[Stk] New STK release (4.4.0)

[John Dey]

Gary,

I have looked at the code and also run Md2Skini with the -f option.   
The deltatime is zero in the file generated by Md2Skini.  It appears  
to be an error in the RtMidi.cpp and its interface with CoreMidi.  Has  
anyone else seen this error?  If not, can anyone verify.  Thanks.  My  
host used is iMac 10.5.6.

John


On May 3, 2009, at 5:56 PM, Gary Scavone wrote:

> Hi John,
>
> If you look at the Md2Skini.cpp code, you'll see that SKINI messages  
> sent to stdout always have a delta time of zero.  This is because in  
> a realtime context, a delta time of zero say "apply the message  
> immediately" (which is usually what one wants).  However, if the  
> output is sent to a file, the delta times are not zero ... they are  
> the actual times between events.
>
> I guess it would help to say that in the usage statement!
>
> Regards,
>
> --gary
>
> On 1-May-09, at 7:13 PM, John Dey wrote:
>
>> Gary,
>>
>> Thanks for making your work available.  You probably have trained  
>> more audio programmers then you ever imagined.  I find it  
>> fascinating seeing how things work.
>>
>> I do have a question:  The output of Md2Skini has the time output  
>> as zero.  Why don't I see the delta time?  I traced it back to  
>> RtMidi and it appears that CoreMidi is returning zero.  Is there an  
>> error in that code or am I missing something?
>>
>> John
>>
>>
>> On Apr 30, 2009, at 3:48 PM, Gary Scavone wrote:
>>
>>> Dear STKers,
>>>
>>> I just finished a new release of STK.  This update has been "in the
>>> works" for more than a year and a half.  The most pressing issue was
>>> to get rid of the virtual tick() function structure, which made  
>>> things
>>> very much slower than they should have been.  That entailed some
>>> significant restructuring of the code and it took a long time to
>>> determine the best way to proceed.  The new distribution should be  
>>> at
>>> least 30-50% faster than before, achieved in large part by inlining
>>> many of the tick() functions.
>>>
>>> Significant changes include:
>>>
>>> - All classes embedded in the "stk" namespace (except RtAudio,  
>>> RtMidi,
>>> and RtError).
>>> - Class WaveLoop renamed FileLoop.
>>> - Significant efficiency improvements via code restructuring and
>>> inlining.
>>> - Some class source (.cpp) files deleted as part of inlining
>>> (Generator, Filter, Function, WvIn, WvOut, Effect, Instrmnt,  
>>> BowTable,
>>> ReedTable, JetTable, Vector3D).
>>> - Updates to RtAudio and RtMidi.
>>> - Previous "tickFrame()" functions renamed "tick" for more  
>>> consistent
>>> API.
>>> - More consistent and scalable approach to multichannel data and
>>> computations.
>>> - Multichannel support added to Granulate class.
>>> - Filter class made abstract. New Iir and Fir classes made for non-
>>> order-specific filtering.
>>> - New TapDelay class.
>>> - SubNoise class deleted (same as sub-sampled "ticking" of Noise  
>>> class).
>>>
>>> The easy way to deal with the namespace is to put a "using namespace
>>> stk;" at the top of your program files.
>>>
>>> As usual, everything is detailed at the website:
>>>
>>>  http://ccrma.stanford.edu/software/stk/
>>>
>>> Happy programming!
>>>
>>> Regards,
>>>
>>> --gary
>>>
>>> _______________________________________________
>>> Stk mailing list
>>> Stk at ccrma.stanford.edu
>>> http://ccrma-mail.stanford.edu/mailman/listinfo/stk
>