Hello everyone,

I've written a Processing/Csoundo sketch to test creating, reading and
writing from a table using CsoundPerformanceThread. I'm intentionally
running a lot of number through to test the load on the system, and my
results shows the input doesn't match the output roughly 2% - 3% of
the time. I'm using CsoundPerformanceThread which supposedly protects
memory and mutex locks thing down automatically. Though
CsoundPerformanceThread has the method inputMessage(), it doesn't have
methods for reading and writing tables directly. Which I'm guessing
means that they are not thread safe.

I have tried using mutex. Though the mutex used by
CsoundPerformanceThread (* queueLock) is private according to the API.
Since I haven't found a way to get that one, I created my own mutex
with csnd.csoundCreateMutex(1) and tried wrapping Csoundo's table
access methods like this:

mutex.lock();
csound.TableSet(t, i, v);
mutex.unlock();


I'm unsure this is even the proper way to use mutex.
Enabling/disabling my implementation of mutex seems to have zero
affect on the outcome.

I've posted the main loop from my sketch towards the bottom. In a
nutshell, this is what happens. Every time draw() is called, a zero-d
f-table is generated. Then for each index in the f-table, the
following happens:

1) A sine wave point is created.
2) The point is written to the current index of the f-table with cs.tableSet()
3) The value is then read from the f-table with cs.tableGet()
4) If the input != the output, an error message is printed. Most
often, the output is 0.0, which indicates that setChn() was not called
by the time getChn() grabbed its value from the f-table. Example:
0.0 != -0.5027427 @index 3825
5) Display the table, to look for the errors visually.  Example:
http://www.thumbuki.com/temp/CsoundoSetGetTableError.jpg


Here's a key to my Csoundo methods:

cs.event() calls CsoundPerformanceThead.inputMessage()
cs.tableLength() csound.TableLength();
cs.tableGet() calls csound.TableGet();
cs.tableSet() calls csound.TableSet();
mutex.lock() calls csnd.csoundLockMutex();
mutex.unlock() calls csnd.csoundUnlockMutex();

Here's the code:

void draw() {
    noStroke();
    fill(0, 8);
    rect(0, 0, width, height);

    // Create empty table
    float size = pow(2, 12);
    cs.event("f 20 0 " + size + " 10 0");
    int length = cs.tableLength(20);

    // Increment phase of sine table input
    int inc = 60;
    float phase = (float) (frameCount % inc) / inc * TWO_PI;

    // Create, write, read and draw sine waveform
    stroke(255, 255);
    beginShape();

    for (int i = length - 1; i >= 0; i--) {

        // Generate sine data point and input into table
        float input = sin(TWO_PI * (float) i / (float) length + phase);
        cs.tableSet(20, length, input);

        // Read the from the table at the same index
        float output = cs.tableGet(20, length);

        // Print message if input != output
        if (output != input) {
            println(output + " != " + input + " @index " + i);
            bugs++;
            println("fail rate = " + ((float) bugs / (float) frameCount));
        }

        // Draw wave form
        vertex((float) i / (float) (length - 1) * (float) width,
        height / 2.0 + height / 2.0 * output);
    }

    endShape();
}

So I guess my question is, does this mean I have to abandon
CsoundPerformanceThread in favor of csound.csoundPerformKsmps()? Or am
I just completely botching this in my code? Really, anything any of
you have to say would help.

Best,
Jake

Hello Jacob

I think that if you want to protect csound execution fron any changes
(change table, channel), and that these changes are not available as
input message for CsoundPerformanceThread, you have to do what
CsoundPerformanceThread does in the Perform method.

Recreate it in Java this class, and extend the message system with new
java classes for table and channels.

I think also there is a mutex class in Java (java.util.concurrent) to
help this tricky subject of thread synchronization.

jp

On Fri, Aug 27, 2010 at 7:28 PM, Jacob Joaquin  wrote:
> Hello everyone,
>
> I've written a Processing/Csoundo sketch to test creating, reading and
> writing from a table using CsoundPerformanceThread. I'm intentionally
> running a lot of number through to test the load on the system, and my
> results shows the input doesn't match the output roughly 2% - 3% of
> the time. I'm using CsoundPerformanceThread which supposedly protects
> memory and mutex locks thing down automatically. Though
> CsoundPerformanceThread has the method inputMessage(), it doesn't have
> methods for reading and writing tables directly. Which I'm guessing
> means that they are not thread safe.
>
> I have tried using mutex. Though the mutex used by
> CsoundPerformanceThread (* queueLock) is private according to the API.
> Since I haven't found a way to get that one, I created my own mutex
> with csnd.csoundCreateMutex(1) and tried wrapping Csoundo's table
> access methods like this:
>
> mutex.lock();
> csound.TableSet(t, i, v);
> mutex.unlock();
>
>
> I'm unsure this is even the proper way to use mutex.
> Enabling/disabling my implementation of mutex seems to have zero
> affect on the outcome.
>
> I've posted the main loop from my sketch towards the bottom. In a
> nutshell, this is what happens. Every time draw() is called, a zero-d
> f-table is generated. Then for each index in the f-table, the
> following happens:
>
> 1) A sine wave point is created.
> 2) The point is written to the current index of the f-table with cs.tableSet()
> 3) The value is then read from the f-table with cs.tableGet()
> 4) If the input != the output, an error message is printed. Most
> often, the output is 0.0, which indicates that setChn() was not called
> by the time getChn() grabbed its value from the f-table. Example:
> 0.0 != -0.5027427 @index 3825
> 5) Display the table, to look for the errors visually.  Example:
> http://www.thumbuki.com/temp/CsoundoSetGetTableError.jpg
>
>
> Here's a key to my Csoundo methods:
>
> cs.event() calls CsoundPerformanceThead.inputMessage()
> cs.tableLength() csound.TableLength();
> cs.tableGet() calls csound.TableGet();
> cs.tableSet() calls csound.TableSet();
> mutex.lock() calls csnd.csoundLockMutex();
> mutex.unlock() calls csnd.csoundUnlockMutex();
>
> Here's the code:
>
> void draw() {
>    noStroke();
>    fill(0, 8);
>    rect(0, 0, width, height);
>
>    // Create empty table
>    float size = pow(2, 12);
>    cs.event("f 20 0 " + size + " 10 0");
>    int length = cs.tableLength(20);
>
>    // Increment phase of sine table input
>    int inc = 60;
>    float phase = (float) (frameCount % inc) / inc * TWO_PI;
>
>    // Create, write, read and draw sine waveform
>    stroke(255, 255);
>    beginShape();
>
>    for (int i = length - 1; i >= 0; i--) {
>
>        // Generate sine data point and input into table
>        float input = sin(TWO_PI * (float) i / (float) length + phase);
>        cs.tableSet(20, length, input);
>
>        // Read the from the table at the same index
>        float output = cs.tableGet(20, length);
>
>        // Print message if input != output
>        if (output != input) {
>            println(output + " != " + input + " @index " + i);
>            bugs++;
>            println("fail rate = " + ((float) bugs / (float) frameCount));
>        }
>
>        // Draw wave form
>        vertex((float) i / (float) (length - 1) * (float) width,
>        height / 2.0 + height / 2.0 * output);
>    }
>
>    endShape();
> }
>
> So I guess my question is, does this mean I have to abandon
> CsoundPerformanceThread in favor of csound.csoundPerformKsmps()? Or am
> I just completely botching this in my code? Really, anything any of
> you have to say would help.
>
> Best,
> Jake
> --
> The Csound Blog - http://csoundblog.com/
> Slipmat - http://slipmat.noisepages.com/
>
>
> Send bugs reports to the Sourceforge bug tracker
>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"
>
>


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

The main thing that you need to achieve is that the table access does  
not happen at the same time
as PerformKsmps(). CsoundPerformanceThread() will call PerformKsmps()  
in a loop in its performance thread. It also calls
a Performance callback. So you can use the Performance callback to  
call any functions that might not be called at the same time,
or to implement your locks. The way it looks, your mutex are not  
protecting anything, because CsoundPerformanceThread just
ignores it.

I need to have a look at your situation a bit more closely for a more  
detailed diagnosis, but that looks like the case to me.

Victor

On 27 Aug 2010, at 18:52, jean-pierre lemoine wrote:

> Hello Jacob
>
> I think that if you want to protect csound execution fron any changes
> (change table, channel), and that these changes are not available as
> input message for CsoundPerformanceThread, you have to do what
> CsoundPerformanceThread does in the Perform method.
>
> Recreate it in Java this class, and extend the message system with new
> java classes for table and channels.
>
> I think also there is a mutex class in Java (java.util.concurrent) to
> help this tricky subject of thread synchronization.
>
> jp
>
> On Fri, Aug 27, 2010 at 7:28 PM, Jacob Joaquin  
>  wrote:
>> Hello everyone,
>>
>> I've written a Processing/Csoundo sketch to test creating, reading  
>> and
>> writing from a table using CsoundPerformanceThread. I'm intentionally
>> running a lot of number through to test the load on the system, and  
>> my
>> results shows the input doesn't match the output roughly 2% - 3% of
>> the time. I'm using CsoundPerformanceThread which supposedly protects
>> memory and mutex locks thing down automatically. Though
>> CsoundPerformanceThread has the method inputMessage(), it doesn't  
>> have
>> methods for reading and writing tables directly. Which I'm guessing
>> means that they are not thread safe.
>>
>> I have tried using mutex. Though the mutex used by
>> CsoundPerformanceThread (* queueLock) is private according to the  
>> API.
>> Since I haven't found a way to get that one, I created my own mutex
>> with csnd.csoundCreateMutex(1) and tried wrapping Csoundo's table
>> access methods like this:
>>
>> mutex.lock();
>> csound.TableSet(t, i, v);
>> mutex.unlock();
>>
>>
>> I'm unsure this is even the proper way to use mutex.
>> Enabling/disabling my implementation of mutex seems to have zero
>> affect on the outcome.
>>
>> I've posted the main loop from my sketch towards the bottom. In a
>> nutshell, this is what happens. Every time draw() is called, a zero-d
>> f-table is generated. Then for each index in the f-table, the
>> following happens:
>>
>> 1) A sine wave point is created.
>> 2) The point is written to the current index of the f-table with  
>> cs.tableSet()
>> 3) The value is then read from the f-table with cs.tableGet()
>> 4) If the input != the output, an error message is printed. Most
>> often, the output is 0.0, which indicates that setChn() was not  
>> called
>> by the time getChn() grabbed its value from the f-table. Example:
>> 0.0 != -0.5027427 @index 3825
>> 5) Display the table, to look for the errors visually.  Example:
>> http://www.thumbuki.com/temp/CsoundoSetGetTableError.jpg
>>
>>
>> Here's a key to my Csoundo methods:
>>
>> cs.event() calls CsoundPerformanceThead.inputMessage()
>> cs.tableLength() csound.TableLength();
>> cs.tableGet() calls csound.TableGet();
>> cs.tableSet() calls csound.TableSet();
>> mutex.lock() calls csnd.csoundLockMutex();
>> mutex.unlock() calls csnd.csoundUnlockMutex();
>>
>> Here's the code:
>>
>> void draw() {
>>    noStroke();
>>    fill(0, 8);
>>    rect(0, 0, width, height);
>>
>>    // Create empty table
>>    float size = pow(2, 12);
>>    cs.event("f 20 0 " + size + " 10 0");
>>    int length = cs.tableLength(20);
>>
>>    // Increment phase of sine table input
>>    int inc = 60;
>>    float phase = (float) (frameCount % inc) / inc * TWO_PI;
>>
>>    // Create, write, read and draw sine waveform
>>    stroke(255, 255);
>>    beginShape();
>>
>>    for (int i = length - 1; i >= 0; i--) {
>>
>>        // Generate sine data point and input into table
>>        float input = sin(TWO_PI * (float) i / (float) length +  
>> phase);
>>        cs.tableSet(20, length, input);
>>
>>        // Read the from the table at the same index
>>        float output = cs.tableGet(20, length);
>>
>>        // Print message if input != output
>>        if (output != input) {
>>            println(output + " != " + input + " @index " + i);
>>            bugs++;
>>            println("fail rate = " + ((float) bugs / (float)  
>> frameCount));
>>        }
>>
>>        // Draw wave form
>>        vertex((float) i / (float) (length - 1) * (float) width,
>>        height / 2.0 + height / 2.0 * output);
>>    }
>>
>>    endShape();
>> }
>>
>> So I guess my question is, does this mean I have to abandon
>> CsoundPerformanceThread in favor of csound.csoundPerformKsmps()? Or  
>> am
>> I just completely botching this in my code? Really, anything any of
>> you have to say would help.
>>
>> Best,
>> Jake
>> --
>> The Csound Blog - http://csoundblog.com/
>> Slipmat - http://slipmat.noisepages.com/
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>            https://sourceforge.net/tracker/? 
>> group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>>
>>
>
>
> Send bugs reports to the Sourceforge bug tracker
>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>



Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

On Fri, Aug 27, 2010 at 11:01 AM, Victor Lazzarini
 wrote:
> So you can use the Performance callback to call any
> functions that might not be called at the same time,
> or to implement your locks.


I've been looking into this, and I'm not certain it's possible right
now. The docs I've been reading on SWIG indicates that special
handling is required for passing C pointers in functions.  For
example:

http://www.opensource.apple.com/source/swig/swig-4/swig/Examples/java/funcptr/index.html?txt

I do not see %constant, %callback or %nocallback directives in
java_interface.i. Which leads me to believe that it may not be
possible with the Csound Java API, or it's doing a great job of hiding
from me.

Do correct me if I'm wrong about this.

Best,
Jake

There is a helper class that should do the trick:  
CsoundCallbackWrapper, using directors. I am not sure how to use it,  
but it
might be  starting point for you.

If you get it working, then you can basically do this:

1. Implement a function to get the table data into a list, protected  
by mutex.
2. In the performance callback read this list and call the csound  
table functions, protected by mutex

This is similar to how CsoundPerformanceThread treats events.

Victor


On 28 Aug 2010, at 00:40, Jacob Joaquin wrote:

> On Fri, Aug 27, 2010 at 11:01 AM, Victor Lazzarini
>  wrote:
>> So you can use the Performance callback to call any
>> functions that might not be called at the same time,
>> or to implement your locks.
>
>
> I've been looking into this, and I'm not certain it's possible right
> now. The docs I've been reading on SWIG indicates that special
> handling is required for passing C pointers in functions.  For
> example:
>
> http://www.opensource.apple.com/source/swig/swig-4/swig/Examples/java/funcptr/index.html?txt
>
> I do not see %constant, %callback or %nocallback directives in
> java_interface.i. Which leads me to believe that it may not be
> possible with the Csound Java API, or it's doing a great job of hiding
> from me.
>
> Do correct me if I'm wrong about this.
>
> Best,
> Jake
> -- 
> The Csound Blog - http://csoundblog.com/
> Slipmat - http://slipmat.noisepages.com/
>
>
> Send bugs reports to the Sourceforge bug tracker
>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>



Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

I don't know the Java syntax for this, but by looking at the C++  
source I reckon what you have to do is:

1. Derive a class from CsoundCallbackWrapper
2. Implement one or more of the functions, which will be your callbacks
     Here you'll be looking to implement the  
CsoundCallbackWrapper::YieldCallback(), which is similar placed to  
PerformanceCallback, but
instead of called by CsoundPerformanceClass, it is called by  
PerformKsmps() at ksmp boundaries.
3. Instantiate the object, pass it a Csound object.
4. Call its SetYieldCallback() method

This should be all you need to do. Any " host data"  your callback  
needs might be placed as class members to your derived class; you will  
not
be able to use 'csoundHostData' to hold host data, because this  
mechanism already uses it to pass the callback wrapper class object to  
Csound.
But you can just add any host data to your callback wrapper class.

This is absolutely untested, since I really know no Java and for  
Python, which I use, I have wrapped things differently in the swig  
interface.

Victor

On 28 Aug 2010, at 10:30, Victor Lazzarini wrote:

> There is a helper class that should do the trick:  
> CsoundCallbackWrapper, using directors. I am not sure how to use it,  
> but it
> might be  starting point for you.
>
> If you get it working, then you can basically do this:
>
> 1. Implement a function to get the table data into a list, protected  
> by mutex.
> 2. In the performance callback read this list and call the csound  
> table functions, protected by mutex
>
> This is similar to how CsoundPerformanceThread treats events.
>
> Victor
>
>
> On 28 Aug 2010, at 00:40, Jacob Joaquin wrote:
>
>> On Fri, Aug 27, 2010 at 11:01 AM, Victor Lazzarini
>>  wrote:
>>> So you can use the Performance callback to call any
>>> functions that might not be called at the same time,
>>> or to implement your locks.
>>
>>
>> I've been looking into this, and I'm not certain it's possible right
>> now. The docs I've been reading on SWIG indicates that special
>> handling is required for passing C pointers in functions.  For
>> example:
>>
>> http://www.opensource.apple.com/source/swig/swig-4/swig/Examples/java/funcptr/index.html?txt
>>
>> I do not see %constant, %callback or %nocallback directives in
>> java_interface.i. Which leads me to believe that it may not be
>> possible with the Csound Java API, or it's doing a great job of  
>> hiding
>> from me.
>>
>> Do correct me if I'm wrong about this.
>>
>> Best,
>> Jake
>> -- 
>> The Csound Blog - http://csoundblog.com/
>> Slipmat - http://slipmat.noisepages.com/
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>>
>
>
>
> Send bugs reports to the Sourceforge bug tracker
>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>



Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

Further info:
http://www.swig.org/Doc1.3/Java.html#java_directors_classes

On 28 Aug 2010, at 10:56, Victor Lazzarini wrote:

> I don't know the Java syntax for this, but by looking at the C++  
> source I reckon what you have to do is:
>
> 1. Derive a class from CsoundCallbackWrapper
> 2. Implement one or more of the functions, which will be your  
> callbacks
>    Here you'll be looking to implement the  
> CsoundCallbackWrapper::YieldCallback(), which is similar placed to  
> PerformanceCallback, but
> instead of called by CsoundPerformanceClass, it is called by  
> PerformKsmps() at ksmp boundaries.
> 3. Instantiate the object, pass it a Csound object.
> 4. Call its SetYieldCallback() method
>
> This should be all you need to do. Any " host data"  your callback  
> needs might be placed as class members to your derived class; you  
> will not
> be able to use 'csoundHostData' to hold host data, because this  
> mechanism already uses it to pass the callback wrapper class object  
> to Csound.
> But you can just add any host data to your callback wrapper class.
>
> This is absolutely untested, since I really know no Java and for  
> Python, which I use, I have wrapped things differently in the swig  
> interface.
>
> Victor
>
> On 28 Aug 2010, at 10:30, Victor Lazzarini wrote:
>
>> There is a helper class that should do the trick:  
>> CsoundCallbackWrapper, using directors. I am not sure how to use  
>> it, but it
>> might be  starting point for you.
>>
>> If you get it working, then you can basically do this:
>>
>> 1. Implement a function to get the table data into a list,  
>> protected by mutex.
>> 2. In the performance callback read this list and call the csound  
>> table functions, protected by mutex
>>
>> This is similar to how CsoundPerformanceThread treats events.
>>
>> Victor
>>
>>
>> On 28 Aug 2010, at 00:40, Jacob Joaquin wrote:
>>
>>> On Fri, Aug 27, 2010 at 11:01 AM, Victor Lazzarini
>>>  wrote:
>>>> So you can use the Performance callback to call any
>>>> functions that might not be called at the same time,
>>>> or to implement your locks.
>>>
>>>
>>> I've been looking into this, and I'm not certain it's possible right
>>> now. The docs I've been reading on SWIG indicates that special
>>> handling is required for passing C pointers in functions.  For
>>> example:
>>>
>>> http://www.opensource.apple.com/source/swig/swig-4/swig/Examples/java/funcptr/index.html?txt
>>>
>>> I do not see %constant, %callback or %nocallback directives in
>>> java_interface.i. Which leads me to believe that it may not be
>>> possible with the Csound Java API, or it's doing a great job of  
>>> hiding
>>> from me.
>>>
>>> Do correct me if I'm wrong about this.
>>>
>>> Best,
>>> Jake
>>> -- 
>>> The Csound Blog - http://csoundblog.com/
>>> Slipmat - http://slipmat.noisepages.com/
>>>
>>>
>>> Send bugs reports to the Sourceforge bug tracker
>>>          https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>> Discussions of bugs and features can be posted here
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>> "unsubscribe csound"
>>>
>>
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>          https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>>
>
>
>
> Send bugs reports to the Sourceforge bug tracker
>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>



Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

But in any case for most applications the simplest way is just to
register a callback for the CsPerformanceThread. This callback is
called whenever a performKsmps finishes and is waited for before
calling another performKsmps. But I'm not sure if this is also
available for java...

Cheers,
Andres

On Sat, Aug 28, 2010 at 11:07 AM, Victor Lazzarini
 wrote:
> Further info:
> http://www.swig.org/Doc1.3/Java.html#java_directors_classes
>
> On 28 Aug 2010, at 10:56, Victor Lazzarini wrote:
>
>> I don't know the Java syntax for this, but by looking at the C++ source I
>> reckon what you have to do is:
>>
>> 1. Derive a class from CsoundCallbackWrapper
>> 2. Implement one or more of the functions, which will be your callbacks
>>   Here you'll be looking to implement the
>> CsoundCallbackWrapper::YieldCallback(), which is similar placed to
>> PerformanceCallback, but
>> instead of called by CsoundPerformanceClass, it is called by
>> PerformKsmps() at ksmp boundaries.
>> 3. Instantiate the object, pass it a Csound object.
>> 4. Call its SetYieldCallback() method
>>
>> This should be all you need to do. Any " host data"  your callback needs
>> might be placed as class members to your derived class; you will not
>> be able to use 'csoundHostData' to hold host data, because this mechanism
>> already uses it to pass the callback wrapper class object to Csound.
>> But you can just add any host data to your callback wrapper class.
>>
>> This is absolutely untested, since I really know no Java and for Python,
>> which I use, I have wrapped things differently in the swig interface.
>>
>> Victor
>>
>> On 28 Aug 2010, at 10:30, Victor Lazzarini wrote:
>>
>>> There is a helper class that should do the trick: CsoundCallbackWrapper,
>>> using directors. I am not sure how to use it, but it
>>> might be  starting point for you.
>>>
>>> If you get it working, then you can basically do this:
>>>
>>> 1. Implement a function to get the table data into a list, protected by
>>> mutex.
>>> 2. In the performance callback read this list and call the csound table
>>> functions, protected by mutex
>>>
>>> This is similar to how CsoundPerformanceThread treats events.
>>>
>>> Victor
>>>
>>>
>>> On 28 Aug 2010, at 00:40, Jacob Joaquin wrote:
>>>
>>>> On Fri, Aug 27, 2010 at 11:01 AM, Victor Lazzarini
>>>>  wrote:
>>>>>
>>>>> So you can use the Performance callback to call any
>>>>> functions that might not be called at the same time,
>>>>> or to implement your locks.
>>>>
>>>>
>>>> I've been looking into this, and I'm not certain it's possible right
>>>> now. The docs I've been reading on SWIG indicates that special
>>>> handling is required for passing C pointers in functions.  For
>>>> example:
>>>>
>>>>
>>>> http://www.opensource.apple.com/source/swig/swig-4/swig/Examples/java/funcptr/index.html?txt
>>>>
>>>> I do not see %constant, %callback or %nocallback directives in
>>>> java_interface.i. Which leads me to believe that it may not be
>>>> possible with the Csound Java API, or it's doing a great job of hiding
>>>> from me.
>>>>
>>>> Do correct me if I'm wrong about this.
>>>>
>>>> Best,
>>>> Jake
>>>> --
>>>> The Csound Blog - http://csoundblog.com/
>>>> Slipmat - http://slipmat.noisepages.com/
>>>>
>>>>
>>>> Send bugs reports to the Sourceforge bug tracker
>>>>         https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>> Discussions of bugs and features can be posted here
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>>> csound"
>>>>
>>>
>>>
>>>
>>> Send bugs reports to the Sourceforge bug tracker
>>>         https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>> Discussions of bugs and features can be posted here
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>>> csound"
>>>
>>
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>          https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
>> csound"
>>
>
>
>
> Send bugs reports to the Sourceforge bug tracker
>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe
> csound"
>
>



-- 


Andrés


Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

The problem is registering the callback. The mechanism allows only  
Csound class callbacks.

Victor
On 28 Aug 2010, at 11:15, Andres Cabrera wrote:

> But in any case for most applications the simplest way is just to
> register a callback for the CsPerformanceThread. This callback is
> called whenever a performKsmps finishes and is waited for before
> calling another performKsmps. But I'm not sure if this is also
> available for java...
>
> Cheers,
> Andres
>
> On Sat, Aug 28, 2010 at 11:07 AM, Victor Lazzarini
>  wrote:
>> Further info:
>> http://www.swig.org/Doc1.3/Java.html#java_directors_classes
>>
>> On 28 Aug 2010, at 10:56, Victor Lazzarini wrote:
>>
>>> I don't know the Java syntax for this, but by looking at the C++  
>>> source I
>>> reckon what you have to do is:
>>>
>>> 1. Derive a class from CsoundCallbackWrapper
>>> 2. Implement one or more of the functions, which will be your  
>>> callbacks
>>>   Here you'll be looking to implement the
>>> CsoundCallbackWrapper::YieldCallback(), which is similar placed to
>>> PerformanceCallback, but
>>> instead of called by CsoundPerformanceClass, it is called by
>>> PerformKsmps() at ksmp boundaries.
>>> 3. Instantiate the object, pass it a Csound object.
>>> 4. Call its SetYieldCallback() method
>>>
>>> This should be all you need to do. Any " host data"  your callback  
>>> needs
>>> might be placed as class members to your derived class; you will not
>>> be able to use 'csoundHostData' to hold host data, because this  
>>> mechanism
>>> already uses it to pass the callback wrapper class object to Csound.
>>> But you can just add any host data to your callback wrapper class.
>>>
>>> This is absolutely untested, since I really know no Java and for  
>>> Python,
>>> which I use, I have wrapped things differently in the swig  
>>> interface.
>>>
>>> Victor
>>>
>>> On 28 Aug 2010, at 10:30, Victor Lazzarini wrote:
>>>
>>>> There is a helper class that should do the trick:  
>>>> CsoundCallbackWrapper,
>>>> using directors. I am not sure how to use it, but it
>>>> might be  starting point for you.
>>>>
>>>> If you get it working, then you can basically do this:
>>>>
>>>> 1. Implement a function to get the table data into a list,  
>>>> protected by
>>>> mutex.
>>>> 2. In the performance callback read this list and call the csound  
>>>> table
>>>> functions, protected by mutex
>>>>
>>>> This is similar to how CsoundPerformanceThread treats events.
>>>>
>>>> Victor
>>>>
>>>>
>>>> On 28 Aug 2010, at 00:40, Jacob Joaquin wrote:
>>>>
>>>>> On Fri, Aug 27, 2010 at 11:01 AM, Victor Lazzarini
>>>>>  wrote:
>>>>>>
>>>>>> So you can use the Performance callback to call any
>>>>>> functions that might not be called at the same time,
>>>>>> or to implement your locks.
>>>>>
>>>>>
>>>>> I've been looking into this, and I'm not certain it's possible  
>>>>> right
>>>>> now. The docs I've been reading on SWIG indicates that special
>>>>> handling is required for passing C pointers in functions.  For
>>>>> example:
>>>>>
>>>>>
>>>>> http://www.opensource.apple.com/source/swig/swig-4/swig/Examples/java/funcptr/index.html?txt
>>>>>
>>>>> I do not see %constant, %callback or %nocallback directives in
>>>>> java_interface.i. Which leads me to believe that it may not be
>>>>> possible with the Csound Java API, or it's doing a great job of  
>>>>> hiding
>>>>> from me.
>>>>>
>>>>> Do correct me if I'm wrong about this.
>>>>>
>>>>> Best,
>>>>> Jake
>>>>> --
>>>>> The Csound Blog - http://csoundblog.com/
>>>>> Slipmat - http://slipmat.noisepages.com/
>>>>>
>>>>>
>>>>> Send bugs reports to the Sourceforge bug tracker
>>>>>         https://sourceforge.net/tracker/? 
>>>>> group_id=81968&atid=564599
>>>>> Discussions of bugs and features can be posted here
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>> "unsubscribe
>>>>> csound"
>>>>>
>>>>
>>>>
>>>>
>>>> Send bugs reports to the Sourceforge bug tracker
>>>>         https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>> Discussions of bugs and features can be posted here
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>> "unsubscribe
>>>> csound"
>>>>
>>>
>>>
>>>
>>> Send bugs reports to the Sourceforge bug tracker
>>>          https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>> Discussions of bugs and features can be posted here
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>> "unsubscribe
>>> csound"
>>>
>>
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe
>> csound"
>>
>>
>
>
>
> -- 
>
>
> Andrés
>
>
> Send bugs reports to the Sourceforge bug tracker
>            https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>



Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"

In Java, I mean. For Python, I've wrapped it conveniently for Python  
functions.

Vicror
On 28 Aug 2010, at 16:28, Victor Lazzarini wrote:

> The problem is registering the callback. The mechanism allows only  
> Csound class callbacks.
>
> Victor
> On 28 Aug 2010, at 11:15, Andres Cabrera wrote:
>
>> But in any case for most applications the simplest way is just to
>> register a callback for the CsPerformanceThread. This callback is
>> called whenever a performKsmps finishes and is waited for before
>> calling another performKsmps. But I'm not sure if this is also
>> available for java...
>>
>> Cheers,
>> Andres
>>
>> On Sat, Aug 28, 2010 at 11:07 AM, Victor Lazzarini
>>  wrote:
>>> Further info:
>>> http://www.swig.org/Doc1.3/Java.html#java_directors_classes
>>>
>>> On 28 Aug 2010, at 10:56, Victor Lazzarini wrote:
>>>
>>>> I don't know the Java syntax for this, but by looking at the C++  
>>>> source I
>>>> reckon what you have to do is:
>>>>
>>>> 1. Derive a class from CsoundCallbackWrapper
>>>> 2. Implement one or more of the functions, which will be your  
>>>> callbacks
>>>>  Here you'll be looking to implement the
>>>> CsoundCallbackWrapper::YieldCallback(), which is similar placed to
>>>> PerformanceCallback, but
>>>> instead of called by CsoundPerformanceClass, it is called by
>>>> PerformKsmps() at ksmp boundaries.
>>>> 3. Instantiate the object, pass it a Csound object.
>>>> 4. Call its SetYieldCallback() method
>>>>
>>>> This should be all you need to do. Any " host data"  your  
>>>> callback needs
>>>> might be placed as class members to your derived class; you will  
>>>> not
>>>> be able to use 'csoundHostData' to hold host data, because this  
>>>> mechanism
>>>> already uses it to pass the callback wrapper class object to  
>>>> Csound.
>>>> But you can just add any host data to your callback wrapper class.
>>>>
>>>> This is absolutely untested, since I really know no Java and for  
>>>> Python,
>>>> which I use, I have wrapped things differently in the swig  
>>>> interface.
>>>>
>>>> Victor
>>>>
>>>> On 28 Aug 2010, at 10:30, Victor Lazzarini wrote:
>>>>
>>>>> There is a helper class that should do the trick:  
>>>>> CsoundCallbackWrapper,
>>>>> using directors. I am not sure how to use it, but it
>>>>> might be  starting point for you.
>>>>>
>>>>> If you get it working, then you can basically do this:
>>>>>
>>>>> 1. Implement a function to get the table data into a list,  
>>>>> protected by
>>>>> mutex.
>>>>> 2. In the performance callback read this list and call the  
>>>>> csound table
>>>>> functions, protected by mutex
>>>>>
>>>>> This is similar to how CsoundPerformanceThread treats events.
>>>>>
>>>>> Victor
>>>>>
>>>>>
>>>>> On 28 Aug 2010, at 00:40, Jacob Joaquin wrote:
>>>>>
>>>>>> On Fri, Aug 27, 2010 at 11:01 AM, Victor Lazzarini
>>>>>>  wrote:
>>>>>>>
>>>>>>> So you can use the Performance callback to call any
>>>>>>> functions that might not be called at the same time,
>>>>>>> or to implement your locks.
>>>>>>
>>>>>>
>>>>>> I've been looking into this, and I'm not certain it's possible  
>>>>>> right
>>>>>> now. The docs I've been reading on SWIG indicates that special
>>>>>> handling is required for passing C pointers in functions.  For
>>>>>> example:
>>>>>>
>>>>>>
>>>>>> http://www.opensource.apple.com/source/swig/swig-4/swig/Examples/java/funcptr/index.html?txt
>>>>>>
>>>>>> I do not see %constant, %callback or %nocallback directives in
>>>>>> java_interface.i. Which leads me to believe that it may not be
>>>>>> possible with the Csound Java API, or it's doing a great job of  
>>>>>> hiding
>>>>>> from me.
>>>>>>
>>>>>> Do correct me if I'm wrong about this.
>>>>>>
>>>>>> Best,
>>>>>> Jake
>>>>>> --
>>>>>> The Csound Blog - http://csoundblog.com/
>>>>>> Slipmat - http://slipmat.noisepages.com/
>>>>>>
>>>>>>
>>>>>> Send bugs reports to the Sourceforge bug tracker
>>>>>>        https://sourceforge.net/tracker/? 
>>>>>> group_id=81968&atid=564599
>>>>>> Discussions of bugs and features can be posted here
>>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>>> "unsubscribe
>>>>>> csound"
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Send bugs reports to the Sourceforge bug tracker
>>>>>        https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>>> Discussions of bugs and features can be posted here
>>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>>> "unsubscribe
>>>>> csound"
>>>>>
>>>>
>>>>
>>>>
>>>> Send bugs reports to the Sourceforge bug tracker
>>>>         https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>>> Discussions of bugs and features can be posted here
>>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>>> "unsubscribe
>>>> csound"
>>>>
>>>
>>>
>>>
>>> Send bugs reports to the Sourceforge bug tracker
>>>          https://sourceforge.net/tracker/?group_id=81968&atid=564599
>>> Discussions of bugs and features can be posted here
>>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>>> "unsubscribe
>>> csound"
>>>
>>>
>>
>>
>>
>> -- 
>>
>>
>> Andrés
>>
>>
>> Send bugs reports to the Sourceforge bug tracker
>>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
>> Discussions of bugs and features can be posted here
>> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
>> "unsubscribe csound"
>>
>
>
>
> Send bugs reports to the Sourceforge bug tracker
>           https://sourceforge.net/tracker/?group_id=81968&atid=564599
> Discussions of bugs and features can be posted here
> To unsubscribe, send email sympa@lists.bath.ac.uk with body  
> "unsubscribe csound"
>



Send bugs reports to the Sourceforge bug tracker
            https://sourceforge.net/tracker/?group_id=81968&atid=564599
Discussions of bugs and features can be posted here
To unsubscribe, send email sympa@lists.bath.ac.uk with body "unsubscribe csound"