xref: /dports/audio/nas/nas-1.9.4/HISTORY

$Id: HISTORY 293 2013-10-07 17:58:33Z jon13 $

Version 1.9.4 (stable) 10/07/2013

  - auerswald

    - Fix several potential security issues in the NAS server reported
      by Hamid Zamani <me@hamidx9.ir> to the NAS mailing list.

      See http://radscan.com/pipermail/nas/2013-August/001270.html for
      a description. In addition, CVE numbers have been defined for
      these issues:

      CVE-2013-4256
      CVE-2013-4257
      CVE-2013-4258

    - Fixed some warnings while building some of the clients.

    - narrow a possible race condition in
      auvoxware.c:disableProcessFlow() reported by Paul Fox.

  - steve_mcintyre

    - add a format string to aulog.c (found using -Wformat-security
      warnings)

    - Don't preprocess man pages

  - Add proper includes in server/os/connection.c to prevent some
    compile warnings and a possible server coredump under the right
    conditions on x86_64 Linux systems.


Version 1.9.3 (stable) 05/13/2011

  - overdue 1.9.3 maint release

  - Older X11 systems used to define SharedLibX and friends if the
    installed X11 system was built with shared lib support.  Newer
    X11's use SharedLibX11 instead.  Same for
    {Normal|Debug|Profile}LibX.  So check for these and set
    appropriately in NetAudio.tmpl.

    Shared lib installations should deal with libXau dependancies
    correctly now.

  - auerswald:

    Added MacOS X aka Darwin to list of systems that need to
    explicitely link libaudio.so against libXau. This enables building
    of the shared library libaudio.so on MacOS X.

    There is currently no NAS server for MacOS X.

Version 1.9.2 (stable) 03/07/2009

  - overdue 1.9.2 release.

  - patch from Aleksey Cheusov:

    "fix for "libpthread: Unlocking unlocked mutex" seen on NetBSD-4"

    libaudio: OpenSvr.c: The server mutex was being unlocked, even
    though it was never locked in the first place.

  - Hugo Vanwoerkom reported thousands of diagnostics in his syslog:
    "doSetup: elementRate == 0! Forcing to 5000".  Fix doSetup to only
    warn about this when the debug level is above 10.

  - auerswald:

    document syslog usage

    use syslog if running as a daemon, stderr otherwise

    add syslog-usage-change to HISTORY file

  - steve_mcintyre:

    Minor man page fix ported from the Debian packages.

    Multiple minor portability changes ported over from the Debian
    packages. Cope with Hurd and GNU/KFreeBSD and other systems using
    glibc

  - apply patch from sourceforge user moinakg:

    Libaudio is built with missing linker flags on OpenSolaris like
    missing lXt, so it results in symbol not found errors at runtime.
    I am including an Imakefile patch below that fixed this.

Version 1.9.1 (stable) 11/10/2007

  - per a request from Frank B�ttner (fedora maintainer), and in the
    interest of providing the 'least surprise' for a user, nasd will
    no longer initialize the mixer settings at start up, by default.
    The default value of 'InitMixer' is now set to 'NO' in the
    /etc/nasd/conf file.

    See https://bugzilla.redhat.com/show_bug.cgi?id=247468 for the
    bug report.

  - based on patch from Frank B�ttner, nasd will create it's local
    socket, and libaudio will look for it, in /var/run/nasd/audio*.
    on Linux systems.  In reality, this should probably apply to all
    'modern' unix systems (Solaris, *BSD, etc), but only Linux is
    affected by this change now.

    See https://bugzilla.redhat.com/show_bug.cgi?id=250453 for the bug
    report.

    The downside is that /var/run is not writable by normal users, so
    if a user is trying to run the nasd server, s/he will get an error
    about being unable to bind to the socket, and nasd will not start.
    The workaround is to run nasd with the '-pn' (partial network)
    option.  This way, nasd will bind to a local TCP transport and
    start successfully even if the local socket transport cannot be
    initialized (though you will still see the error when nasd
    starts).

    Additionally, users on the local host trying to play sounds
    through a server started this way may need to set their
    AUDIOSERVER environment variable to 'localhost:0' for example,
    rather than the default ':0' in order to play sounds.

    Of course if nasd is started as root at system startup time, then
    this is not a problem.

  Version 1.9a (devel) 10/27/2007

  - this is a possible 1.9.1 (stable) release in a couple weeks.

  - modify intervalProc to stop/start the timer around AuProcessData()
    calls, rather than simply disable the signal handler.  This should
    ensure that we never lose a timer signal, and seems to fix the
    reported nasd hangs under newer linux kernels (2.6.21+).

    The *BSD folks should make sure everything still works, it
    should...

  - rework signal handling in the voxware (OSS) server.  This also
    allows DIA to block/unblock interrupts at he appropriate times
    now.

  - apply patch from Yarda that corrects a problem with blocking and
    unblocking signals in IntervalProc() (auvoxware). [jet - this
    patch was removed in the patches further up]

  - remove mention of the auvoxware manpage from the filelist (for src
    packaging), and remove the doc/html/auvoxware.1.html file.

  - Apply patches from Erik Auerswald:

    - the recent bug report from a fedora 7 user showed a small bug
      with the input mixer code: If the input mixer cannot control the
      input channels a wrong error message is printed. The attached
      patch fixes this.

    - remove auvoxware.man from the voxware Imakefile

    - aupanel: add option -interval to synopsis section of man page

    - add automatic querying of device attributes to aupanel with a
      default query interval of 10 seconds

    - make the nasd.conf man page formatting more consistent

  - more cleanup and removal of old/useless files.

  - remove useless nas.lsm file.

  - sync up for copying repository to sourceforge.

  - Patch from Stefan Huehner:

    - remove SleepQueue functionality which is apparently not being
      used anywhere.

  - remove useless 'register' declarations

  - convert Swap32Write for K&R to ANSI.  Also, define it's buffer
    pointer as AuUint32 rather than 'long', which was wrong, but
    worked accidentally.


Version 1.9 (stable) 04/07/2007

    - stable 1.9 release.  No code changes (other than RELEASE) since
      1.8b devel release.

    - In short: Fixes to several DOS attacks that could be carried out
      aginst a nasd server, ANSI'fication of most of the server and
      client code, and significant rework of input and output mixer
      handling in the voxware (OSS) server.

      Read below for more detail on the changes since the last stable
      release (1.8).


    Version 1.8b (devel) 03/25/2007

    - 1.8b devel release.  Final 1.9 in about 2 weeks, barring any
      catastrophes.

    - fix a variety of problems that could result in a denial of
      service by crashing the nasd server.  These attacks were
      researched by Luigi Auriemma, who also provided a description of
      the attacks and an exploit program, 'nasbugs'.

      I have added his emailed report and the test attack code to the
      nas repository in contrib/nasbugs if you are interested.  Thanks
      to Luigi for finding these problems.  It sucked fixing them :)

      Here is a list of the bugs tested as output by the nasbugs
      program:

      1 = accept_att_local buffer overflow through USL connection
      2 = server termination through unexistent ID in AddResource
      3 = bcopy crash caused by integer overflow in ProcAuWriteElement
      4 = invalid memory pointer caused by big num_actions in ProcAuSetElements
      5 = another invalid memory pointer caused by big num_actions in
          ProcAuSetElements
      6 = invalid memory pointer in compileInputs
      7 = exploits bug 3 in read mode (requires something playing on
          the server)
      8 = NULL pointer caused by too much connections

      Note on bug #2, X11 display servers should be vulnerable to a
      DOS of this type as well (causing fatal 'client not in use'
      errors in AddResource()).

      Note on bug #8, the nasd server will not be able to accept
      further client connections when the client table is full, until
      the rejected clients disconnect their end of the socket and the
      neccessary fd's are freed up.  It's better than coring though.

    - set the default open modes for the output audio device to
      write-only, and for the input device, use read-only (already the
      default).

      These can still be changed in the nasd.conf file, if needed.

      Previously, the default was to open the output device
      read-write, causing various issues with different
      hardware/driver configurations.  There is no need to open this
      device read-write anyway.

    - apply patch from Paul Fox, correcting a typo in auvoxware.c

    - apply patch from Erik Auerswald

      "With the attached patch the output and input device can be
      disabled by specifying an empty string as device name. Yes, when
      using an empty string for both devices there will be a NAS
      server that can neither play nor record anything."


    Version 1.8a (devel) 11/27/2006

    - Updated the Amd.h (machine definition file) with modern X11
      (X11R6.8) contents so it can build with Qt and other X11
      software that makes use of X11's Xmd.h header file.  Problem
      reported by Bernard Leak.

    - ANSI'fication of DeleteTypes (dia/resource.c) and NoopDDA.
      Add proper header #includes where needed.

    - patch from Petr Salinge (via Steve McIntyre) adding GNU/kFreeBSD
      support.

    - Patches from Paul Fox:

      - allow the specification of a scaling factor to apply to the
        output gain

      - add a proper return value in auvoxware.c:initMixer()

      - mondo indentification of server/*

    - Applied patches from Stefan Huehner:

      - marks some read-only string and function parameters as 'const
        char*' instead of 'char *'. In addition an unused buffer
        'errfile' from aulog.c was removed.  As the two release.h
        files are generated the NetAudio.tmpl was modified to emit the
        'const char *' declaration.

      - part two of the _.*Const removal patch. Some more defines of
        _.*Const were removed which have been missed by the last patch
        and the interval usage of _.*Const has been simplified to just
        use const.

      - removes more unused defines and functions in server/ .
        Additionally a k&r style function was convertd to ansi c in
        swaprep.c

      - converts more function declarations and their prototypes
        without parameters from () to the (void) form

      - attached patch changes all of the conditional i.e. _AuConst
        definitions to only list const in the externally visible
        header files. This is to preserve compatibility if someone
        uses these definitions while using the libaudio library.

        [JET] The use of AuConst is deprecated (and is probably not
              being used by anyone anyway).

      - removes the defined VENDOR_STRING and VENDOR_RELEASE from
        server/include/site.h as they aren't used in the 3 files
        including site.h.  Additionally the inclusions of site.h was
        removed from server/dia/main.c as none of its defines is used
        in main.c

      - removes the 2 unused functions:

        - server/dia/dispatch.c: void UpdateCurrentTime()
        - server/dia/dixutils.c: TimeStamp ClientTimeToServerTime

          and it changes one more occurence of () to (void).

      - corrects some function declarations by changing () to (void)

      - removes some unsued variables in lib/audio/nameaddr.c and
        and unused function in server/os/utils.c

      - removed some old (and stupid) debug code for AIX: in
        server/os/utils.c there is some code to redirect error logging
        from stderr to and file in tmp on AIXV3. As OpenDebug isn't
        called anywhere this is broken when AIXV3 is
        defined. Additionally this logfile is fixed in /tmp which
        permissions 00777

      - removed the declaration of FreeResource from the opaque.h
        header file. All files which include opaque.h don't use this
        function and this declaration is copied from resource.h

      - corrects several prototypes in resource.[ch] to match the
        actual function definitions inside resource.c

      - patch that modifies the two users of ostruct.h to directly
        include os.h (which includes misc.h) and removes the
        osstruct.h line in config/filelist. If an additional "svn
        remove server/include/osstruct.h" is used we can get rid of
        osstruct.h completely.

      - removes servermd.h. removes the inclusion of servermd.h in 3
        files and its reference in 'config/filelist'.

      - clean up several of the client programs. Mostly adding const
        to parameter/variable declarations. Additionally an unsused
        variable was removed and some prototype in audemo.c were
        fixed.

      - remove some unused variables

      - remove server/include/miscstruct.h

      - various include file cleanups - removing unused
        variables/structures, etc left over from X11.

      - removing old 'const' determination, correcting some proper
        uses of const.  'const' is expected to be supported by your
        compiler.

      - ANSIfication of various NAS components.

      - remove unused portion of X11 server components (fonts, GC's,
        etc).

    - applied patches from Erik Auerswald:

      - Add patch adding a proper 'reset' action to contrib/rc/nasd.

      - a patch that:

        Adds support for a different mixer for the input device to the
        voxware server. Without this patch the mixer device specified
        in the input section of the config file is not used.

        To use the same mixer for input and output just specify the
        same device in the input- and output-section of the config
        file. The empty string "" can be used to specify no mixer
        device for the input- or output-section.  This is documented
        in the nasd.conf man page.

        The outputsection keyword is added to the nasd.conf man page.

        fixed a bug in the support of two devices: If two devices are
        used, and the sample rate of the input device can be changed,
        the wrong sample size would be set by openDevice().

        remove the unused "stereodevs" variable.

        NOTE: Prior to this patch, the mixer device parameter for the
              input section was completely ignored.  The default value
              for this ignored parameter was "/dev/mixer1".

              With this patch, this parameter is honored.  If your
              configuration depends on the input mixer being the same
              as the output mixer, you will need to edit your
              /etc/nas/nasd.conf and set the mixer parameter in the
              inputsection to '/dev/mixer', or to whatever mixer
              device your outputsection is using.

              If your system contains a /dev/mixer1 device (the
              default input mixer device), nasd will now use it.  This
              is a good thing. :)

      - preserve configuration of input device in the voxware server
        when ReleaseDevice is on, and no flow is active.

      - document nasd '-V' option in the man page

      - opening the mixer device should be allowed to fail according
        to http://radscan.com/nas/nas-ml/msg01121.html. But right now
        it may only fail on startup, not when re-opening the audio
        device. The attached patch corrects this issue.

      - adding an option to aupanel to specify the initially
        controlled device.

      - keeping gain and input mode changes consistent between voxware
        server and mixer if possible.

      - changing recording level controls of the voxware server to use
        IGain or RecLev when available and select only the specified
        recording source.

      - fixing MixerInit option of the voxware server.

      - Adding a KeepMixer option (on by default) that causes NAS to
        always keep the mixer device open so it can track external
        changes to mixer settings.

      - patch adding ReInitMixer option that init's the mixer to
        defaults every time nasd re-opens the audio device.

      - patches ANSI'fying the nas clients

      - extending the gainScale functionality to apply to the input
        gain as well.


Version 1.8 (stable) 06/03/2006

        - time for another stable release.

        - Correct problem in auvoxware, where if MixerInit is set to
          false, remote volume (mixer) requests are ignored.
          Reported by Paul Fox, with a fix determined by Steve
          McIntyre and Erik Auerswald.

        - Applied patch from Nicolas Escuder correcting a Cygwin
          issue in auvoxware.c

        - applied patch from Erik Auerswald adding a Mute button and
          functionality to aupanel.

        - applied patch from Nicolas Franois  (via Steve McIntyre)
          correcting some typos in the AuSetBucketAttributes and
          AuSoundCreateFileFromBucket man pages.

        - applied patch from 'Inma', correcting a connection hang in
          Solaris 9.  This reworks some of the very old and primitive
          fd_set stuff in Aulibint.c .

        - updated config/config.sub and config.guess to properly
          detect DragonFly systems.

        - Applied patch from Mark Davies and Joerg Sonnenberger
          correcting some BSD issues and adding DragonFly support.

          The nasd.conf man page is always installed in
          $FILEMANDIR for all systems.

        - applied patch from Mark Davies correcting audio/Imakefile
          to properly define BSD44SOCKETS on NetBSD systems.

        - Applied patch from Steve McIntyre correcting CYWIN -> CYGWIN
          typo in server/dda/voxware/auvoxware.c

        - applied patch from Steve McIntyre Correcting some Gcc 4.0
          and AMD64 issues.


Version 1.7 (stable) 11/14/2004

        - 1.7 release :)

        - fixed config/Imakefile so 'configure' command would work
          under cygwin.  Tested clients/server under cygwin.

        10/06/2004

        - 1.7 release candidate

        - corrected a possible race condition in auvoxware that could
          cause the server to hang under certain conditions.  I do not
          know why I haven't seen this problem before, but I could
          recreate it easily by killing auplay while playing a long
          tune.

          If the system restarts syscalls after a signal handler has
          run, a situation could arise where the previous timer was
          never disabled.  This could cause a previous ioctl(DSP_SYNC)
          called from setSampleRate() to be constantly interrupted and
          restarted, causing an apparent hang.

          Now we just turn off the timer as the first thing in
          setSampleRate() so DSP_SYNC can actually finish.

          We also turn off the timer when openDevice is called, so the
          'wait for device to become available' code can work too.

        - Patch from Tobias Diedrich that allows mixer init to
          fail in the voxware server.

          Quote:

          With this patch "aoss nasd" works, without it would fail
          (aoss is the alsa oss wrapper script, which simply preloads
          libaoss.so, a wrapper library similar to libaudiooss).

          Combining this with the alsa dmix feature should allow nas
          to coexist with other daemons like artsd.

          See http://radscan.com/nas/nas-ml/msg01121.html for more
          details.

        - redid the FAQ in an attempt to make it more useful.  I'll
          try to keep it up to date.

        - updated libtool due to a problem(s) reported by Steve
          McIntyre.

        08/20/2004

        - clear CFLAGS before building config/config.h

        - emit the release string to the log and stdout when
          verbose is enabled.

        - if '-V' specified, emit release string and stdout and exit.

        - add a limit of 5 retries on failure to open input device.

        - barring major problems, this is a release candidate.

        06/20/2004

        - add autoconf/configure support for config/. Now, during the
          'includes' phase of the nas build, configure will try to
          figure out some things about your system and build a
          config.h file that the rest of nas can use.   Prompted by
          John Wehle's patch below.

          NOTE: you still build nas like always with the imake
          facility.  configure in this case is only concerned with
          locating various header files, functions, and the like.

        - added FreeBSD 5.x patches from John Wehle correcting
          malloc.h/stdlib.h use and other issues.

        - remove ioctl(devAudio, AUDIO_DRAIN, 0) call from
          disableProcessFlow() in dda/sun/ausuni.c that seems to cause
          problems (server hangs) with certain sun audio
          hardware/kernel drivers.

        - applied patch from James Lee correcting a coredump in
          audemo.

        - corrected problem with dia/auprocess.c that caused SIGFPE's
          in nasd when an element with a samplerate of 0 was
          processed. Thanks to Tobias Diedrich for a test case.

        - removed some pre-R5 cruft from lib/audio/ConnSvr.c that
          caused build problems on some solaris systems.

        02/14/2004

        - Fixed up some Cygwin issues... All of the clients and the
          server seem to compile and work fine now.  I tested both
          local and remote, to and from unix and cygwin.

        - Added patch from Steve McIntyre, adding a new '-config'
          option to nasd.  This allows you to start nasd with
          alternate configuration files rather than the default of
          /etc/nas/nasd.conf.  Useful if you have multiple sound
          devices.

        - Added 64bit fixes from Per Oyvind Karlsen

        12/31/2003

        - corrected a coredump with the clients in 1.6a when using a
          non-zero display number. GetAuthorization() was a bit
          confused about the order of it's arguements...

        12/24/2003

        - preparing a long overdue development release

        - ANSI'ification of libaudio.

          There seemed to be some confusion with the order of
          arguments to the AuSoundRecordToFile*() functions.

          The function declarations were made to match the prototype
          definition and the man page.

        - Modified Scott Presnell's patch to not enable BSD44SOCKETS
          on Linux or SVR4 systems. There are some structure members
          used that are not present on these (and perhaps other)
          systems.

          If you run into compilarion problems with ConnSvr.c relating
          to invalid structure members, look at the define of
          BSD44SOCKETS in Imakefile and correct for your system as
          needed.  Oh, send me the patch as well? :)

        - auvoxware

          - added some more useful debugging info.  If you get the old
            'cannot create audio block connection info" try running
            nasd with '-v -d 99' and see what the real error is.

          - certain ioctl() requests are no longer fatal (like
            SOUND_MIXER_WRITE_RECSRC and SOUND_MIXER_READ_RECSRC).
            Recording may not work properly if these fail though.

        9/14/2003

        - added patch from Marty Leisner cleaning up a couple of error
          messages in server/os/connection.c

        3/15/2003

	- added a patch to the contrib section by Laurent Vivier - this
          patch adds NAS support to the esound daemon on AIX systems.
          Should be reasonably portable to other platforms as well.

          Rumor has it that the Ximian folks are adding this support
          officially into esound.

        3/15/03

	- set timer in auvoxware.c to trigger when 50% of the samples
          are predicted to be consumed... This hopefully will help
          with underruns on some systems/OS's...  I'm not sure why
          this is happening on those systems (it's not happened on any
          of the systems I've used NAS on), so this should probably
          be considered a harmless workaround.

        3/4/03

	- Patch from Scott Presnell to set BSD44SOCKETS properly for
          libaudio.

        12/11/2002

	- patch from Steve McIntyre correcting a typo in Xtutil.h


Version 1.6 10/23/2002

	- Release!

	## 1.5h    10/12/2002

	- Added SoundFileGetProc() to get access to the toSound and
          fromSound procs in _SoundFileInfo[].  Needed for xemacs.

	- Added AudioLibraryVersionMajor, AudioLibraryVersionMinor,
          and AudioLibraryVersionTeeny to audiolib.h.  Somewhat more
          informative than just the old AudioLibraryVersion which was
          set at '2'.

	  The old AudioLibraryVersion define is still there however,
          and is defined to be equal to AudioLibraryVersionMajor.

	## 1.5g    10/7/2002

	- changed inputsection defaults in nasd.conf.eg to 44.1Khz, 2
          channel 16b sound.

	  outputsection also defaults to 44.1Khz now, instead of 33Khz.

	- began some CC ANSI'fication of libaudio.  So far only a
          couple of files have been 'upgraded' just to see what the
          implications are.  It's going to be a bitch ;-) And there is
          the likelyhood of introducing errors, so I'll get a stable
          release out first before finishing this task.  The result
          however should be worth it.  I've a feeling a few obscure
          bugs will be found and fixed doing this.

	- some cleanup in server/os/utils.c and server/include/os.h.
          removing K&R style prototype decls, #including unistd.h, and
          setting Xalloc() et. al. to use proper void * rather than
          unsigned longs.

	- changed hardcoded '-L/usr/X11R6/lib' for REQUIRED libs in
	  lib/audio/Imakefile to '-L$(USRLIBDIR)'

	- it doesn't have the HPUX 11 build fixes - I think I will
          really need access to a machine, or someone who does have
          access to one and wants to fix the build, in order to solve
          this... The bcopy problem seems to be in the HPUX X11
          config/ stuff.

	- We really should release this one unless there are big
          problems. ;-)

	8/26/2002

	- time for another development update.  Should probably go
          final with a 1.5.1 version soon...

        7/23/2002

	- removed fdset.h stuff for cygwin - no longer needed.

	- auvoxware - added __CYGWIN__ checks (equal to linux)

	- added __CYGWIN__ to dia/auservertype.h

	- changed server imake to build nasd.exe on cygwin rather than nasd.


        7/20/2002

	- added cygwinArchitecture to the server Imakefile so
          auvoxware will be built on cygwin systems.

	- Removed the SoundFileInfo[] array in sound.h so cygwin would
          be happy.  Apps should not have been accessing that array
          directly anyways.

	  Added the following new functions to sound.c (that used to
	  be macros in sound.h):

	     SoundFileFormatString, SoundValidDataFormat,
	     SoundFileFormatToString, SoundFileFormatToAbbrev,
	     SoundFileFormatToSuffixes

	  Assuming your code did not attempt to directly access
	  SoundFileInfo[], no changes should be required.

	  libaudio.so revision bumped to 2.3

        7/15/2002

	- patch from Nicholas Wourms.

	  - fixed FIONREAD in Alibint.c for cygwin
	  - fixed Imake for libaudio on cygwin

        7/9/2002

	- rename dia/nasconfig.* to dia/nasconf.* to avoid name
          collision with NasConfig.h on windows systems.

	## 1.5e    7/7/2002

	- added patch from Nicolas Escuder that adds support for
          Windows operating systems using Cygwin.  Even the NAS server
          is now supported.  There are still issues with the clients.

	  I will provide a link on the nas-links page to precompiled
	  windows binaries for libaudio.dll, nasd, and some clients -
	  provided by Nicolas.

	- based on idea from Nicolas, added background (-b) mode to
          nasd, to force it to run in the background.  Actually, he
          supplied a patch, but I just redid it to work properly with unix
          systems.

	  If you want to see error messages from nasd while starting
	  up, do not run it in daemon mode or you won't see the error
	  messages as stdout/stderr are closed.  Someday, I'll track
	  down all of these and have them use osLogMsg(), which would
	  work regardless.

	- added NAS_AUDIOMSGFILE in config/NasConfig.h to specify
          where audio messages are stored (ADMPATH in os/osinit.c).

	- Yes, we are at 1.5e now. Sorry I didn't release 1.5b-d.  Was
          a little busy with the day job ;-)

	## 1.5d    6/1/2002

	- added .ps -> .pdf conversions in doc/pdf/

	- added html versions of man pages in doc/html/

	## 1.5b    3/16/2002

        - added patch from Marco Molteni to add a '-local' option to
	  nasd.  Specifying this option allows only local clients to
	  connect.

        - fix nasd.man to describe this and other supported options.

        - moved allow_any_host into the NasConfig struct.  Along with
          the new '-local' option.

	## 1.5a    2/23/2002

	- more STARTSERVER fixes - there was still one case where when
          STARTSERVER was defined, nasd could be called with the wrong
          display number due to a bug in _AuConnectServer.

	- changed LEX/YACC DEBUG token to CDEBUG to avoid conflicts
          when building nas with '-DDEBUG'

	- DEFAULT_AUSERVER is now ":0" rather than
          "tcp/localhost:8000".  A local socket connection will startup
          faster than a tcp connection.

	- mutex.h
		- corrected unixware 7.x build problem
                  (XMUTEX_INITIALIZER) in mutex.h
		- added same check for sun/solaris boxes
		- got rid of XUSE_MTSAFE_API - now controlled by
                  NAS_USEMTSAFEAPI defined in config/NasConfig.h

		  If you think you are having mutex problems, you can
		  edit config/NasConfig.h and '#undef'
		  NAS_USEMTSAFEAPI, then rebuild everthing.  This will
		  turn off threadsafety in libaudio.

	- added patch from Tobias Diedrich to fix deadlock in
          AuNextEvent().

	- added new mutex in audio/Alibint.c to protect
          _AuReadEvents() since it can no longer rely on the server
          lock as a result of Tobias's patch above.

	- added new option, 'gain', to config file in the input/output
          sections.  This sets the default gains when nasd starts up
          (if MixerInit is enabled).  Only in voxware for now.

        1/5/2002

	added -lXt and -lXau to the lib dependancy for libaudio.so on
	Linux systems per a request from Lorenzo Delana.

        1/20/2002

	- fixed bug in ConnSvr.c that could cause the wrong server
          number to passed to nasd when STARTSERVER is defined.


Version 1.5 12/26/2001

	## 12/1/2001

	- Corrected manpage html generation issue with the nasd.conf
	  manpage via patch from Erik Inge Bolso


	## 1.4.2d  11/25/2001

	- implemented locking in libaudio via Xthread mutexes.

	  This is kind of a big change, but the result should be a
	  threadsafe libaudio.  I used Xthreads, because then
	  theoretically I shouldn't have to care about the system's
	  actual thread implementation. Right?? I've tested so far on
	  Linux (uses pthread_*) and Unixware 2.03 (uses thr_*).  So
	  far so good.

	- bumped up the libaudio rev to 2.2

	## 1.4.2c  11/4/2001

	- turned off streams support on SVR4 systems for libaudio.
	  UNIX and TCPCONN work just fine.  The nasd server hasn't
	  used streams support since the 1.3 timeframe anyway.  This
	  will work around local connection issues on UnixWare and
	  probably Intel Solaris as well.

	- added patches from Steve McIntyre to work around an XFree86
	  issue, and prevent the HPUX server from being built on HPPA
	  systems running Linux.

	- add change to AuServerName() in libaudio to default to
	  'tcp/localhost:8000' if neither AUDIOSERVER or DISPLAY are
	  set as per a suggestion by Erik Inge Bolso.

	## 1.4.2b  10/6/2001

	- Added patches from Peter Berg Larsen to the Sun server that
	  adds support for $AUDIODEV on Sun Ray platforms.

	- Added patches from Peter Berg Larsen that corrects some
	  issues dealing with dead clients, and improper server resets
	  when persistent flows are still active.  Also, some
	  uninitialized variables are now properly initialized.

	## 1.4.2a  4/21/2001

	- patch from Steve McIntyre to resolve man page issues, and
	  add a libm dependancy to libaudio for linux platforms.

	- patches from Philippe Jouguet

		new options in nasd.conf for the voxware server:
		- add 'mixer' device option to secify the mixer.
		- add 'forcerate' option to force the sample rate
		  to the current rate at all times
		- add 'autoOpen' option to specify whether the audio
		  devices are opened at init time.
		- add 'readwrite' option to specify whether devices
		  are opened read/write (default) or
		  readonly/writeonly for the input and output devices,
		  respectivly.


	- added patch from Nick Ing-Simmons to pass the correct port
	  offset when starting nasd within libaudio (if START_SERVER
	  is defined at build time).

	- always output verbose/debugging info to stderr if debugging
          is enabled (even if you normally use syslog).  stderr is not
          closed when debugging is turned on.


Version 1.4.2 4/8/2001:

        ## 1.4.1b 3/9/2001

	- added patch from Paul Fox allowing auctl to set gains based
		in an increment

	- added a patch to lib/audio/wave.c from Philippe Jouguet that
          corrects an infinite loop that can occur when auplay is fed
          a corrupted wave file.

	- added patches to auplay from Paul Fox.  auplay will now stop
          the flow when a signal is caught.  Also a new option '-l'
          allows auplay to play filenames specified on stdin.

	- set libaudio's SO revision to 2.1 from 1.4 to avoid
	  confusion with older apps built when the older libaudio
          was at 2.0 (debian).

	## 1.4.1a 12/10/2000

	- autool - fixed -volume option

	- added cast from a patch by Sun Ireland for their OpenOffice
		(StarOffice) project.

	- fixed libaudio to start up 'nasd' instead of 'au' when
		START_SERVER is defined.

	- applied Erik Inge Bolso's patch for properly opening 2
          channels instead of 3 for stereo output devices.

	- added HPUX 10.20.x patches from Klaus Dittrich

	- fixed problem in lib/audio/Alibint.c regarding ambiguous
          casts with void pointers in the buffer padding code, and to
          make it more portable with other operating systems/compilers
          (HPUX 10.20, UW7).

	- lib/audio/soundlib.c - fixed cases where a client (like
          auplay) would hang if the element Import size matched the
          complete datasize of a flow.  An EOF would never be sent.
          Thanks to Shigeharu Takeno for providing a reproducible test
          case.

	- added Unixware 7.1 patches from Robert Roselius.

 	- added patch to sun/ausuni.c from John Wehle correcting a
          problem on some solaris machines that don't allow sbpro
          devices to be open for both reading and writing.

	- add patches from Steve McIntyre for creating sticky socket
          directories, and to close stdin, stdout, and stderr.



Version 1.4.1 9/4/2000:

	- this release is really only neccessary if you are using the
          NAS server on a Sun, and have frequent dropouts.

	- fix to the sun server from Charles Levert correcting a Sun
          problem with 'late SIGPOLLS'.

	- minor documentation fixes - new mailing list archive
          location, etc.

Version 1.4 2/27/2000:

	- Release of 1.4

	- Removes USE_BSD* dependancies on Linux builds.

	## 1.3e 02/12/2000:

	- This is a pre-release 1.4 candidate.

	- corrected a problem in the voxware server where when
		ReleaseDevice was enabled, the device would be held
		open when nasd was first started until the first
		connection was made.

	- added config/NetAudio.def to set various configurables like
		NasConfigSearchPath.

	- changed documentation regarding the new location of the web
		site (http://radscan.com/nas.html)

	## 1.3d 01/05/2000:

	- corrected some build problems on Sparc Solaris 2.6 with native
		compiler/X11 env

	- corrected build problems on native Unixware compiler and X11R5
		Note, to get this to work, you need 3 corrected
		/usr/X/lib/config/ files from me...

	- integrated config handling in all servers, based on a
		lex/yacc parser from Stephen Hocking that was in the
		voxware server.

		- config file is /etc/nas/nasd.conf
			an example config file (/etc/nas/nasd.conf.eg)
				is installed.
		- parser errors are a bit more informative ;-)
		- no more parser in voxware server... moved into dia/
			for general purpose use.
		- each server now has a config.c and config.h file that
			provides the ddaSetConfig() functionality required
			of each server by dia/.  In the SGI case,
			it's just a stub. ddaSetConfig() ignores options
			it doesn't understand (required behavior).
		- server cmdline options '-v' (verbose) and '-d <num>'
			(debug messages) added.  These override
			nasd.conf settings.
		- hooks added to all servers in config.c - ddaProcessArg()
			and ddaUseMsg() for integrated arg handling.


	- voxware server:
		- release device functionality now controlled by
			nasd.conf
		- mixer init policy now controlled by nasd.conf

	- sun server:
		- added integrated sun (ausuni) module (now built by
			default) into the distribution.
		- release device functionality now controlled by
			nasd.conf

	- hpux server:
		- added device release (ReleaseDevice) capability to hpux
			server.  All servers except for sgi now
			support this functionality.
		- fixed sample rate change problem in hpux server
		- separated hardware init code for ReleaseDevice
			functionality
		- added new option in nasd.conf to select either the
			EXT (external - headphones/speakers) or INT
			(internal speaker) output device by default

	- renamed generated server binary from au* to nasd for all
			servers
		- man pages updated (auserver.man and
				AUVoxConfig.man	respectively).

	## 1.3c 10/24/99:

	- Added patches from Mark Davies for NetBSD.

	- changed BIG_ENDIAN ->NAS_BIG_ENDIAN and LITTLE_ENDIAN to
		NAS_LITTLE_ENDIAN per Mark's suggestion globally

	- Added global NasConfig struct for configurables global to all
		servers.

	- More prep work for unified arg/config handling

	- Some bug fixed related to the 'mondo' patch.

	- Replacement of PRMSG() macros with a varargs logging function:
		osLogMsg(...).  If your working on a server and want to
		use these, #include "aulog.h" (in 1.3c+) and go for it.

	- I've added a NasConfig.h file to config/ that defines (or will)
		a few things (like whether you want to use syslog,
		etc).  Initially you will need to tweak this manually
		for your site...

	- Changes made to the way various include files are generated
		during a build... Basically any module that contains
		header files needed by other modules have the
		appropriate header files exported into TOPLEV/include
		during the includes phase of the build using the imake
		BuildIncludes() macros.

	## 1.3b 9/4/1999:

	- Addition of Steve McIntyre's Mondo patch.

	## 1.3a 8/1/1999:

	- First test version incorporating many patches from around
		the planet, based on original 1.2p5 sources.

#
#
############################################################################
# the following is included for historical reasons, and concerns
#  changes to 1.2p5 and earlier versions.
#
# $NCDId: @(#)HISTORY,v 1.27 1996/09/05 18:46:47 greg Exp $
#

** 1.2 - PATCH 05 **

Server
	o voxware bug fixes, configuration file, and man pages.
	  (Thanks to Stephen Hocking)

	o fixed some HP server compilation problems.

	o fix a potential problem when reading from a client.  under
	  some circumstances the read pointer could be left on
	  a non-sample size boundry resulting in white noise.

	o add a bit to the component use field to indicate that using
	  that component will override the use of other components.  This
	  bit is used on input devices when the audio hardware does not	  support recording and playing simultaneously.

	o fix a problem with limiting samples on waveform devices.  if you
	  used anything other than AuUnlimitedSamples, bad things would
	  happen.

Library
	o fixed a problem reading some wave files.

Clients
	o auphone - check for the exclusive use bit.

	o auscope - display the exclusive use bit.

Misc
	o support library and client compilation under Windows NT with
	  X11R6.1.

** 1.2 - PATCH 04 **

Server
	o fixed a bug in the Sun dbri driver that caused the speaker
	  to get turned off.  (Thanks to Charles Levert)

	o fixed an ancient server bug that caused buffers to get
	  corrupted when a flow is stopped and restarted.

Clients
	o some more minor bug fixes to auscope.

** 1.2 - PATCH 03 **

Server
	o HPUX server now functional.  (Thanks to Jay Brister)

	o Fix non-rentrant malloc problem for SGI server under
	  IRIX (5.3).  (Thanks to Scott Byer)

	o removed obsolete SVR4 voxware code.

	o catch bad protocol requests.

Library
	o prevent AuSoundPlayFromFile() and AuSoundPlayFromData() from
	  changing the line mode.  A bug introduced in patch 2.

	o fixes for shared audio library under X11R6.  (Thanks to
	  Bill Sebok)

	o clean up some defines in the STARTSERVER code. (Thanks to
	  Bill Sebok)

	o added AuSoundRecordToDataLength() which can be used to find the
	  length of data recorded when recording to memory.
	  (Thanks to Jonathan Thompson)

Clients
	o minor bug fix to auscope.

	o audemo - cut off filename extensions from the right instead
	  of the left. (Thanks to Raymond Toy)

Misc
	o Updated SunOS 4.1.3 OpenWindows notes in BUILDNOTES.
	  (Thanks to Curt Schroeder)

	o Added notes and patch for running on a SS5 to BUILDNOTES.
	  (Thanks to Mikael Pettersson and Curt Schroeder)

	o Added note for building clients under AIX to BUILDNOTES.
	  (Thanks to Harald Milz)

	o Added note for building on DEC Alpha (OSF1 V3.2).
	  (Thanks to Andrew Daviel)

** 1.2 - PATCH 02 **

Server
	o caused the server to return to the lowest sampling rate when a
	  flow terminates while playing multiple flows simultaneously.

	o improved voxware dda driver.
	  (Thanks to Stephen Hocking and Shawn M.)

	o experimental HP dda driver.  (Thanks to William J. Hunter)

	o add a -timeout option to the server to allow the server to
	  terminate after a period of time with no clients.
	  (Thanks to William L. Sebok)
	  To enable this, you must build the NAS tree using the following:

	  make WORLDOPTS='-k CDEBUGFLAGS="$(CDEBUGFLAGS) -DSTARTSERVER"' World

Library
	o fixed a bug in the wave file reader.

	o fixed a bug in the aiff file reader.

	o new soundlib/sound functions AuSoundPlay(), AuSoundRecord(),
	  and SoundTellFile() with man pages.  (Thanks to Yoav Eilat)

	o attempt to launch an NAS server on the local host if client
          cannot initially connect to a local server.
          (Thanks to William L. Sebok)
	  To enable this, you must build the NAS tree using the following:

	  make WORLDOPTS='-k CDEBUGFLAGS="$(CDEBUGFLAGS) -DSTARTSERVER"' World

Clients
	o add *input resources to all X clients to correct a problem
	  with some window managers not giving focus.

	o fixed bug in auscope when displaying actions or parameters.

	o fixed typo in auscope that caused it not to work with Perl 5.

	o added playRaw example client.


** 1.2 - PATCH 01 **

Server
	o added output mode control (Speaker/Headphone/Line-Out) to the server
	  and the Sun dda drivers. (thanks to Ian Donaldson)

	o fix for multiple bundle elements in a flow. (thanks to Steven King)

	o improved voxware dda driver.
	  (thanks to Christian Schlichtherle and Stephen Hocking)

	o the server now does a range check on the output data and
	  clips correctly.  Due to the possible performance impact,
	  this can be disabled by commenting out the #define
	  OUTPUT_RANGE_CHECK in server/dia/auprocess.c.  ** ATTENTION
	  DDA DEVELOPERS ** this fix required a change to all of the
	  dda drivers.  The output minibufs must be twice as large as
	  they were previously.  (thanks to Steven King)

	o voxware dda driver - Fix problem (apparently from
	  SoundPlayFromBucket) where tail of data was not flushed
	  through card until start of next operation.  (Thanks to Paul
	  F. Werkowski)

	o only call dda set sample rate callback if the rate has changed.

Library
	o SVR3.2 with STREAMSCONN wouldn't work with XDMCP due to a
   	  bug in nameserver.c.  (required XDMCP to be disabled; this
   	  won't affect users of stock NAS 1.2 but would affect those
   	  audio servers that implement XDMCP authorization checking)
	  (thanks to Ian Donaldson)

	o fatal errors now lookup AuErrorDB for more detail.
   	  (like Xlib does with XErrorDB) (thanks to Ian Donaldson)

	o Implementation of user callable error catching functions:
		AuSetErrorHandler()
		AuSetIOErrorHandler()
	  (thanks to Ian Donaldson)

	o added default error handler variables so you can catch errors
	  during the initial server open.  (Thanks to Richard Caley)

	o added some setjmp gunkus to Xtutil.c for use by error handlers.
	  (Thanks to Richard Caley)

	o C++ compatibility. (Thanks to Yoav Eilat)

	o New SoundSeekFile() and SoundFlushFile() functions.
	  (Thanks to Yoav Eilat)

	o SoundNumSamples() value is now updated in SoundWriteFile().
	  (Thanks to Yoav Eilat)

Clients
	o added re-written auscope.

	o modified aupanel to support output mode controls.

General
	o made all the Imakefiles use an explicit path for NetAudio.tmpl
	  to fix possible includes of the wrong NetAudio.tmpl.

	o some notes on building under SunOS 4.1.3 OpenWindows from
	  Ian Donaldson.

	o new API man pages from Steven King.  (Thanks!)

	o made man page installation use standard InstallManPage macro.

	o use AuInstallScript macro to install scripts.

	o fixed some build and install problems under FreeBSD 1.1.5.1.
	  (Thanks to Paul F. Werkowski)

** Release 1.2 **

General
	o A few cosmetic changes.

** PATCH 02 **

Server
	o fixed a bug with monitor elements not returning the correct
	  data.  This caused the meters in audemo to be broken.

	o patches from Steven King

	    - In server/os/connections.c, EstablishNewConnections
              probably ought to check that newconn is less than
              lastfdesc before it is used in the BITSET macro (this in
              the mit X11r5 source, havent looked at r6 yet), tho
              you'd need lots and lots of connections to tickle
              this. Also in connections.c, CheckConnections should
              probably check for EINTR if select returns an error, I
              dont know if any implementation of select would actually
              EINTR for if it got a SIGPOLL or SIGALRM if the timeout
              is zero, but better safe than sorry.

            - FallbackDefineSelf exists and is only referenced in
              os/access.c. Its part of the localconn patch which
              obviously isnt handling non 386 svr4's correctly.

Library
	o fixed a bug in AuSoundRecordToData().  The flow wasn't getting
	  stopped when the recording was finished.

	o fixed compile problem in ConnSvr.c under Dell 2.2 Xfree86-2.1.1.

	o fixed compile problems under some X11R4s that don't define _Xconst.

Clients
	o fixed problem with audial -recognize messing with line mode and
	  gain attributes without checking to see if the device was capable.

	o fixed range/clipping problem in effectsAmplitude in auedit.

General
	o changed occurances of the name "NetAudio" to "Network Audio System"
          or "NAS" due to a name conflict.

	o patches from Ian Donaldson

	    - Solaris 2 server failed to link because of undefined
	      symbol FallbackDefineSelf().

	    - Shared library compilation broke completely due to
	      changes in NetAudio.tmpl and because the include of
	      <Library.tmpl> was moved too high up in
	      lib/audio/Imakefile, and missed out on the
	      DoSharedLib/DoNormalLib flags etc.

** PATCH 01 **

Server
	o fixed a couple of problems with speed changes in the
	  SVR4 VOXware driver.

** Release 1.2b BETA **

Library
	o fixed possible bogus free in sound.c.

Clients
	o fixed auwave dropping core (due to bugs in the slider widget).

General
	o lots of stuff from Steven King.  See CHANGES.sk below for details.

	o The file server/PATCHES.sk contains patches from Steven King to
	  improve the handling of audio clipping in the server. I've
	  not applied them to the standard distribution yet because
	  I'm concerned about their possible performance impact.

	o added some documentation about actions.

** PATCH 01 **

Server
	o fixed a bug with looping flows.

Library
	o fixed problem with X11R4 build.

** Release 1.2a ALPHA **

Server
	o fixed a bug in ReadElement that would occasionally cause
	  protocol errors during recording.

	o wake up the server when there are events to send out.

	o fixed a bug in readBucket() that caused noise at the end of
	  a bucket.

	o get rid of minibuf padding to eliminate clicks.

	o optimized sample rate conversion.

	o added FreeBSD/Linux VOXware driver (thanks to Amancio Hasty).

	o added SVR4 VOXware driver (thanks to Stephen Hocking).

	o added line mode control to Sun dbri driver.

Library
	o add AuXtAppRemoveAudioHandler().

	o fixed memory leaks and other problems with the bucket attribute
	  cache.

	o fix some nits with waveform durations.

Clients
	o new auphone sample app.

	o added some simple examples.

General
	o much cleaning and various OS support fixes from
	  Ian M. Donaldson.  See CHANGES.labtam below for details.

** Release ?.? ALPHA (Feburary 11, 1994) **

Server
	o fixed a bug where the last little bit of audio data would be
	  lost.

	o fixed a bug in readClient where the buffer was being incorrectly
	  padded, possibly resulting in static at the end of the audio data.

	o fixed two track monitor elements.

	o Sun driver

		- fixed a bug that could cause the output to get jumbled.

		- wait for audio to drain after flows finish.

		- Sparc 10 dbri driver.

Library
	o added API for set and get close down mode.

	o added Amiga IFF/8SVX sound file support.

	o close the sound file before the callback is called for
	  AuSoundRecordToFile().

	o make ListBuckets add to the bucket cache.

	o added file format suffixes to the SoundFileInfo structure.

	o added AuSetString() macro.

	o fixed "strdup" bug in SoundCreate.

	o re-arranged some utility routines into separate modules.
	  (NOTE: possible incompatibility - AuMonitorRate and AuMonitorFormat
	         are no longer externed in audioutil.h)

	o add AuWaveFormToString() and AuStringToWaveForm().

	o add AuSoundRecordToData().

	o add stereo support for voc files (thanks to Steven King).

Clients
	o auedit

		- deal with sound files with an unknown number of samples
		  (like those from stdin).

		- add record capability.

	o utilities

		- various new utilities.

	o new auwave sample app.

	o new slider widget for volume/gain controls.

General
	o re-worked the includes of NetAudio.tmpl so it will (hopefully)
	  work on most platforms.

	o builds under SVR3.2 (thanks to Steven King).

** Release 1.1 **

Server
	o Protected malloc/free from audio signals.

	o Fixed setting parameters of waveforms and buckets.

	o New dia/dda interface.

	o Fixed a bug where the flow was re-compiled after
	  an event was requested even if that event didn't change
	  the state of the flow.

	o Fixed possible memory trashing in readClient.

	o Sun driver

		- Don't initialize output gain or internal/external speaker
		  mode.

		- Get rid of some 'clicks'.

		- Deal with the situation when there are no physical outputs
		  in a flow.  This fixes the core dumps when you record before
		  having played anything.

		- Fixed erroneous calling of AuProcessData() after changing
		  gain values.

		- Fixed various bugs in recording (how did it ever work?!)

	o New SGI driver (thanks to Reiner Ernst)

Library
	o Fixed a bug in AuConvertShortToData().

	o Fixed function prototypes in Xtutil.h.

	o Fixed a problem in Xtutil.c where events would not get processed
	  if they were enqueued while awaiting a reply.

Clients
	o auedit
		Fixed a bug with temporary and backup file names.

	o auconvert
		Fixed a bug with converting raw data.

General
	o Solaris 2.2 support


********************************************************************************
			      CHANGES.labtam
********************************************************************************

Many thanks to Ian M. Donaldson at Labtam for the following:

- Now compiles cleanly on all these without any compiler warnings
  whatsoever (well the odd one or two under X11R4).

	- Solaris 2.3 on SS1000 under Sun provided Open Windows 3.2
	  Server untested.
	      (We only have a SS1000 running 2.3 which has no audio device)

	  Porting required because Sun changed the audio interface
	  slightly in Solaris 2 versus SunOS 4.

	  Clients tested reasonably.

	- AIX 3.2 on a RS6000/220 under IBM provided X11R4
	  No server port done.  Don't have access to any IBM audio hardware.
	  Clients tested reasonably.

	- SunOS 4.1.2  on a Sun IPX under MIT provided X11R5.
	  (no porting required specific to this platform)
	  Both clients and server tested reasonably.

	- SVR4.0/486  on one of our own computers under MIT provided X11R5.
	  (as a guess comparable to Dell SVR4.0 or ESIX SVR4.0)
	  We used native STREAMSCONN TLI TCP/IP transport.
	  No server port done.  This computer never had audio hardware.
	  Clients tested reasonably.

	- SVR3.2/486  on one of our own computers under MIT provided X11R4.
	  This O/S uses Lachman provided TCP/IP, and we used
	  STREAMSCONN TLI TCP/IP transport.
	  No server port done.  This computer never had audio hardware.
	  Clients tested reasonably.

	  Unfortuantely every SVR3.2 system has some different
	  quirk since TCP was non-standard then.   No guarantees
	  it will compile better on other types.

- Bugfixes
	- server had byte swapping  error that crashed playbucket(1)
	  when client run from opposite endian host

	- autool crashed on little endian host when volume slider was
	  used when stopped (hung on big endian host)

	- various signal mis-use.  Background processes from non-job
	  control shells don't die when ^C hit anymore.

	- "SVR4" is not a reliable symbol for use in Imakefiles; use SystemV4
	  instead.  (broke audial).

	- in Netaudio.tmpl, don't redefine MKDIRHIER or BINDIR by
	  default as this breaks the bootstrap in subtle ways on
	  certain systems.  Audio clients now get installed in the
	  same directory that X clients do rather than /usr/local/bin.

	  On Solaris 2.3 with OpenWindows, this means /usr/openwin/bin,
	  otherwise it usually means /usr/bin/X11.
	  Similarly for header files, app-defaults and libraries.

	- build will not fail on systems that don't have servers ported
	  to them.  This allows the clients and libraries to be built
	  on all platforms and servers on a subset.

- libaudio is built as a shared library (currently rev 2.0) on systems
  where libX11 is built shared.  In the list above that means
  Solaris 2.3, SunOS 4.1, SVR4.0.
  I used rev 1.0 for the September 93 netaudio release but changes
  between then and netaudio 1.1 looked somewhat incompatable so I bumped
  the revision number then.

- X11R4 compatability re-added in so compiliation in those
  environments is now possible.

- STREAMSCONN support for SVR3.2 and SVR4.0 for libaudio.
  This is a direct port from MIT X11R5 with the odd fix-tracker applied.
  NB: Not done for auscope(1).  sockets used in that case.
  (this is ok but if people want binary compatability from SVR3.2
  to SVR4 sockets won't work because SVR3.2 has no standard socket ABI)

- Better System-V and SVR4 support

	- Use System-V reliable signals.  (ie: sigset(2)).
	  signal(2) is unreliable on System-V systems.
	  Didn't bother with SVR4/POSIX signals; SVR3 signals work
	  fine on SVR4 and are easier to understand and code.

	- don't use libucb at all on SVR4.  This library is bad news
	  on most SVR4 systems and should be avoided like the plague!

- Better ANSI support (SVR4 + Solaris)

	- use ANSI memcpy/memset exclusively on ANSI systems instead
	  of bcopy/bzero

	- eliminated all SVR4.0/486 cc's reported warnings of callback
	  function type declaration mismatches.  (this is the bulk
	  of the changes).  Solaris 2 cc doesn't report half these things
	  unfortunately.

- auscope(1) now has a '-n' flag so it can be run in background reliably.

- 'make install' now fully installs correctly.  Even app-defaults!

- 'make clean' now properly cleans

********************************************************************************
			      CHANGES.sk
********************************************************************************

Thanks to Steven King for the following:

- building on ISC

    I had a few problems building 1.2a on my system running ISC SVR3.2r4.0
    with the imake enviroment provided by XFree86 (It looks to be more
    work than its worth to get netaudio to build with the vendor supplied
    X11R4).

    In server/dia/au.h, ISC doesnt have memmove, however bcopy and friends
    are in the normal library search path.

    In lib/audio/Alibnet.h, the omission of the #define for _AuWriteV
    resulted in link errors for all clients.

    In lib/audio/Imakefile, the SYSV_DEFINES macro causes major grief if
    one has installed XFree, I would expect this to be true for any SVR32
    that has XFree.  Also, I found it usefull to include NetAudio.tmpl
    after Library.tmp so that I could redefine things like CDEBUGFLAGS and
    such in NetAudio.tmpl

    In clients/audio/audemo/audemo.c, ISC needed sys/time.h included, I
    suspect most SYSV's will as well.

    In clients/audio/audial/audial.c, the compile breaks on the inclusion
    of sys/file.h, I think this is a ISC 4.0 specific problem.

- localconn patches

    These patches are derived from the XFree86 2.0 release.  I pretty much
    just merged the XFree patches into netaudio.  My principle interest
    was in the streams localconn code for ISC; besides greatly improving
    the performance for local clients, they allow the server to run with
    the system at run level 2, or on systems where full networking isnt
    availible (an extra $$$ option on ISC and lots of other PC sysVs).
     In addition to the ISC code, theres also support for other systems, SVR4,
    SCO, linux and net/free bsd, which may or may not be usefull depending
    on the system (ie, is anyone doing an independent SCO port?).  There
    is also alot of code for a minix and amoeba port.  I can pretty much
    guarantee that this will not work, but I left it in as it doesnt affect
    the other systems and might give a leg up to anyone attempting a minix
    or amoeba port.

- NetAudio.tmpl

    Make clients get rebuilt after libaudio.a changes.