What I discovered is that apparently the Windows binary runs fine (and with
ASIO drivers only) when there is *no* separate portaudio.dll (or similar)
present. (I did a search for all files with "portaudio" in their names,
changed their names, and Csound5 still ran fine.) I gather that whatever the
Windows binary needs with respect to audio is self-contained; or is it
something else?

I also gather that Victor's enhancements (relating to additional drivers)
are *not* included in Michael's new release. If this is so, when might they
be - or is there some problem with this? (I'm wondering particularly about
the current limitation of -b and -B to ksmps - which I also hope could soon
be dealt with. OTOH, I'm not the one doing the work!)

Thanks, folks, for all your wonderful contributions.

Art Hunkins

----- Original Message -----
From: 
To: "Csound Developers Discussion List" 
Sent: Wednesday, December 15, 2004 10:36 AM
Subject: [CSOUND-DEV:5536] Re: New file release of Csound 5 and CsoundVST
for Windows


The 2.7 meg is debug info. You can strip it.

Original Message:
-----------------
From: Art Hunkins abhunkin@uncg.edu
Date: Tue, 14 Dec 2004 22:44:04 -0500
To: csound-dev@eartha.mills.edu
Subject: [CSOUND-DEV:5534] Re: New file release of Csound 5 and CsoundVST
for Windows


Victor,

I tried your suggestion - substituting the MyCsound5 portaudio.dll for that
in Michael's Csound5.

Believe it or not, both worked fine and the same; that is, both allowed ASIO
drivers only. (This unless I did something wrong in the attempted
substitution - but I don't think so.)

What seems particularly strange to me is that the MyCsound5 portaudio.dll is
77K, while that of Csound5 blossoms to 2.7M.

So I gather (hesitantly) that it must take something besides just the
portaudio code to implement other driver models. Any further ideas as to how
to get more drivers into Csound5?

Art Hunkins

----- Original Message -----
From: "Victor Lazzarini" 
To: "Csound Developers Discussion List" 
Sent: Monday, December 13, 2004 12:23 PM
Subject: [CSOUND-DEV:5532] Re: New file release of Csound 5 and CsoundVST
for Windows


> I haven't checked the new release, but I'd say it just needs a version of
> portaudio.dll built for all types of devices (MME, WDM (on w2k) and ASIO).
> If the library is not statically linked as in the previous version, it's
> probably
> just a case of trying it with a different DLL (Art you can try with the
> portaudio.dll
> that comes with mycsound5, however because it's msvc-built, it might clash
> with the migwin binary).
>
> Victor
>
>
> At 16:38 13/12/2004, you wrote:
> >Michael,
> >
> >I installed your new Windows file release, which seems to be working fine
> >with ASIO drivers on my fairly updated machine.
> >
> >Only one install hitch: Error message "A required .DLL file,
> >LIBSNDFILE-1.DLL, was not found." Changing the filename of libsndfile.dll
to
> >libsndfile-1.dll fixed this.
> >
> >A realtime run is now aborted cleanly by CTRL-C in the main performance
> >window, or by closing same window (it shuts down FLTK windows at the same
> >time - which is very nice).
> >
> >One caveat: don't try to abort the run with CTRL-BREAK; the system will
lock
> >up. (If this could be "fixed," it would be nice.)
> >
> >I had hoped maybe you'd try to get the non-ASIO drivers working again, as
I
> >think Victor has done in his MyCsound5 release (perhaps I am mistaken
about
> >this). If I *am* correct, perhaps Victor can pass on some hints as to how
he
> >did it?
> >
> >I'll get out my slower machine to see if ASIO will run on it - but from
my
> >previous experience, I very much doubt it. (As you'll recall, latency is
not
> >an important consideration for me.)
> >
> >Art Hunkins
> >
> >----- Original Message -----
> >From: 
> >To: "Csound Developers Discussion List" 
> >Sent: Sunday, December 12, 2004 8:44 PM
> >Subject: [CSOUND-DEV:5525] New file release of Csound 5 and CsoundVST for
> >Windows
> >
> >
> >This is a new build of Csound 5 and CsoundVST with prebuilt binaries for
> >Windows with MinGW, and sources for all platforms.
> >
> >     csoundvst package at http://www.sourceforge.net/projects/csound
> >         csound5-mingw-2004-12-12.zip Windows binaries and all sources
> >         csound-2004-12-12.pdf Csound reference manual and API
documentation
> >
> >Built with updated versions of libsndfile, PortAudio, FLTK, and Loris.
> >
> >New features include user-defined function tables in Csound, and a
> >counterpoint generator in CsoundVST. Experimental PortMIDI support.
> >
> >Several bug fixes. Removed most compiler warnings (with MinGW, anyway).
> >
> >Changes:
> >
> >2004-12-12  Michael Gogins 
> >
> >* Updated to FLTK 1.1.6, libsndfile 1.1.11, CVS PortAudio,
> >CVS Loris for Windows pre-built binaries.
> >
> >* Added Loris API documentation to Csound reference manual.
> >
> >2004-12-10  John ffitch  
> >
> >* InOut/libsnd.c: and code
> >
> >* OOps/sndinfUG.c (getsndinfo): Removed unused called to readheader
> >
> >2004-12-03  Michael Gogins 
> >
> >* Tried to remove most remaining compiler warnings: mostly returning
> >int from main, sscanf for %lf with USE_DOUBLE, type casting, etc.
> >
> >2004-12-02  John ffitch  
> >
> >* SConstruct (else): Added PortMIDI build option to aid experimentation.
> >
> >2004-11-30  John ffitch  
> >
> >* Top/main.c (signal_handler): Move message to after check for alarm
> >
> >2004-11-25  Michael Gogins 
> >
> >* Initial implementation of CounterpointNode --
> >generates species 1, 2, or 3 counterpoint for cantus firmi
> >selected from notes generated by child nodes.
> >Based on code from Bill Schottstaedt's species counterpoint generator
> >(thanks, Bill!).
> >
> >* Optimized the build for boost::ublas matrix code
> >in CsoundVST Node classes.
> >
> >2004-11-28  John ffitch  
> >
> >* OOps/pvfileio.c (write_pvocdata, pvoc_readfmt): Use a union
> >rather than dubious casts.
> >
> >2004-11-23  John ffitch  
> >
> >* OOps/midirecv.c (m_chanmsg): Removed initial values for
> >volume/balance/pan/expressions/bend/mastvol
> >
> >2004-11-16  John ffitch  
> >
> >* OOps/ugens2.c (itable, itabli, itabl3): Need to check OK rather
> >than assume true=OK
> >
> >* Engine/fgens.c: All over removed statics into the environment or
> >into arguments.  Make ftfind etc take extra argument.  Should
> >assist with reentrance.
> >
> >2004-11-14  John ffitch  
> >
> >* util2/exports/het_import.c (main): Open output as binary stream
> >
> >* Top/main.c (mainRESET): ftRESET takes environment argument
> >
> >* Engine/fgens.c: Move many statics into a single static.  Adjust
> >args to gens to reflect this -- statics almost unnecessary
> >
> >2004-11-11  Michael Gogins 
> >
> >* Changed the Python interface to CsoundVST to allow Python
> >programmers to derive new classes in Python from CsoundVST.Node,
> >and these new classes will function as C++ Node * objects in the
> >CsoundVST C++ internals. This enables Python-derived Nodes to
> >function just like regular C++ Nodes in music graphs.
> >
> >2004-11-11  John ffitch  
> >
> >* Engine/fgens.c: Removed static variables ftp and nargs and used
> >arguments instead.  Simplifications all over
> >
> >2004-11-10  John ffitch  
> >
> >* H/csdl.h (FLINKAGE): defined
> >
> >* Top/dl_opcodes.c: Allow definitions of fgens
> >
> >* H/csoundCore.h: Added NGFENS structure
> >
> >2004-11-09  John ffitch  
> >
> >* Engine/fgens.c (fgens): Started named fgens
> >(fterror): Allow named gens in error messages
> >(allocgen): New function to allocate new names
> >
> >
> >
> >--------------------------------------------------------------------
> >mail2web - Check your email from the web at
> >http://mail2web.com/ .
>
> Victor Lazzarini
> Music Technology Laboratory
> Music Department
> National University of Ireland, Maynooth
>


--------------------------------------------------------------------
mail2web - Check your email from the web at
http://mail2web.com/ .

At 16:40 15/12/2004, you wrote:
>What I discovered is that apparently the Windows binary runs fine (and with
>ASIO drivers only) when there is *no* separate portaudio.dll (or similar)
>present.

Aha! Portaudio has been statically linked.

>I also gather that Victor's enhancements (relating to additional drivers)
>are *not* included in Michael's new release. If this is so, when might they
>be - or is there some problem with this? (I'm wondering particularly about
>the current limitation of -b and -B to ksmps - which I also hope could soon
>be dealt with. OTOH, I'm not the one doing the work!)

There were no enhacements. I just built portaudio for all driver types and
dynamically linked it.

Victor

I noticed that the pmidi.c source in the latest releases has
not being changed to my proposed fixes. Is there any reason
for that? I know that there were some questions about it, but
the current portmidi implementation is broken and my fixes
addressed the problems, even if in a provisory form.

Without portmidi, there is no MIDI support for ALSA, so I'd
suggest that my fixes should be included before the whole
code is reviewed.

Victor



Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 

Could any of the other cvs developers upload some of my
changes to the CVS repository? Apart from two, these changes only affect
MSVC builds, so they are safe for everyone else.

The other two are:

(a) in libsnd.c commented out line 362,
  rtin_dev = 0;
  this makes the default value 0, rather 1024, with implications in rtpa.c
  I think it's a safe change

(b) in oload.c line 404
    {}  becomes {0}

    {} doesn't compile in MSVC, but {0} seems to initialise FGDATA ok.

Anyway, if anyone can upload these files, please let me know and I'll
send them to you:

csoundcore.h
sysdep.h
widgets.cpp
ugens1.cpp
main.c
libsnd.c
oload.c
mididevice.c

It'll mean I don't need to be editing these files every time
I get a new updated archive from the tarball or from the releases.

Thanks

Victor


Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 

I notice that some changes were made to enable plugin ftables. These
might have introduced a bug, which is leading to a crash on my windows
msvc build.

Basically, using the opcode ftgen seems to crash csound, whereas score-defined
tables are OK.

However, I tested this with Michael's mingw binary and it seems to run OK . 
Howvever,
with his binary I am experiencing a crash with a FOF-based instrument, only
when using realtime audio -odac, a crash that of course happens only 
intermittently
with my binary, aagh!

Anyway, the ftgen problem I identified in line 2234 of fgens.c:

(*csound->gensub[genum])(ftp, csound);     /* call gen subroutine  */

which when commented out, doesn't crash csound.

I'll investigate, but perhaps you might have some clues as to what is the
problem. Ah, one main difference between my binary and Michael's is that
I am using float samps instead of doubles; but it shouldn't matter.

Victor


Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 

It appears that csound->gensub[genum]  is not set to any gen function,
so it causes the crash. It's only happening in hfgens(), in fgens() is
business as usual, no problems... csound->gensub[genum] has the
right function.

Victor



Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 

I know what's happening now with ftgen: if there is a plugin
ftable lib in the OPCODEDIR, it works. Otherwise it doesn't,
and in Michael's release there is a ftest.dll, which I didn't
build on mine.

Now that's a bug. I don't know exactly what to do, but in fgens()
there is code to load the gen functions array, which is missing
in hfgens(). I suppose John Ffitch will know what to do, so I won't
bother finding out.

all the best,

Victor
At 15:54 17/12/2004, you wrote:

>It appears that csound->gensub[genum]  is not set to any gen function,
>so it causes the crash. It's only happening in hfgens(), in fgens() is
>business as usual, no problems... csound->gensub[genum] has the
>right function.
>
>Victor
>
>
>
>Victor Lazzarini
>Music Technology Laboratory
>Music Department
>National University of Ireland, Maynooth

Victor Lazzarini
Music Technology Laboratory
Music Department
National University of Ireland, Maynooth 

Mea culpa; think fixed now. (need initial allocation in hfgens)
==John ffitch