AudioMatt

I'm writing a timeline object which, to the user, stores a dictionary in each audio region. so internally, I just have a void* in each region struct called "content." The idea is that I'd store a reference to a dict and that's where I'd put a buffer name and anything else.

What happens if I split a region and copy the void*?

Now I have two pointers to the same dictionary. If I store a *t_dictionary in the void*, I can't re-retain it without first searching for its associated t_symbol using namefromptr. That seems inefficient.

The other option is to store a *t_symbol with the dictionary name. Now I can't call release without re-calling register which defeats the purpose of the function completely.

Does anyone see what I'm getting at here? Do I have to store the *t_dict and the *t_symbol?

I'm writing a timeline object which, to the user, stores a dictionary in each audio region.  so internally, I just have a void* in each region struct called "content."  The idea is that I'd store a reference to a dict and that's where I'd put a buffer name and anything else.

What happens if I split a region and copy the void*?    

Now I have two pointers to the same dictionary.   If I store a *t_dictionary in the void*, I can't re-retain it without first searching for its associated t_symbol using namefromptr.  That seems inefficient.   

The other option is to store a *t_symbol with the dictionary name.   Now I can't call release without re-calling register which defeats the purpose of the function completely.

Does anyone see what I'm getting at here?   Do I have to store the *t_dict and the *t_symbol?


I'm writing a timeline object which, to the user, stores a dictionary in each audio region.  so internally, I just have a void* in each regi

storing-dict-references

I can't catch the point which you are telling. Can you please tell me in details? Thanks and Regards, 

I can't catch the point which you are telling. Can you please tell me in details? Thanks and Regards, panda helper app emu4ios app gbwhatsapp app  

I'll be honest and say that this approach scares me. The memory/resource ownership of dictionaries can be delicate and the API for dictionaries is in place to provide safety. By copying an anonymous and untyped pointer the entire reference counting system of dictionaries is rendered helpless.

There are multiple options here. Using the existing SDK, I suggest looking at the functions for cloning or referencing dictionaries.

Another option is to look at our (still experimental) C++ API for Max which avoids using pointers altogether to make it harder to goof up the resource ownership of dictionaries. An example of a dictionary object using the C++ API can be seen at 

https://github.com/Cycling74/min-devkit/blob/master/source/projects/min.dict.join/min.dict.join.cpp

I'll be honest and say that this approach scares me. The memory/resource ownership of dictionaries can be delicate and the API for dictionaries is in place to provide safety. By copying an anonymous and untyped pointer the entire reference counting system of dictionaries is rendered helpless. 

Another option is to look at our (still experimental) C++ API for Max which avoids using pointers altogether to make it harder to goof up the resource ownership of dictionaries. An example of a dictionary object using the C++ API can be seen at https://github.com/Cycling74/min-devkit/blob/master/source/projects/min.dict.join/min.dict.join.cpp


Got it, thanks Timothy. All good things to know!

Got it, thanks Timothy.  All good things to know! 

Storing dict references