Recent changes to patches

RPM .spec for mpgtx 1.3

Andrew Ziem — Thu, 30 Mar 2006 14:11:57 -0000

RPM .spec for mpgtx 1.3 which works on Fedora Core 4
with GCC 4.0. The sed stuff fixes some build problems.

infinite loop when demuxing a truncated file

Stephen Beahm — Thu, 24 Mar 2005 16:55:17 -0000

When demuxing a file that has been truncated, mpgtx
goes into an infinite loop. This is a quick way to fill
up a hard drive ;)

This patch tests for an end-of-file condition, issues a
warning and breaks from the loop.

I chose to write the partial chunk before breaking from
the loop.

MPEG audio bugfix

Matthew Lieder — Fri, 11 Mar 2005 00:50:00 -0000

When using mpgtx on MPEG-2 videos created by
SnapStream BeyondTV3's software encoder, the
resulting videos often have problems with the audio
being out of pitch. I discovered that the problem was
partly because of two audio packets being between the
system header and the first video packet; copying those
two audio packets to the edited video fixed the
problems. Attached is a patch file containing that
copying code.

MPEG video metadata copying

Matthew Lieder — Fri, 11 Mar 2005 00:45:35 -0000

SnapStream BeyondTV3 (not sure about any other
programs) stores metadata in MPEG-2 videos, inside a
Private Stream 2 (0xBF) packet. The attached patch file
contains changes that allow mpgtx to copy that
metadata to the edited video it outputs.

CVS: mpgtx can write the ouput

Evan Meakyl — Mon, 31 Jan 2005 10:32:30 -0000

With the cvs source from 2005/31/1 at 11am, mpgtx can't
write its outputs when creating an mpg file... Here is
the fix (some tests were doing the opposite of what
they were supposed to do):

Index: mpegOut.cxx

RCS file: /cvsroot/mpgtx/mpgtx/mpegOut.cxx,v
retrieving revision 1.30
diff -r1.30 mpegOut.cxx
409c409
< if ( fwrite(&high, 1, 1, MpegOut) ) {
---
> if ( fwrite(&high, 1, 1, MpegOut)!=1 ) {
413c413
< if ( fwrite(&low, 1, 1, MpegOut) ) {
---
> if ( fwrite(&low, 1, 1, MpegOut)!=1 ) {

Bugfixes

Matthew Lieder — Sat, 22 Jan 2005 00:11:43 -0000

This patch enables mpgtx to handle weird system
headers (specifically the headers of WinTV PVR-
250/350/etc... MPEG-2 recordings), fixes some bugs
which caused the duration of MPEG-2 files to be
reported incorrectly, and fixes some bugs with properly
reading and writing timestamps from GOP's.

A couple fixes might be considered hacks, but so far my
changes have worked properly on every MPEG-1/MPEG-2
file I've tested. Please let me know if you encounter any
new problems because of these changes.

I've also only tested these changes on Windows, in
conjunction with my Win32 patch. Linux testers would
be appreciated.

OS/2 executable not built out of the box

Dave Yeo — Wed, 19 Jan 2005 05:10:49 -0000

On OS/2 we end up with an a.out binary instead of an
executable. The simplist fix is much the same as the
cygwin fix. Add this code to configure, perhaps under
the test for cygwin

# OS/2 support
if [ "${HOSTNAME}" = "i386-pc-os2-emx" ]; then
OFLAGS="-Zomf"
EXEEXT=".exe"
fi

The -Zomf is only important for a debug build, causes
the object files to be in omf format instead of a.out.
Debugger only supports omf
Dave

Win32 patch w/ LFS

Matthew Lieder — Thu, 13 Jan 2005 20:13:33 -0000

A Win32 port with some changes to add 64bit file i/o.

Also includes solution and project files for building under
MSVC++ .NET 2003.

Fix for GCC detection in configure

Wed, 27 Nov 2002 17:12:06 -0000

I'm attaching a fix to the gcc detection in configure.
The way it's being done now, is not tolerant to any
changes in how GCC reports the version. This patch is
independent of display, it uses "gcc -dumpversion"
which should work on any gcc >=2.

My patch basically changes:
gcc_major=`gcc --version | cut -b 1`
to
gcc_major=`gcc -dumpversion | cut -b1`

and it also fixes the test line which doesn't appear to
work (only a single '=' sign is present)

if test "$gcc_major" == "3"; then
optimization_cflags="-O2";
fi

is what it should be.

After this change, mpgtx compiles fine with gcc >= 3
and gcc 2.

And yes, this is a patch against mpgtx 1.3 which still
exhibits the problem.

This should close the bugs #642694 and #641185

Set manpage path installation directory

Eric Sokolowsky — Tue, 08 Oct 2002 18:17:38 -0000

I submit this patch to fix the manual page installation
for the configure script.

Below is a patch that will correctly install the manual
pages into the correct directory. If no --prefix is
given the manual path defaults to the same as the
regular installation directory. If the --prefix option
is given, then the manual page prefix becomes the new
installation path prefix. If the new option
--manprefix is given, then the manual path will be set
to the new option.

I think that this patch is important because it results
in more standard behavior. I want all of the files to
fall into a particular hierarchy (especially a local
directory when I might not have permission to install
man pages in /usr/local which is the default) without
having to edit the make file after I run the configure
script. This patch makes this possible.