Where's the attachment?

Regards,
Mike

----- Original Message ----- 
From: "Iain Duncan" 
To: 
Sent: Monday, August 22, 2005 6:56 PM
Subject: [Cs-dev] API wackiness for real time


> Thanks Istvan and Michael for the help, I think I have got things written 
> correctly now. If you could look at the attachment that would be great. 
> Here's the catch, it only works when csound is rendering to file. 
> Everything behaves as I would expect under those circumstances. When 
> using -odac, things get odd. I get no print out from the csound thread, 
> and if enabled, the display thread just gets zeros. Same result if I turn 
> off the display thread entirely, no print out from the csound thread.
>
> Thanks
> Iain
>
>
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO
> September 19-22, 2005 * San Francisco, CA * Development Lifecycle 
> Practices
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel
> 




-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Someone didn't get much sleep last night, oops.

Iain

Michael Gogins wrote:
> Where's the attachment?
> 
> Regards,
> Mike
> 
> ----- Original Message ----- From: "Iain Duncan" 
> To: 
> Sent: Monday, August 22, 2005 6:56 PM
> Subject: [Cs-dev] API wackiness for real time
> 
> 
>> Thanks Istvan and Michael for the help, I think I have got things 
>> written correctly now. If you could look at the attachment that would 
>> be great. Here's the catch, it only works when csound is rendering to 
>> file. Everything behaves as I would expect under those circumstances. 
>> When using -odac, things get odd. I get no print out from the csound 
>> thread, and if enabled, the display thread just gets zeros. Same 
>> result if I turn off the display thread entirely, no print out from 
>> the csound thread.
>>
>> Thanks
>> Iain
>>
>>
>> -------------------------------------------------------
>> SF.Net email is Sponsored by the Better Software Conference & EXPO
>> September 19-22, 2005 * San Francisco, CA * Development Lifecycle 
>> Practices
>> Agile & Plan-Driven Development * Managing Projects & Teams * Testing 
>> & QA
>> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
>> _______________________________________________
>> Csound-devel mailing list
>> Csound-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/csound-devel
>>
> 
> 
> 
> 
> -------------------------------------------------------
> SF.Net email is Sponsored by the Better Software Conference & EXPO
> September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
> Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
> Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
> _______________________________________________
> Csound-devel mailing list
> Csound-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/csound-devel
> 

After a few minor changes to make it compile with the latest CVS
version of Csound, I could not reproduce the problem. The attached
file (requires an up to date Csound library) works OK, but again,
I had no problems - other than the compile errors - with the original
either.

Thanks Istvan, I will try a new build of csound and see how that goes. 
Could you tell me if you have time why the changes are made? And why do 
I need to use csound thread and mutex functions for threads that are not 
to do with csound? What specifically prevents use of pthreads and 
pthread mutex's, or how are the csound ones better? Also, why are the 
return types changed to uintptr_t?

Thanks so much for your time on this
Iain

Istvan Varga wrote:
> After a few minor changes to make it compile with the latest CVS
> version of Csound, I could not reproduce the problem. The attached
> file (requires an up to date Csound library) works OK, but again,
> I had no problems - other than the compile errors - with the original
> either.
> 
> 
> ------------------------------------------------------------------------
> 
> // cs_simple_3.c
> 
> // adding a value output structure for sending values to and from csound
> // protected by a mutex
> 
> #include 
> #include 
> #include 
> 
> // csound table number for output values
> const int out_tbl = 2;
> const char *csd_file = "cs_simple_lfo.csd";
> 
> // array and mutex for the csound output values
> struct csound_output
> {
> 	void	*mtx;
> 	float	data[16];
> } cs_out;
> 
> // Csound state variables:
> 
> // pointer to an instance of csound
> CSOUND *csound;
> 
> // boolean to tell us if csound is running
> volatile bool display_thread_stop = false;
> 
> // an empty user_data pointer ( would be the data for the csound thread routine )
> void *user_data = NULL;
> 
> // the routine for the csound thread
> uintptr_t csound_thread_routine( void *user_data );
> 
> // the display thread
> uintptr_t display_thread_routine( void *cs_out_data );
> 
> // the table value we want to get from csound
> MYFLT table_value; 
> float result;
> 
> int main( void )
> {
> 	
> 	// initialize the mutex for the output values
> 	cs_out.mtx = csoundCreateThreadLock();
> 
> 	// make our display thread
> 	void *display_thread;
> 	display_thread = csoundCreateThread( display_thread_routine, NULL );
> 	
> 	// create an instance of csound with no user data
> 	csound = csoundCreate( NULL );
> 
> 	// cb_thread_yield will get called whenever csound has spare time to do other stuff
> 	// csoundSetYieldCallback(csound, &cs_yield_routine); 
> 	
> 	// create a csound thread
> 	// returns a thread handle pointer, takes pointers to the thread routine and to user data
> 	void *cs_thread = csoundCreateThread( csound_thread_routine, user_data );
> 	
> 	printf( "\n\n SPAWNING CSOUND THREAD.\n\n" );
> 
> 	//printf("LFO value is: %5.2f.\n", float( table_value ) );
> 	// wait on the csound thread until it is done
> 	int cs_result = (int) ((unsigned int) csoundJoinThread( cs_thread ));
> 	display_thread_stop = true;
> 	// wait for our display thread to exit too
> 	csoundJoinThread( display_thread );
> 	printf("Csound thread returned %d\n", cs_result);
> 	return cs_result;
> }
> 
> 
> uintptr_t csound_thread_routine(void *user_data)
> {
> 	printf("Csound thread executing.\n");
> 	
> 	// the below does not allow console output for some reason
> 	const char *argv[] = {"csound", "-+rtaudio=alsa", "-odac", csd_file };
> 	//csoundCompile(csound, 4, (char **)argv);
> 	
> 	// the below makes everything work
> //	const char *argv[] = {"csound", "cs_simple_lfo.csd" };
> 	int retval = csoundCompile(csound, sizeof(argv) / sizeof(char*),
> 					   (char **)argv);
> 	
> 	// go is set by the user
> //	int index = 0;
> 	double prvTime, curTime;
> 	RTCLOCK rt;
> 	timers_struct_init(&rt);
> 	prvTime = timers_get_real_time(&rt);
> 	while( ! retval ) {
> 		// perform csound for a ksmp
> 		retval = csoundPerformKsmps(csound);
> 		if (retval) break;
> 
> 		// get all our output values, protected with mutex
> 		int result = csoundWaitThreadLock( cs_out.mtx, 1000 );
> 		if ( result ) {
> 			printf("Timeout in Csound thread\n");
> 			return (uintptr_t) ((intptr_t) -1);
> 		}
> 		
> 		for ( int i = 0; i < 16; i++ )
> 		{
> 		 	table_value = csoundTableGet( csound, 2, i );
> 		 	cs_out.data[i] = float(table_value);
> 		}
> 		
> 		csoundNotifyThreadLock( cs_out.mtx );
> 
> 		// the below works fine when not using -odac
> 		// with -odac, no printing ever happens, even with spawining other threads
> 		curTime = timers_get_real_time(&rt);
> 		if (curTime - prvTime >= 0.1) {	// print 10 values per second
> 			printf( "Csound thread. LFO value is: %5.3f\n", cs_out.data[0] );
> 			prvTime = curTime;
> 		}
> 
> 	}    
> 	// we are done, so clean up
> 	csoundCleanup(csound);
> 	printf( "Csound thread all done!\n" );
> 	retval = (retval == CSOUND_EXITJMP_SUCCESS ? 0 : retval);
> 	return (uintptr_t) ((intptr_t) retval);
> 
> }
> 
> // this is running but it sure doesn't seem to get the table data
> uintptr_t display_thread_routine( void *cs_out_data )
> {
> 	MYFLT lfo_value;
> 	void *mtx = csoundCreateThreadLock();
> 	csoundWaitThreadLock(mtx, 1);
> 
> 	while ( ! display_thread_stop )
> 	{
> 		// get the lfo value from the csound output structure
> 		int result = csoundWaitThreadLock( cs_out.mtx, 1000 );
> 		if ( result ) {
> 			printf("Timeout in display thread\n");
> 			return (uintptr_t) ((intptr_t) -1);
> 		}
> 		lfo_value = cs_out.data[0];
> 		csoundNotifyThreadLock( cs_out.mtx );
> 
> 		// without -odac the below prints a moving lfo value
> 		// with -odac it only ever prints a 0.00
> 		printf("Display thread. LFO value is: %5.3f\n\n", lfo_value);
> 		// hack to implement 100 ms sleep (does not work on Mac)
> 		csoundWaitThreadLock(mtx, 100);
> 	}
> 
> 	return 0;
> }
> 


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net

Iain Duncan wrote:

> Could you tell me if you have time why the changes are made?

None of the changes have anything to do with fixing the problem you
reported (as I simply could not reproduce it). I only fixed the
compile errors resulting from recent API changes, reduced the
rate of printing (which, originally done at the control rate, caused
a lot of dropouts in real time sound output), and made the program
exit more cleanly, particularly in the case of errors and Control-C.

> And why do I need to use csound thread and mutex functions for
 > threads that are not to do with csound? What specifically prevents
 > use of pthreads and pthread mutex's, or how are the csound ones better?

This was only done to make the code more consistent. There is no problem
with using pthreads (but you should not mix the use of the two different
interfaces on the same thread or mutex object, even if you know that the
Csound interface just wraps pthreads anyway) - it is even slightly faster,
using the Csound functions is just cleaner and more portable.

> Also, why are the return types changed to uintptr_t?

The thread function is expected to return a void* pointer when using the
pthread interface (that is, any platform other than Windows), and a void*
may not always match the size of an int, but it should match the size of
the uintptr_t type. Win32 threads return an unsigned int which is also the
same size as uintptr_t, although only on a 32 bit system; on Win64, I think
unsigned int is 32 bits, while uintptr_t is 64 bits, so the result may be
truncated to 32 bits.


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
Csound-devel mailing list
Csound-devel@lists.sourceforge.net