NOTE THAT THERE HAVE BEEN SOME MAJOR CHANGES IN THE NAMING CONVENTION OF THE 
	FUNCTION CALLS WITHIN THIS RELEASE. THIS WILL AFFECT ALL PROGRAMS
	CURRENTLY USING PANDA... READ ON FOR DETAILS...

Release 0.3 -> Release 0.4
  * Imagebox file not found error messages for TIFFs improved (filename that 
	was bad is now included)

  * Discovered a wierd bug in acroread (I think). See huh.pdf... This seems to
	be acrobat related. It doesn't happen in gv.

  * PDF doesn't support / as a character in the name of an object such as an 
	image. Therefore when we add an image into a document we now have to 
	go through the name string and change all of the /'s to -'s. Then 
	again, this might not be a problem with PDF itself, but is certainly 
	a problem with the lexers in xpdf, gv and acroread and therefore is 
	an issue.

  * Moved all the headers into panda/, which means the demo app can now use 
	the same path for the includes as the real applications that users
	would be writing. This means that the demo application had to be
	renamed to pandademo (from panda).

  * extern "C"'ed all of the headers so that they are accessible from C++

  * Defined #defines so that headers are only included once at compilation
	time

  * Fixed the Makefile.am so that the man pages are installed as well...

  * Discovered that the Makefile.am needs to be changed for that the headers
	will install from their new location in panda/, so I had a look into
	that as well...

========================== 0.3.1 Released =====================================

  * Implemented xmalloc in memory.c so that I could simplify the malloc calls
	in the rest of the code in the library... Also changed all of the 
	malloc calls in the library to use it...

  * Resolved the namespace problems by renaming all function calls to use
	the naming scheme "panda_<<>>". There are #defines to the old names in
	panda/legacy.h

  * Note that the namespace changes were partially made by a script, so if 
	 there are any typos, could you please let me know... Things still
	 compile, so at least the typos are consistent.
	
  * Implemented template pages for PDF documents. These are created via the
	panda_newpage() call with a page size of gPageSizeTemplateA4. They can
	then be asserted on a page using panda_asserttemplate().

  * Fixed the namespace on variable names as well... This included moving some
	stuff out of the gcc reserved namespace (__*) in objects.h.

  * The namespace changes were lots of fun (not). I suggest that no-one ever
	do them again...

  * Added legacymain.c to check that the legacy function mappings and stuff
	actually work

  * BACKPORTED FROM PANDA 0.3.2: Fixed some small problems with the
	configure script, added the Makefile.am to the distribution and made
	some small changes requested by the users...

  * Started rearranging the demo document so that it shows the features of
	Panda in a less confusing manner.

  * If you failed to setup font details before writing a string, Panda would
	crash. It now works on reasonable defaults.

  * Made dmalloc stop complaining about freeing NULL pointers, I assume this 
	makes the code more portable.

  * The saga of the vertical line in JPEG images (reported by dave@valinux.com)
	- a vertical line appears down the LHS of JPEG images embedded by Panda
	- visible in: acroread (linux)
	- not visible in: acroread (win32) gv (linux) xpdf (linux)

	As a sanity check, I created a PDF using acrobat on a win32 box of just
	the JPEG image in question... I then compared the streams. They are 
	definitely different, but it would appear that acrobat is converting
	the image into a TIFF stream before insertion, possibly because of this
	sort of problem.

  * Uncompressed TIFF images may now be inserted into PDF documents.

  * Grayscale TIFF images now can be embedded as well...