The following are some notes on the current state as it is some months
since the last report.  I am grateful to Michael Gogins and Steven Yi
for their input to this report (and indeed to the whole project), but
any errors or omissions are mine.
==John ffitch

REPORT ON CSOUND5 PROJECT -- SEPT 2004

a) Parser

   Some internal reworking but this is not integrated and is not
   progressing as it should

b) Sound file library

   Apart from dnoise this is believed to be completed

c) Completion of Portaudio v19

   Runs acceptably on Linux.  OSX causes problems as the blocking code
   for PA is missing.  A callback mechanism has been written.  Still
   some problems.  On Windows there are no problems with ASIO, and the
   latency is excellent, comparable to Linux.  There remains a problem
   with breaking out of an ASIO performance, often causing the
   computer to hang.  There is no problem if the performance finishes
   normally. 

d) Software bus

   The code is written.  Perhaps the names need to be revised, and
   there is currently no API code using it.  There are alternative
   ideas so the current implementation cannot be taken as settled.

e) Check mac build

   Not sure of Mac status; there are conflicting reports.  It builds
   for me but I have not stressed it at all

f) Check Windows build

   The scons build on Windows appears to be be in pretty good shape,
   except the NSIS installer target suggested by Steven Yi has not yet
   been implemented.  Also, CsoundVST built with useDouble=1 used to
   function in Cubase SX and now does not (must be built with float
   samples to work), although useDouble=1 works in all other contexts.

g) Reorganisation of opcodes

   The move to a plugin-based architecture is essentially complete.
   This could be revisited, but this is low priority.

h) separation of API

   Largely complete.  Still in minor development

j) MIDI

   A first attempt at deploying portMIDI has been made, but the whole
   MIDI area is not substantially changed.  This needs to be improved
   before we can release.

k) Making re-entrant

   Slow progress; needs some higher priority given to it.  There are
   some positive suggestions on a mechanism.

l) Loris opcodes

   Loris opcodes are functional with CsoundVST.  Michael has ported
   them to Csound5.  Steven Yi has suggestions for improving the Loris
   target for non-CsoundVST builds.

m) Python opcodes

   The Python opcodes are functional, and Oeyvind Brandtsegg is
   playing with these opcodes and providing feedback on them.

n) Documentation

   Micael has begun working on folding the new documentation on Csound
   5 found in csound.pdf into the Kevin Conder documentation system.
   When this is done the documentation target from scons willbe
   removed (or changed...).  There will still be a csound.pdf, but it
   will produced by DocBook and contain all Csound documentation in
   one place, including the Conder manual, up to date installation and
   build instructions, new features, extending Csound with plugin
   opcodes, and documentation of the API.

p) SConstruct changes

   File to be reorganized to have build targets as well as to handle 
   collecting data for installers.  install-opcodes, install-executables, 
   and install targets already integrated.

q) Installers

   Not much progress here, mostly dependent on SConstruct changes 

And no doubt there are other things I have forgotten
==JPff

 previous> May 2004:

 previous>  a) Parser
 previous>     ------
 previous>     Little or no progress.  I have improved lexer and it is now
 previous>     compatible with loadable opcodes.

 previous>  b) Sound file library
 previous>     ------------------
 previous>     This is mainly done.  Some of the utilities (in particular dnoise)
 previous>     have not been converted yet

 previous>  c) Completion of Portaudio v19
 previous>     ---------------------------
 previous>     There have been major improvements here.  On Linux we have a
 previous>     working system with some concerns about performance.  On Windows
 previous>     Portaudio suffers from high latency, and at present the ASIO
 previous>     variant is not functional.  The mac is even more problematic as
 previous>     the pa_v19 port to OSX is far from done.

 previous>  d) Software bus
 previous>     ------------
 previous>     Designs have been floating about but I do not get a clear feeling
 previous>     on consensus.  The proposals have been mainly easy to code, so it
 previous>     is a policy issue

 previous>  e) Check mac build
 previous>     ---------------
 previous>     Thanks to steven and others there is progress here.  It is up to
 previous>     them to report.  No progress of integrating a Mac-like user
 previous>     interface onto the API (I think) but a simple command-line build
 previous>     works in delayed time.

 previous>  f) Check Windows build
 previous>     -------------------
 previous>     I think Michael has sthis under control now; I do not have the
 previous>     space to build on Windows now

 previous>  g) Reorganisation of opcodes
 previous>     -------------------------
 previous>     This needs review.  The code works, but I am not totally happy
 previous>     with the specification method for loading.  There is a suggestion
 previous>     that there should be a specified core, and the rest should be
 previous>     loadavle but there is no progress here

 previous>  h) separation of API
 previous>     -----------------
 previous>     I think this is progressing welll there are uncertainties about
 previous>     division into external/internal parts which needs more thought.

 previous>  j) MIDI
 previous>     ----
 previous>     Little or no progress has been made with PortMIDI; some older MIDI
 previous>     code is still working

 previous>  k) Simple build for Linux
 previous>     ----------------------
 previous>     Awaits software bus in part

 previous>  l) Making re-entrant
 previous>     -----------------
 previous>     Progress but needs a further push; there may be significant things
 previous>     to do here.

May I suggest that we also formally add increasing the maximum ftable size
to the agenda for Csound 5?  And if someone REALLY feels brave, they could
remove the power of 2 size restriction as well.

Anthony Kozar
anthony.kozar@utoledo.edu
http://akozar.spymac.net/


On 9/28/04 2:30 AM, jpff@codemist.co.uk etched in
stone:

> The following are some notes on the current state as it is some months
> since the last report.  I am grateful to Michael Gogins and Steven Yi
> for their input to this report (and indeed to the whole project), but
> any errors or omissions are mine.
> ==John ffitch

[...]

> And no doubt there are other things I have forgotten
> ==JPff