git.net

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

open, close



> On 1 Sep 2019, at 17:57, MRAB <python at mrabarnett.plus.com> wrote:
> 
> On 2019-09-01 16:46, Barry wrote:
>>> On 31 Aug 2019, at 15:41, Manfred Lotz <ml_news at posteo.de> wrote:
>>> When you say COULD this sounds like it is a matter of luck. My thinking
>>> was that USUALLY the file will be closed after the statement because
>>> then the file handle goes out of scope.
>> It all depends on the way any python implementation does its garbage collection. The file is closed as a side effect of deleting the file object to reclaiming the memory of the file object.
>> At the start of python 3 people where suprised when files and other resources where not released at the same time that python 2 released them.
> Is that true?

Yes. I recalling having to fix some code, but that was over 10 years ago.
I have been using python 3 since the alphas of 3.0.

> 
> I thought that it was because other implementations of Python, such as Jython and IronPython, don't use reference counting, so files and other resources aren't necessarily released as soon as an object loses its last reference, and, moreover, it's not required to do so by the language definition.

You can also see delayed rsource freeing with cpython. I think it was
an fd leak and lsof showed the files where not closing. Adding with fixed it.
But it was so long ago...

> 
> Adding the 'with' statement added determinism.

Barry

> 
> See PEP 343 -- The "with" Statement
> -- 
> https://mail.python.org/mailman/listinfo/python-list
>