Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #10440 > unrolled thread

monotonically increasing memory usage

Back to article view | Back to comp.lang.python

Contents

  monotonically increasing memory usage Pedro Larroy <pedro.larroy@gmail.com> - 2011-07-28 11:52 +0200
    Re: monotonically increasing memory usage Nobody <nobody@nowhere.com> - 2011-07-28 18:01 +0100
    Re: monotonically increasing memory usage Ulrich Eckhardt <ulrich.eckhardt@dominolaser.com> - 2011-07-29 13:35 +0200

#10440 — monotonically increasing memory usage

Hi

pickling

Just crossposting this from stackoverflow:

http://stackoverflow.com/questions/6857006/python-monotonically-increasing-memory-usage-leak

Any hints?


Pedro.

-- 
Pedro Larroy Tovar   |    http://pedro.larroy.com/

[toc] | [next] | [standalone]

#10465

On Thu, 28 Jul 2011 11:52:25 +0200, Pedro Larroy wrote:

> pickling
> 
> Just crossposting this from stackoverflow:
> 
> http://stackoverflow.com/questions/6857006/
> 
> Any hints?

AFAIK, it's because the Pickler object keeps a reference to each object so
that pointer-sharing works; if you write the same object multiple times,
you get multiple references to a single object, not multiple objects.

This means that the dictionaries aren't deleted while the Pickler object
lives.

It would seem that this issue could be avoided if pickle used weak
references, but there may be issues which I have overlooked.

[toc] | [prev] | [next] | [standalone]

#10511

Pedro Larroy wrote:
> Just crossposting this from stackoverflow:
> 
> http://stackoverflow.com/...
> 
> Any hints?

At first I was just too lazy to visit stackoverflow and skipped this 
posting. Then I thought: Why didn't you include the content, so people can 
actually answer this question here? Therefore the only answer you got here 
is utterly useless because the whole context is missing!

Then, something worse dawned upon me: Since stackoverflow is a "free" site, 
their business model probably involves data mining or advertising, and since 
you just posted a link to that site with the sole goal of getting people to 
visit it, you have effectively (But I'm definitely not saying intentionally 
and knowingly!) become a spammer.

I'm really not blaming you, but this is puzzling me and I'm not sure how to 
react at the moment...

Cheers!

Uli

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web