[Numpy-discussion] Memory leak when looking .flags

[Robert Kern]

Cyrille Rosset wrote:
> Hi,
> 
> I'm not sure this is the right mailing list for this, but it seems 
> there's a memory leak when looking at flags :
> 
>  >>> from numpy import *
>  >>> x=ones(50000000)       #==> python use 25% of memory (ok)
>  >>> del x
> #==> memory usage fall back to almost zero (as seen in top)
> Thqt's good.
> 
> but if I look at flags before the del :
>  >>> x=ones(50000000)
>  >>> x.flags
>    C_CONTIGUOUS : True
>    F_CONTIGUOUS : True
>    OWNDATA : True
>    WRITEABLE : True
>    ALIGNED : True
>    UPDATEIFCOPY : False
>  >>> del x
>  >>> who()
> 
> Upper bound on total bytes  =       0
> 
> That looks nice, but the memory usage by python (in top) is still 25%... 
>   Isn't it a bug ?

No, x.flags is still being stored in _. It still has a reference to x. Evaluate
something else (e.g. ">>> 1") to clear that out and the memory should be released.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco