the diff was forgotten, it's discussed here:
http://marc.info/?l=openbsd-ports&m=145505706611958
It's about playing 24-bit files with full precision (assuming audio
sub-system supports it), i.e. without converting samples to 16-bit
first and loosing 8-bits of precision.
OK?
Index: Makefile
===================================================================
RCS file: /cvs/ports/audio/sox/Makefile,v
retrieving revision 1.65
diff -u -p -r1.65 Makefile
--- Makefile 12 Sep 2016 22:20:36 -0000 1.65
+++ Makefile 14 Sep 2016 06:22:03 -0000
@@ -5,6 +5,7 @@ BROKEN-hppa = bend.c:155:12: internal co
COMMENT= Sound eXchange, the Swiss Army knife of audio manipulation
DISTNAME= sox-14.4.2
+REVISION= 0
SHARED_LIBS += sox 4.0 # 3.0
CATEGORIES= audio
Index: patches/patch-src_sndio_c
===================================================================
RCS file: patches/patch-src_sndio_c
diff -N patches/patch-src_sndio_c
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-src_sndio_c 14 Sep 2016 06:22:03 -0000
@@ -0,0 +1,25 @@
+$OpenBSD$
+--- src/sndio.c.orig Mon Jan 30 04:01:44 2012
++++ src/sndio.c Tue Feb 9 23:23:00 2016
+@@ -113,8 +113,6 @@ static int startany(sox_format_t *ft, unsigned mode)
+ else
+ reqpar.rchan = ft->signal.channels;
+ }
+- if (ft->signal.precision > 0)
+- reqpar.bits = ft->signal.precision;
+ switch (ft->encoding.encoding) {
+ case SOX_ENCODING_SIGN2:
+ reqpar.sig = 1;
+@@ -127,6 +125,12 @@ static int startany(sox_format_t *ft, unsigned mode)
+ }
+ if (ft->encoding.bits_per_sample > 0)
+ reqpar.bits = ft->encoding.bits_per_sample;
++ else if (ft->signal.precision > 0)
++ reqpar.bits = ft->signal.precision;
++ else
++ reqpar.bits = SOX_DEFAULT_PRECISION;
++ reqpar.bps = (reqpar.bits + 7) / 8;
++ reqpar.msb = 1;
+ if (ft->encoding.reverse_bytes != sox_option_default) {
+ reqpar.le = SIO_LE_NATIVE;
+ if (ft->encoding.reverse_bytes)
