Here is a list of the changes and steps that I had to take to get 4.23f10 to
compile (and perform well) on Mac OS 9:

0.  (Replace the csound_src directory in Perf4.23 sources with contents of
the 4.23f10 tarball of course).

1.  Copy the gets.r file from the 4.23f07 tarball to the 4.23f10 folder.

2.  Correct two strings in gets.r which were missing their final quotation
marks '"' (String #s 1809 and 1812).

3.  Comment out (or #ifndef mills_macintosh) line 681 of insert.c:
  if (!POLL_EVENTS()) longjmp(cglob.exitjmp,1); /* PC GUI needs attention */

4.  Add the following line after line 35 in sftype.h:
  #  define     PACKED            // akozar, 022604

5.  Copy the file macmidirec.c from the Perf 4.23 source archive because the
CodeWarrior project file was looking for it.  (This may not be strictly
necessary since there are two other files named MacMidi.c which are
virtually identical).

6.  Create a dummy file named autoheader.h since most Csound source files
now include it.  (Would it be better to conditionally include this?)

Since I don't have developer priveleges, could someone please see that the
necessary changes be applied to CVS so that 4.23f11 will compile "out of the
box" for the Mac.

Thanks!

Anthony Kozar
anthony.kozar@utoledo.edu

Anthony,

It would be nice if the version of Csound in the csound module
supports Mac OS9.  I need your help to make this work.  Unless and
until you are granted develeper privileges, please perform an
anonymous checkout of the sources in the CVS repository.  Make the
needed changes, which must include adding an OS9 specific Makefile
into the csound directory.  Then send me the results of performing a
cvs diff with both the -u and -N options (unified format and include
new files).  I'll commit the improvements.

John

Anthony Kozar  writes:

> Here is a list of the changes and steps that I had to take to get 4.23f10 to
> compile (and perform well) on Mac OS 9:
> 
> 0.  (Replace the csound_src directory in Perf4.23 sources with contents of
> the 4.23f10 tarball of course).
> 
> 1.  Copy the gets.r file from the 4.23f07 tarball to the 4.23f10 folder.
> 
> 2.  Correct two strings in gets.r which were missing their final quotation
> marks '"' (String #s 1809 and 1812).
> 
> 3.  Comment out (or #ifndef mills_macintosh) line 681 of insert.c:
>   if (!POLL_EVENTS()) longjmp(cglob.exitjmp,1); /* PC GUI needs attention */
> 
> 4.  Add the following line after line 35 in sftype.h:
>   #  define     PACKED            // akozar, 022604
> 
> 5.  Copy the file macmidirec.c from the Perf 4.23 source archive because the
> CodeWarrior project file was looking for it.  (This may not be strictly
> necessary since there are two other files named MacMidi.c which are
> virtually identical).
> 
> 6.  Create a dummy file named autoheader.h since most Csound source files
> now include it.  (Would it be better to conditionally include this?)
> 
> Since I don't have developer priveleges, could someone please see that the
> necessary changes be applied to CVS so that 4.23f11 will compile "out of the
> box" for the Mac.
> 
> Thanks!
> 
> Anthony Kozar
> anthony.kozar@utoledo.edu

On 2/28/04 9:38 AM, John D. Ramsdell etched in stone:

> Anthony,
> 
> It would be nice if the version of Csound in the csound module
> supports Mac OS9.  I need your help to make this work.

It looks like John ff. has already made the changes in the new release
(4.23f11) that was posted over the weekend on Sourceforge.  (Thanks John!!).

However, I am a litte confused on the relationship between these releases
and the CVS repository.  Are the tarballs in the download section based on
CVS?  If not, do the changes from John ff. in the tarballs eventually make
their way into CVS?  How does this happen?

And a related question:  How are the bug fixes being made in the csound
module being incorporated into the csound5 module?

> Unless and
> until you are granted develeper privileges, please perform an
> anonymous checkout of the sources in the CVS repository.

I do not yet have dev access.  But I can do this if it is still necessary.

> Make the needed changes, which must include adding an OS9 specific Makefile
> into the csound directory.

The Mills version of Csound has always been compiled using the CodeWarrior
IDE.  CodeWarrior does not use makefiles.  It has binary project files
instead which store a list of files, targets, and settings (dependencies are
resolved automatically).  These files probably cannot be managed in CVS,
correct?

I only have CodeWarrior 4 and 8.  The other developers seem to have been
using either version 5 or 6.  The differences are that CW8 does not come
with support for building 68K Mac targets.  However, CW8 has the ability to
save its project files as XML (i.e. text which could be subjected to CVS),
but CW4 does not have this feature.  So, I could send you an XML project
file but that would mean possibly abandoning the 68K targets (and MacOS 7 as
well, which I still use on my laptop).  Obtaining an XML project file from
CodeWarrior 6 from someone else would probably be more ideal (CW6 was the
last to have 68K support, I believe).

The other option -- which I have been considering for awhile -- would be for
me to figure out how to build Csound with Apple's Macintosh Programmer's
Workshop (MPW).  This would be nice since it is available at no cost from
Apple (versus CW's $400 price tag).  MPW does have a make system, but it is
incompatible with Unix make.  Still, it would be no problem to store the
makefile in CVS.  Other benefits of MPW include being able to use gcc,
bison, sed, gawk, etc. as well as stdio redirection.  A drawback may be the
lack of certain Unix header files which may be necessary to build things
like libsndfile.  I haven't investigated quite enough yet to know for sure.

Another issue is the couple dozen Mac-specific source files that are not in
CVS at all.  These files create the Mac GUI for the application called
'Perf' and are currently required (although they are likely unneccessary for
Matt's CsoundLib version).  One of these files is a binary resource file.
Plus there is the entire code base for the Mac front-end application.  None
of these files are LGPL.  Matt has indicated that there is no license for
them which I suppose would make them "open source" for Sourceforge purposes.
It might be a good idea to create a new "MillsMacintosh" module in CVS to
store and manage all of these files.

> Then send me the results of performing a
> cvs diff with both the -u and -N options (unified format and include
> new files).  I'll commit the improvements.

I am a CVS newbie, but I am willing to try this if it is still needed.

Thanks for your help.  I'm interested to hear what others think might be the
best solution here.  (Sorry for the lengthy reply with much unnecessary
detail no doubt).

Anthony Kozar
anthony.kozar@utoledo.edu

Anthony Kozar  writes:

> However, I am a litte confused on the relationship between these
> releases and the CVS repository.  Are the tarballs in the download
> section based on CVS?  If not, do the changes from John ff. in the
> tarballs eventually make their way into CVS?  How does this happen?

What is in the csound module of the csound CVS repository is an
attempt to support both a Csound distribution that can be built using
the original makefiles and one that is built using the usual
configure, make, make install sequence of commands, however, the
directory reorganization required to make the later distribution seems
to have interfered with John ffitch's ability to use it to maintain
the former distribution.  As such, it has become the place where the
GNU Build System (GBS) version of Csound 4.23 is maintained.
Distributions are released in the wxcsound package of the Csound
project.  In that package, the current source distribution is in
csound-4.23f10gbs.0.tar.gz.  John ffitch usually sends me instruction
on what to change, but the current arrangement could cause a
divergence between the two distributions.

John

Have a little free time yesterday I downloaded a complete csound4 and
performed diffs against my sources.  As a result I updated a number of
files with bug fixes and in one case a minor efficiency hack.  I have
not confirmed that all is still well.  The files should be in step
with my files, OEE.

On a different track, I am still working on tidying up LibSndFile
deployment.  I am now doing without wave.c, aiff.c, aifc.c, soundio.c
and a bunch of other code.  So far I am without the dither-on-output
option but am adding the other formats allowed in the library.

==John ffitch

Wow, this all sounds great!  I did a CVS update of csound4, compiled,
and installed without any hitches (used GNU build to compile).  I
noticed there were changes to the pluck opcode: I couldn't see what the
major changes were (it's late and my eyes are pretty blurry) but the
instruments I have that use pluck worked without any problems.  All
seems well to me.

Thanks very much!
steven

^_^


On Thu, 2004-03-04 at 22:58, jpff@codemist.co.uk wrote:
> Have a little free time yesterday I downloaded a complete csound4 and
> performed diffs against my sources.  As a result I updated a number of
> files with bug fixes and in one case a minor efficiency hack.  I have
> not confirmed that all is still well.  The files should be in step
> with my files, OEE.
> 
> On a different track, I am still working on tidying up LibSndFile
> deployment.  I am now doing without wave.c, aiff.c, aifc.c, soundio.c
> and a bunch of other code.  So far I am without the dither-on-output
> option but am adding the other formats allowed in the library.
> 
> ==John ffitch
> 
> 

On 3/5/04 1:58 AM, jpff@codemist.co.uk etched in stone:

> Have a little free time yesterday I downloaded a complete csound4 and
> performed diffs against my sources.  As a result I updated a number of
> files with bug fixes and in one case a minor efficiency hack.  I have
> not confirmed that all is still well.  The files should be in step
> with my files, OEE.

Great!  Thanks.  I will confirm that it works for me tomorrow.  (Like
Steven, I am getting blurry-eyed ...)

> On a different track, I am still working on tidying up LibSndFile
> deployment.  I am now doing without wave.c, aiff.c, aifc.c, soundio.c
> and a bunch of other code.  So far I am without the dither-on-output
> option but am adding the other formats allowed in the library.

More options, yea!  If you need more single-character option flags, I think
we could probably change a number of the "Mac-only" options to be
"--whatever" style flags.  Some of the existing options aren't even being
used and there is no need to clutter up the command-lines of platforms that
actually care about such things!

Can you tell that it is late?
:)

Anthony Kozar
anthony.kozar@utoledo.edu

I did not merge the Changeling, but the changes to wgpluck were to
remove unused code and to initialse an unitialise field.  Changes to
repluck was intonation -- not yet finished. 
==John ffitch

2004-03-03  John ffitch  

	* scxtract.c (scxtract): Added call to sread_init() as suggested
	by Anthony Kozar

2004-02-27  John ffitch  

	* version.h: VERSION 4.23f11

	* insert.c (kperf): Removed call to POLL_EVENTS which was OTT

2004-02-24  John ffitch  

	* dumpf.c (nkread): Reading of text numbers was wrong.

2004-02-20  John ffitch  

	* grain.c (ags): Use memcpy rather than a copy to copy back

2004-02-03  John ffitch  

	* rtlinux.c (recopen_): Removed setting of ishift
	(playopen_): and oshift as not used, and hence getshift is
	removed.

2004-01-29  John ffitch  

	* version.h: VERSION 4.23f10

	* insert.c (ingoto): At matt's suggestion check that the boolean
	is i-time

2004-01-07  John ffitch  

	* insert.c (ingoto): Should set ids not pds

	* ugrw1.c: Changes to sprints so rounding works for negative values

2004-01-01  John ffitch  

	* wavegde.c (filter3Set): Need to initialise filt->x1 and ->x2 to zero

	* flanger.c (wguide2): Avoid frequencies below 5Hz (reciprocal of
	MAXDELAY) 

	* version.h (BUGFIX): fix 8 -- avoiding octal feature!

	* OSC-sock.c: 
	* metro.c: Add casts to avoid warnings

	* flanger.c (wguide1): Avoid frequencies below 5Hz (reciprocal of
	MAXDELAY) 

John,

Thank you for your contribution.  I did a "cvs diff -u" between the
current contents of the csound module and the version tagged
csound-4_23f10_1.  Everything looked fine except that the ChangeLog
contained none of your contributions!  I added the content given in
your message, and released the distribution.   Thanks again.

John

jpff@cs.bath.ac.uk writes:

> I did not merge the Changeling, but the changes to wgpluck were to
> remove unused code and to initialse an unitialise field.  Changes to
> repluck was intonation -- not yet finished. 
> ==John ffitch
> 
> 2004-03-03  John ffitch  
> 
> 	* scxtract.c (scxtract): Added call to sread_init() as suggested
> 	by Anthony Kozar
> 
> 2004-02-27  John ffitch  
> 
> 	* version.h: VERSION 4.23f11
> 
> 	* insert.c (kperf): Removed call to POLL_EVENTS which was OTT
> 
> 2004-02-24  John ffitch  
> 
> 	* dumpf.c (nkread): Reading of text numbers was wrong.
> 
> 2004-02-20  John ffitch  
> 
> 	* grain.c (ags): Use memcpy rather than a copy to copy back
> 
> 2004-02-03  John ffitch  
> 
> 	* rtlinux.c (recopen_): Removed setting of ishift
> 	(playopen_): and oshift as not used, and hence getshift is
> 	removed.
> 
> 2004-01-29  John ffitch  
> 
> 	* version.h: VERSION 4.23f10
> 
> 	* insert.c (ingoto): At matt's suggestion check that the boolean
> 	is i-time
> 
> 2004-01-07  John ffitch  
> 
> 	* insert.c (ingoto): Should set ids not pds
> 
> 	* ugrw1.c: Changes to sprints so rounding works for negative values
> 
> 2004-01-01  John ffitch  
> 
> 	* wavegde.c (filter3Set): Need to initialise filt->x1 and ->x2 to zero
> 
> 	* flanger.c (wguide2): Avoid frequencies below 5Hz (reciprocal of
> 	MAXDELAY) 
> 
> 	* version.h (BUGFIX): fix 8 -- avoiding octal feature!
> 
> 	* OSC-sock.c: 
> 	* metro.c: Add casts to avoid warnings
> 
> 	* flanger.c (wguide1): Avoid frequencies below 5Hz (reciprocal of
> 	MAXDELAY) 

On 3/5/04 4:23 AM, Anthony Kozar etched in stone:

> On 3/5/04 1:58 AM, jpff@codemist.co.uk etched in stone:
> 
>> Have a little free time yesterday I downloaded a complete csound4 and
>> performed diffs against my sources.  As a result I updated a number of
>> files with bug fixes and in one case a minor efficiency hack.  I have
>> not confirmed that all is still well.  The files should be in step
>> with my files, OEE.
> 
> Great!  Thanks.  I will confirm that it works for me tomorrow.

Well, tomorrow has come, and I am very happy to report that the Csound4 CVS
snapshot that I checked out on Mar. 7th compiled like a charm with zero
changes!  The problems that I reported seem to be fixed as well.

Thanks, John, for the changes.

Anthony Kozar
anthony.kozar@utoledo.edu