Re: [foo-devel] memory leak, WAS: Some buggies, examples to show

[Ramon Gonzalez-A. <ar...@ma...>]

Dear Martin,

Thanks for your quick answer.

With respect to snd-region/snd-extract.

> Did some tests and they look fine -- could you please test for you
> also?
>


  I don't understand what do you mean. I have already done new tests  
with the just recently downloaded foo and as my report said, I have  
had problems when snd-region is applied after snd-extract, but not  
the other way round, which was the case with the previous foo.

Are you not having problems?

...

With respect to memory-leak. Let me see if my reasoning is right…

What I am doing is iteratively repeating the evaluation of the same  
patch (actually three different patches Test3a&b (sampling with  
forced gc/no forced gc), Test3c (synthesis with forced gc), Test3d&e  
(very simple synthesis (no sampling) with forced gc/no forced gc)).

My foo has a default-heapsize=65536.

In the cases where gc is forced once per cycle, since the patches are  
quite simple, never more memory is required.

In the case where I do not force gc, it happens only when it is  
needed, which is probably when the heap is filled (* 2 65536)=131072  
and then it remains constant.

As you can see, in the old foo memory was constantly increasing,  
whether or not forcing gc.

This could mean that actually there is no memory leak.

But maybe before being so forward I should do more complex tests. And  
also tests where I do not kill-context every cycle.

Does it make sense my reasoning?

All the best,

Ramon