[squeak-dev] Re: [Pharo-project] #ensure: issues
leves at elte.hu
Thu Mar 4 21:41:44 UTC 2010
On Thu, 4 Mar 2010, Bert Freudenberg wrote:
> On 04.03.2010, at 21:57, Levente Uzonyi wrote:
>> On Thu, 4 Mar 2010, Igor Stasenko wrote:
>>> 2010/3/4 Levente Uzonyi <leves at elte.hu>:
>>>> Here's a simple example (replace file with any external resource):
>>>> My process opened a file, a termination block will close it if it's
>>>> evaluated. If I send #terminate to the process I expect my file to be
>>>> closed, so I won't leak a file descriptor.
>>>> If the file can't be closed (aka the termination block raised an error) then
>>>> there's a serious problem. It doesn't really matter what happens then.
>>>> But I don't have to try to convince you anymore, because Andreas is about to
>>>> solve the issue.
>>> I'm not trying to convince anyone, i just wanted to show you that
>>> there is no good solution in that plane.
>>> More workarounds means more code to run (and makes things more complex, btw).
>>> But you will be still unsafe.You will be safe, once you stop relying
>>> on #ensure: in your code and use different approach.
>>> As for your example: use weak finalizer to close your file.
>> I'm about to get rid of the weak finalizer in cases where we know the exact scope when a file is used (as with #*FileNamed:do:). There's no point in registering and unregistering a file to the WeakRegistry if #ensure: closes the file for us at the right time.
>>> This will make sure that no matter what were happen, you wont leave
>>> the file open. Working with external resources is a pain. But lets try
>>> to not poison ourselves with manual resource management, which comes
>>> from C world.
>> That's exactly what I'm about to do, keeping the resource management automatic (from the user POV) while making it fast.
>> You can read my monologue about these plans here: http://lists.squeakfoundation.org/pipermail/squeak-dev/2010-February/144993.html
> Well, the finalizers are a safety net for file handles only anyway. Developers are advised to close files asap, and any code I have seen at least tries to do that. Why would you want to remove those finalizers?
There are several places where the files are only closed when no error
occurs, for example: AIFFFileReader >>
If an error occurs, the finalizer will close it. There are more than 200
methods which possibly have the same kind of code.
I want to remove the registration to the WeakRegistry when a file is
opened via #*FileNamed:do:, because registering is slow and superfluous
because the termination block (#ensure:) will close them anyway.
Currently registering and unregistering takes ~27% of the total time of
opening and closing a file if the WeakRegistry is almost empty (2
elements). If it contains 1002 elements, then registering and
unregistering takes ~37% of the total time. This could be decreased to 0
in both cases, and the number of open files wouldn't affect the
performance of file opening and closing in this case.
> - Bert -
More information about the Squeak-dev