in libmpeg2/motion_comp_arm_s.S have addresses in .text, which is bad
for shared libraries.  Some environments demand that .text actually be
read-only all the time, yet MC_put_o_16_arm etc require that the addresses
be modified by the dynamic linking mechanism (dlopen, LoadLibrary, etc.)
Even in those environments which permit the dynamic linker to modify the
.text segment, the runtime cost of doing the relocation can be noticeable.

This commit rewrites the linkage, discarding the tables of addresses
in favor of tables of offsets.  All transfers are local within each individual
function, so there can be no interference by processing that occurs
after assembly, such as link-time re-ordering (even of individual functions.)

Patch by John Reiser <jreiser@bitwagon.com>

configure option, due to AC_TRY_COMPILE not declaring arguments (not even void)
for the main() function it generates.

want header.c to manipulate the decoder structures directly.

The idea is that we don't want header.c to know that level of detail
about the decoder's internals.

have header.c poke into the mpeg2_decoder_t structure nowadays.

CPU intensive part of libmpeg2) from the rest of the decoder
(i.e. the chunk gathering, header parsing and state machine).

Eventually I want to get to a point where decode.c and header.c do not know
about the internals of the mpeg2_decoder_t type. Longer term, I want to
be able to switch the slice decoders, for example to support GPU acceleration.

The FSF changed headquarters a long time ago.  Here is a patch to update
the license headers in libmpeg2 to reflect that change.  It also
includes a similar change for the COPYING file.  I just replaced yours
with the copy in /usr/share/common-licenses/GPL-2 on my Debian system.

they depend on the caller knowing implementation details of the
mpeg2_decoder_t type. Move the function declarations to mpeg2_internal.h...

* Allocate aligned buffers using mpeg2_malloc(), to avoid reimplementing
  this in the samples (and demonstrate usage - mpeg2_malloc() is public
  just so people can do this)

* In a few places, remove the local 'sequence' variable and replace with
  info->sequence - the main reason here being to limit differences between
  the various code samples.

This is both to test that they do compile, and to give
easy access to the binaries for running them.

Patch from Diego Biurrun

  * configure.ac: remove include constructs because aclocal isn't clever
    enough to look for missing macros in these files.

    of DIST_SUBDIRS) and putting them back in $(includedir)/mpeg2dec (instead
    of $(includedir)/libmpeg2).

   regression tests) ;
 * ChangeLog, NEWS: sync with the 0.4.1 branch, and prepare for the 0.5.0
   release ;
 * configure.ac, vc++/config.h: bump up revision to 0.5.0.

    have been too many breakage reports.

    they are doing.

    remaining compatible with the previous interface.

    the empty argument list.

  * configure.ac Makefile.am: hide autotools files in .auto, only recurse
    "make" into directories that build stuff, and replace INCLUDES with
    more modern AM_CPPFLAGS.

   rates, spotted by Craig Milo Rogers.

   cleaned up by Diego Biurrun and Alexander Strange.