I've posted a macintosh VM (3.8.6b5) that includes the buffered display improvements, changes from Andreas to view certain GC events/features and my changes from last fall to collect GC statistics & alter the VM to bias for growth versus compaction. Plus changes to sound input/output.
I've stuck them on my idisk http://homepage.mac.com/johnmci/FileSharing.html
See the experimental directory and the squeakGarbage sub-directory. or see the following ftp directory for zip files.
ftp://ftp.smalltalkconsulting.com/squeakGarbage/ ftp://ftp.smalltalkconsulting.com/
I have not build an os-9 version yet.
3.8.6b5 uses the Unix CoreAudio sound plugin which apparently allows you greater flexibility in recording rates. Folks should confirm recording and sound playback works as in the past since we have dropped the ancient sound input and carbon based double buffering for sound playback.
- ======================================================================== === John M. McIntosh johnmci@smalltalkconsulting.com 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===
Am 05.02.2005 um 07:54 schrieb John M McIntosh:
3.8.6b5 uses the Unix CoreAudio sound plugin which apparently allows you greater flexibility in recording rates. Folks should confirm recording and sound playback works as in the past since we have dropped the ancient sound input and carbon based double buffering for sound playback.
Recording works fine. However, on playback I sometimes get a a very sharp, short noise - most often when playing the first sound, once when quitting, sometimes in between.
It's a bit hard to reproduce, but I get it fairly consistently when dragging an object from the object catalog and immediately dropping it back onto the catalog (don't know which sound that is), the first time after starting the image.
This is a G4 Titanium running 10.3.7 ...
- Bert -
Found it.
In the Unix code sqUnixSoundMacOSX.c
OSStatus bufferDataProc(AudioConverterRef inAudioConverter, UInt32 *ioDataSize, void **outData, void *context) { ... if (!n1) { char empty[256]; *ioDataSize= min(256, *ioDataSize); *outData= (void *)empty; # if (DEBUG) putchar('-'); fflush(stdout); # endif } ...
One should NOT assume that char empty[256] will in fact result in a byte string containing 256 zero bytes, rather we get something not-zero, thus a buzz. I also wonder about the *outData = (void *) empty if *ioDataSize is actually < 256 bytes. Seems to me one should do a memset(outData,0,*ioDataSize); to ensure zeros and not walking off the end of a small outData buffer...
A 3.8.6b6 is pending, awaiting further feedback.
On Feb 5, 2005, at 7:46 AM, Bert Freudenberg wrote:
Am 05.02.2005 um 07:54 schrieb John M McIntosh:
3.8.6b5 uses the Unix CoreAudio sound plugin which apparently allows you greater flexibility in recording rates. Folks should confirm recording and sound playback works as in the past since we have dropped the ancient sound input and carbon based double buffering for sound playback.
Recording works fine. However, on playback I sometimes get a a very sharp, short noise - most often when playing the first sound, once when quitting, sometimes in between.
It's a bit hard to reproduce, but I get it fairly consistently when dragging an object from the object catalog and immediately dropping it back onto the catalog (don't know which sound that is), the first time after starting the image.
This is a G4 Titanium running 10.3.7 ...
- Bert -
-- ======================================================================== === John M. McIntosh johnmci@smalltalkconsulting.com 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===
On Feb 8, 2005, at 12:07 AM, John M McIntosh wrote:
Found it.
In the Unix code sqUnixSoundMacOSX.c
OSStatus bufferDataProc(AudioConverterRef inAudioConverter, UInt32 *ioDataSize, void **outData, void *context) { ... if (!n1) { char empty[256]; *ioDataSize= min(256, *ioDataSize); *outData= (void *)empty; # if (DEBUG) putchar('-'); fflush(stdout); # endif } ...
One should NOT assume that char empty[256] will in fact result in a byte string containing 256 zero bytes, rather we get something not-zero, thus a buzz. I also wonder about the *outData = (void *) empty if *ioDataSize is actually < 256 bytes. Seems to me one should do a memset(outData,0,*ioDataSize); to ensure zeros and not walking off the end of a small outData buffer...
Oops not quite correct one is setting the void **outData so we just need to point to a zero byte buffer containing enough zeros. Still this is a local variable returned to the caller and the processed, wonder who or where the stack memory lurks. Mmm perhaps it should become a static zero byte structure somewhere.
-- ======================================================================== === John M. McIntosh johnmci@smalltalkconsulting.com 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===
Hi,
Thanks, this is at
ftp://st.cs.uiuc.edu/Smalltalk/Squeak/3.8/mac/squeak3.8.6b5-mac-vm.app.sit
cheers
bruce
On Fri, Feb 04, 2005 at 10:54:56PM -0800, John M McIntosh wrote:
I've posted a macintosh VM (3.8.6b5) that includes the buffered display improvements, changes from Andreas to view certain GC events/features and my changes from last fall to collect GC statistics & alter the VM to bias for growth versus compaction. Plus changes to sound input/output.
I've stuck them on my idisk http://homepage.mac.com/johnmci/FileSharing.html
See the experimental directory and the squeakGarbage sub-directory. or see the following ftp directory for zip files.
ftp://ftp.smalltalkconsulting.com/squeakGarbage/ ftp://ftp.smalltalkconsulting.com/
I have not build an os-9 version yet.
3.8.6b5 uses the Unix CoreAudio sound plugin which apparently allows you greater flexibility in recording rates. Folks should confirm recording and sound playback works as in the past since we have dropped the ancient sound input and carbon based double buffering for sound playback.
========================================================================
John M. McIntosh johnmci@smalltalkconsulting.com 1-800-477-2659 Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===
squeak-dev@lists.squeakfoundation.org