This fixes an issue with some Kate tracks not playing with liboggplay,
reported by both Viktor Gal and ogg.k.ogg.k

Adapated from patch by ogg.k.ogg.k

Adapted fix for regression introduced in 8c2da1

When using a recent liboggz, liboggplay fails to play some files. It
appears it's only files with Kate streams, however some files with
Kate streams do play fine, but liboggplay reports unknown
category/language, so that's probably still failing, but in a way that
liboggplay can recover from.

A sample file which worked before and does not work now is the good old:
http://people.xiph.org/~oggk/Stephen_Fry-Happy_Birthday_GNU-hq_600px_780kbit.ogv
This file is seen as correct by oggz-validate.

Issue:

* when oggz_read() sees the first few packets before it sees a
granulepos, it buffers them in a dlist
* then, oggz_read_deliver_packet() is a dlist callback which goes
through these buffered packets and calls the read callback for each
* the commit you found by bisection returns an error if the
read_callback returns non-zero; previously the read_callback return
values were ignored
* liboggplay's read_predetected(), the read_callback in use during
stream detection, returns 1 (OGGZ_STOP_OK) when it has done what it
needs to do

... hence, the STOP_OK value gets turned into an error value by the
bad commit.

Reported-by: ogg.k.ogg.k@googlemail.com
Signed-off-by: Conrad Parker <conrad@metadecks.org>

Patch by timeless

When implementing anything listed in this file, please update it by deleting
that entry, and include that as part of the commit that implements the fix.

This commit changes the policy of fs_comment_validate_byname() to
allow NULL-value comments, and changes fs_comment_new() to handle
the case where value=NULL, and where the comment has no value but
has an = sign.

This commit should pass "make check", including the policy change
introduced in the previous commit.

This test changes the policy on what comments are valid, to allow
plain comments, not of the form KEY=VALUE (the condition reported
in Mozilla #520500).

As this commit only modifies the test suite, 'make check' is
expected to fail.

Modifies packet buffer dlist cancellation to use oggz_packet

Adds oggz_dlist cancellation, and uses that to bail out during
packet delivery if eg. header data is missing.

script

The problem is the condition used in the configure script for choosing whether
use a version script or not:
  case "$target_os" in
    linux* | solaris*|k*bsd*-gnu*|gnu-*)
given $target_os is "gnu0.3" for GNU/Hurd, the condition is not valid
for it.
Signed-off-by: Conrad Parker <conrad@metadecks.org>

script

The problem is the condition used in the configure script for choosing whether
use a version script or not:
  case "$target_os" in
    linux* | solaris*|k*bsd*-gnu*|gnu-*)
given $target_os is "gnu0.3" for GNU/Hurd, the condition is not valid
for it.
Signed-off-by: Conrad Parker <conrad@metadecks.org>

Patch by Matthew Gregan

and alphebetize