Various macOS fixes

See merge request !1

Currently this works-around lack of M_PIl and PATH_MAX which are
both not available on Darwin.

git-svn-id: https://svn.xiph.org/trunk/postfish@19176 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@19175 0101bb08-14d6-0310-b084-bc0e0c8e3800

Explicitly request a 50ms buffer latency from libao.  This is about double the current default in ao, but Pulse (surprise!) randomly refuses to work properly with a 20ms latency.



git-svn-id: https://svn.xiph.org/trunk/postfish@18785 0101bb08-14d6-0310-b084-bc0e0c8e3800

Don't write update requests to the UI pipe when there is no UI; it will fill and block.  In fact, don't open the UI pipe at all.
Don't install a SIGINT handler for batch mode, since the ^C handling was depending on the gtk mainloop


git-svn-id: https://svn.xiph.org/trunk/postfish@18784 0101bb08-14d6-0310-b084-bc0e0c8e3800

Does automatic device discovery for OSS, ALSA and Pulse



git-svn-id: https://svn.xiph.org/trunk/postfish@18782 0101bb08-14d6-0310-b084-bc0e0c8e3800

applied to any input without popping the UI.



git-svn-id: https://svn.xiph.org/trunk/postfish@18490 0101bb08-14d6-0310-b084-bc0e0c8e3800

low levels


git-svn-id: https://svn.xiph.org/trunk/postfish@18220 0101bb08-14d6-0310-b084-bc0e0c8e3800

compression/expansion together for all channels.



git-svn-id: https://svn.xiph.org/trunk/postfish@17366 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@14732 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@12241 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@10418 0101bb08-14d6-0310-b084-bc0e0c8e3800

Anyway, it both compiles and works now.

Monty



git-svn-id: https://svn.xiph.org/trunk/postfish@8773 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@8588 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@8587 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@8582 0101bb08-14d6-0310-b084-bc0e0c8e3800

In previous versions of Postfish, the stereo reverb output of a
channel reverb had to be explicitly mixed into an output channel using
the direct mixdown blocks or the reverb is inaudible.  You needed to
use the direct mixdown blocks to get reverb even when otherwise only
using the stereo crossplacer to produce a mixdown.

As of now, when using the Crossplacer in a mixdown panel, a channel's
reverb A output will be mixed directly into all active 'output A'
outputs, and the channel's reverb B output likewise mixed directly
into all active 'output B' outputs.  It is mixed according to the
master delay/attenuation sliders on the attenuation panel, but it does
not use the delay/attenuation settings of the crossplacer; this way
the channel can be mixed anywhere into a stereo image and the reverb
will still present the same image of space.



git-svn-id: https://svn.xiph.org/trunk/postfish@8581 0101bb08-14d6-0310-b084-bc0e0c8e3800

Handle this situation properly by having the sigint handler submit a
termination request via the event queue for the UI thread to handle in
proper order.



git-svn-id: https://svn.xiph.org/trunk/postfish@8580 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@8579 0101bb08-14d6-0310-b084-bc0e0c8e3800

Replace existing SWH reverb unit with an extension of Freeverb3



git-svn-id: https://svn.xiph.org/trunk/postfish@8578 0101bb08-14d6-0310-b084-bc0e0c8e3800

deal properly with displaying peak followers.  Fixed.



git-svn-id: https://svn.xiph.org/trunk/postfish@8036 0101bb08-14d6-0310-b084-bc0e0c8e3800

indication is potentially useful



git-svn-id: https://svn.xiph.org/trunk/postfish@8035 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@8034 0101bb08-14d6-0310-b084-bc0e0c8e3800

the companders.  

a) Soft knees are now computed as a freebie during filter following.
No explicit soft-knee means no sqrt() means considerable optimization

b) streamlined the computation pathway of the followers, partially as
an optimization, partially to prepare for adding the scond tier
over/under filters before release

c) found a bug in the behavior of the new peak-follower math (that was
instigated to repair a more serious bug in the original version);
decay was still using a second order bessel which could still ring
about zero.  Restored the peak-hold behavior to near-original
behavior, but now use the same first-order-decay-limited filters as
the RMS follower.

Depending on platform and compander usage, these changes are good for
getting back 10-25% of the CPU.



git-svn-id: https://svn.xiph.org/trunk/postfish@8033 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@7982 0101bb08-14d6-0310-b084-bc0e0c8e3800

playback device that doesn't exist.



git-svn-id: https://svn.xiph.org/trunk/postfish@7981 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@7967 0101bb08-14d6-0310-b084-bc0e0c8e3800

Fix several average-slider bugs in multicompand



git-svn-id: https://svn.xiph.org/trunk/postfish@7966 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@7955 0101bb08-14d6-0310-b084-bc0e0c8e3800

git-svn-id: https://svn.xiph.org/trunk/postfish@7953 0101bb08-14d6-0310-b084-bc0e0c8e3800

reverb was not being reported correctly to the display.



git-svn-id: https://svn.xiph.org/trunk/postfish@7952 0101bb08-14d6-0310-b084-bc0e0c8e3800

and deverber

Correct a bug where the metrics calculation was being run in freq.c
even when metrics were not being displayed.

Disable default soft-knee.



git-svn-id: https://svn.xiph.org/trunk/postfish@7941 0101bb08-14d6-0310-b084-bc0e0c8e3800