2011-02-20 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- magick/magick.c (MagickPanicSignalHandler,MagickSignalHandler): Don't invoke DestroyMagick() since there may be OpenMP worker threads still running which are using data which would be deallocated. Instead we invoke PurgeTemporaryFiles() to remove any existing temporary files. Valgrind will report leaks if the program is terminated by a signal but this causes no actual harm. Resolves SourceForge issue 3165456 "^C causes semaphore failure in MacOSX". (MagickPanicSignalHandler): Invoke abort() in panic signal handler so that we will reliably get a core dump.
- magick/tempfile.c (PurgeTemporaryFiles): New private function to remove any existing temporary files but without destroying temporary file semaphore.
2011-02-13 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- coders/pnm.c (ReadPNMImage): Fix mis-placed break in PAM header parser.
- wand/magick_wand.c (MagickWriteImageBlob): Improve the documentation to mention the related use of MagickSetImageFormat() and MagickResetIterator().
2011-02-07 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- magick/xwindow.c (MagickXBestFont): Check for a few more common font names, and ensure to always check for "fixed" as a final fallback.
2011-02-06 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- config/delegates.mgk.in: Added gs-cmyk entry. Used if '-type ColorSeparation' is specified on the command-line prior to the PDF or Postscript file name. This entry specifies use of the Ghostscript PAM driver which is capable of supporting CMYK output. This may be useful if it is desired to apply CMYK color profiles to the image returned from the PDF. As fair warning, it seems that Ghostscript 8.62 outputs CMYK even if the PDF was in RGB space if the PAM driver is used.
- coders/pnm.c (ReadPNMImage): Add support for reading netpbm's PAM format.
2011-02-01 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- tests/rwblob.c, tests/rwfile.c: Fixes to help tests work when testing with multiple frames.
- coders/sgi.c: SGI format is not documented to support multiple frames. Remove the half-baked extension for it.
2011-02-01 Glenn Randers-Pehrson <glennrp@simple...>
- coders/bmp.c (ReadBMPImage): Changed file_size greater than expected from a corrupt-image error to a debug log entry. File_size too small is still an error, and made that so also for BI_RGB images which were previously exempted from the test.
2011-01-30 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- tests/rwblob.c, tests/rwfile.c: Validate the data in each image frame, validate that each read returns the same number of frames, and validate that the correct number of frames was ultimately returned.
- magick/blob.c (SyncBlob): Disable bogus code which attempted to replicate the blob I/O object across all images in the list when the blob is synced. Leave a less bogus bit of code in place (but commented out) in case such functionality is deemed to actually be needed in the future. The previous code was copying structs on top of each other, including a pointer member to a semaphore.
2011-01-23 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- www/Changes.rst: Add a new Changes page to wrap up the yearly change logs to lessen download size.
- scripts/changelog2rst.sh: Simple utility to format ChangeLog format into something resembling reStructuredText.
- www/Makefile.am: Use reStructuredText to format the ChangeLog files to HTML so that we can inherit the improved formatting and page style.
- coders/pnm.c (ReadPNMImage): Support for multi-frame PNM was botched due to on-going edits to support PAM format.
2011-01-16 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- wand/magick_wand.c (MagickDescribeImage): Was sending descriptive output to stdout rather than returning it in an allocated string as intended.
2011-01-09 Bob Friesenhahn <bfriesen@simple.dallas.tx.us>
- magick/draw.c, wand/drawing_wand.c (MvgPrintf): Update to handle C99 vsnprintf() return values.
- magick/draw.c, wand/drawing_wand.c (DrawAnnotation): Linux glibc does not pass extended text characters if "%.1024s" formatting convention is used. Apparently it assumes that such characters may be UTF8 and returns -1 rather than outputting the string, even if it is assured to fit.