[Magick-announce] ImageMagick 5.5.4 released

[psyche at studio.imagemagick.org]

ImageMagick 5.5.4 is primarily a bug fix release as detailed below. The
release is currently mirroring and should be available at all the standard
archive sites by tommorrow.  See

  http://www.imagemagick.org/www/archives.html

Bug fixes and enhancements include: 

  * Temporary cache files were not being removed under Windows.
  * Fixed a small memory leak in magick/random.c.
  * Pixel area computations in the cache changed from unsigned long to an
	  extended signed integral type.
  * TransformHWB function sometimes failed due to rounding errors (patch by
    Artur Zaprzala).
  * SVG => MVG viewbox was not computed correctly (patch by Artur Zaprzala).
  * Added the -convolve option to the convert and mogrify programs.
  * Some DCM grayscale images did not display correctly.
  * An incorrect conditional compile define prevented the TIFF coder from
    writing JPEG-compressed TIFF image files.
  * -geometry with offsets did not resize properly (bug report by Bob
    Friesenhahn).
  * Circles no longer render with an artifact at 360 degrees.
  * Added -list resource to the convert/mogrify/identify programs.
  * Recognize new gimp xcf image file magic.
  * Merged geometry.h with image.h to prevent Visual C++ build errors.
  * Set image colorspace to RGB when profiles transform CMYK => RGB.
  * The ps, pdf, and ept coders now note the page offsets of any embedded
    bounding box.
  * The TIFF coder now handles colormapped images with a alpha mask.
  * +page correctly resets image page info (bug report by Glenn).
  * Corrected (x,y) offset computation in RotateImage().
  * ImageMagick coder files are now stored in lib/ImageMagick/?.?.?-Q?/coders
    (previously stored in lib/ImageMagick-?.?.?/modules/coders).
  * Corrected (x,y) offset computation in RotateImage().
  * Minor mods to blob.c, studio.h, and CoderInfo.cpp to enable a Borland
    C++ build (patches by Mike Chiarappa).
  * Properly detect POSIX threads on Tru64 (patch by
    china at thewrittenword.com).
  * QuantumDepth 32 produced non-compliant XPM images (bug report by
    Bob Friesenhahn).
  * -trim (and -crop 0x0) now returns the correct image offsets.
  * Reading the DPX header was off by 4 bytes.
  * Compaq TruUnix doesn't like random.h-- moved defines to signature.h.
  * SetImageType(...,PalleteType) did not set the storage class properly.
  * Changed from ftime() to gettimeofday() and clock_gettime() for better
    nanosecond handling for the random number generation.
  * Moved all image/option geometry methods to a new module, geometry.c.
  * Added method ParseGeometry() and replace most scanf() with this method.
  * 'identify' recognizes the -list option (suggested by Glenn
    Randers-Pehrson).
  * Use the ShadowFacter rather than ShadowModule define in RaiseImage()
    (bug report by dbryson at techass.com).
  * Added member error_number to the ExceptionInfo structure to note
    the errno at the time an exception is thrown.
  * Optimized the cloning of the pixel cache.
  * QuantizeImage() no longer hangs for a quantum depth of 32.
  * 'identify' improperly returned an error for PCD images.
  * Deprecated QuantizationError(), use GetImageQuantizeError() instead.
  * Added GetImageDynamicThreshold() method to segment.c (invoked with
    convert option -threshold 0).
  * Fixed bug in png.c, introduced on 12/18/02, that caused assertion
  * Updated TransformRGBImage() and RGBTransformImage() to use the MaxMap,
    ScaleQuantumToMap() and ScaleMapToQuantum() macros.  Previously all
    computations were scaled unnecessarily to [0..65536].
  * A hack allowed TransformRGBImage(image,RGBColorspace) to convert
    the image to the RGB colorspace.  The hack was removed and we now
    correctly use TransformRGBImage(image,image->colorspace) to perform
    the conversion.
  * Fixed an an initialized filename problem in MogrifyImage().
  * Previously RGBTransformImage() assumed the image colorspace was RGB
    (bug report by Bob Friesenhahn).