
  separate operations should have been done by hand (as instructed in the
  old manual).

  All the build steps now report what was done and suggest the next
  step. This feature can be disabled with the `--disable-guide-message' at
  configure time.





* Copyright notice
Copyright (C) 2015-2019 Free Software Foundation, Inc.

Permission is granted to copy, distribute and/or modify this document under
the terms of the GNU Free Documentation License, Version 1.3 or any later
version published by the Free Software Foundation; with no Invariant
Sections, with no Front-Cover Texts, and with no Back-Cover Texts.
