Removed leftover midi code. io-audio doesn't officially support midi
so no point in leaving it in here.

Removed leftover eclipse project files and build scripts

Add support for QCC compiler

This patch adds a small waf tool to add support for the QCC compiler,
which has a somewhat different command line syntax from GCC.  It also
fixes some small C++ issues which arise when trying to compile with the
QCC C++ library.

Change-Id: Id6ee4284095198bff206e45aa0e9a5fa5a2f8b74

Fix io-audio Integration Issues

Add systemd support

This patch adds optional support for systemd, by adding use of the
sd_notify() function to allow jackd to report when the daemon has
started and stopped.  This is similar to the Apple notification
support which is already present.

The JACK ring buffer requires that a buffer be a power of two in size.
This is not possible if the number of voices is not a power of two, so
I have removed the JACK ring buffer from the io-audio JACK driver. I
have replaced it with use of the raw DMA buffer along with a current
fragment value to track where we should be reading from the DMA buffer.

snd_pcm_read() was failing because we were not attempting to read a
multiple of the fragment size.  This patch fixes that so we read a
whole fragment at a time instead of the fragment size / frame size.

Fix io-audio JACK Driver Port Name Convention

Generally, JACK input ports are called "capture" and a number, while
output ports are called "playback".  The io-audio JACK driver for QNX
looks like an io-audio playback device, but to JACK, it looks like a
capture device.  So I have updated the driver to call its ports
"capture" rather than "playback".

Fix io-audio JACK Driver Playback with Fixed Point

When playing a fixed-point wave file, the JACK io-audio driver was
only playing every fourth sample.  This was due to a math error in the
io-audio driver code that was incrementing by the sizeof a sample
rather than just incrementing by 1.

Qnx

Removed a number of commented sections
Made a few sample types explicitly sized
Removed debugging statements

More QNX build fixes

This patch adds a missing case to the tests wscript, to handle the QNX
platform.  It also adds code to the setup for the various directory
macros, to handle the case when the build is taking place on Windows,
to ensure that the directory paths still come across in POSIX format.

Change-Id: If8054284d390e40c97c8c585af2851e94275aad8

Adds a driver for QNX io-audio system that acts as a JACK client.

Adds a JACK internal client to interact with QNX io-audio system.

Adds enough support and platform-specific files to get JACK building
on QNX SDP 6.6.0. This does not include any integration with the QNX
audio system, io-audio.

The expression tries to take the absolute value of an unsigned
(jack_nframes_t) which is the difference between two unsigned.

Cast the unsigned to signed before taking the difference so a
cur_buffer_size larger than delta_time actually produces a meaningful
result.

struct timeval is defined in <sys/time.h>. Including is explicitly here
fixes a compile error on QNX.

The program is C++ but includes C headers and uses extern "C"
clauses around the ExternalMetro class.

According to the posix standard, a named semaphore needs to begin with a
slash in order to guarantee the same semaphore will be opened by that name
by multiple processes.

This is an issue on QNX where a named semaphore that does not begin with a
slash has the current working directory prepended to the name as a kind of
namespacing.

jack_ringbuffer_reset_size changed the tracked size of the internal
buffer but didn't actually resize the buffer itself, potentially
leading to buffer overflows.

Forwarded from
<https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=728710>

Use _WIN32 instead of WIN32 in header files since WIN32 might not alw…

Implement new jack_port_rename API

JACK2 added this function first, but the int return has been always wrong.
When JACK1 added JackPortRenameCallback it used the proper return.

Now that JACK1 supports port renames, devs will start to use it.
(previously it didn't work properly because of the missing jack_client_t* arg)

Some code might be broken because of this, but it's a very simple change,
and existing code would have been broken when changing JACK1 to JACK2.

Finally, JACK2 code never uses the int return value of this callback.
So there's no real reason to NOT change this.

For instance, if compiling with gcc and the option --stc=c++11, WIN32 is not defined.
See http://webcache.googleusercontent.com/search?q=cache:amvfgAUCdsgJ:sourceforge.net/p/mingw-w64/mailman/message/31850993/+&cd=4&hl=en&ct=clnk&gl=no
http://sourceforge.net/p/mingw-w64/mailman/message/31850993/

Correct netmanager latency reporting

Fix buf nullptr dereferencing in jack_midi_event_write()

[jack_bufsize] Fix off by 10 error

Since help and error messages suggest, that 8192 samples is a valid
buffer size, this fixes the value in the check to 8192 instead of 8182.

Fix return value of SetTimebaseCallback()

return value of JackGenericClientChannel->SetTimebaseCallback() instead
of -1 where there is a conditional request