Asp Forum - Simple lock - comp.lang.python

moerchendiser2k3

3/20/2010 1:58:00 AM

Hi,

I have a common question about locks:

class SetPointer
{
private:
void *ptr;

MY_LOCK lock;

public:
void SetPointer(void *p)
{
Lock(this->lock);
this->ptr = p;
}

void *GetPointer()
{
Lock(this->lock);
return this->ptr;
}
};

Just a question, is this lock redundant, when the Pointer can be set/
get from different threads?
Thanks a lot!! Bye, moerchendiser2k3

6 Answers

Ben Finney

3/20/2010 2:25:00 AM

moerchendiser2k3 <googler.1.webmaster@spamgourmet.com> writes:

> I have a common question about locks:

You'd be best to ask in a forum related to the language you're using.
This (â??comp.lang.pythonâ??) is a forum for users of the Python language.

--
\ â??If you can do no good, at least do no harm.â? â??_Slapstick_, |
`\ Kurt Vonnegut |
_o__) |
Ben Finney

Chris Rebert

3/20/2010 2:25:00 AM

On Fri, Mar 19, 2010 at 6:58 PM, moerchendiser2k3
<googler.1.webmaster@spamgourmet.com> wrote:
<question about using locks in C++ snipped>
>
> Thanks a lot!! Bye, moerchendiser2k3

This is the **Python** mailinglist/newsgroup; and your question isn't
about Python.

The C++ one is over there:
http://groups.google.com/group/comp.lang.c++.modera...

Regards,
Chris

MRAB

3/20/2010 2:31:00 AM

moerchendiser2k3 wrote:
> Hi,
>
> I have a common question about locks:
>
> class SetPointer
> {
> private:
> void *ptr;
>
> MY_LOCK lock;
>
>
> public:
> void SetPointer(void *p)
> {
> Lock(this->lock);
> this->ptr = p;
> }
>
> void *GetPointer()
> {
> Lock(this->lock);
> return this->ptr;
> }
> };
>
>
> Just a question, is this lock redundant, when the Pointer can be set/
> get from different threads?
> Thanks a lot!! Bye, moerchendiser2k3

1. That's C++. What does it have to do with Python?

2. The value you're accessing is a simple pointer which you're either
setting or getting, so a lock shouldn't be necessary.

3. You're locking, but never unlocking. The sequence should be: lock, do
stuff, unlock.

Gabriel Genellina

3/20/2010 3:10:00 AM

En Fri, 19 Mar 2010 23:31:23 -0300, MRAB <python@mrabarnett.plus.com>
escribió:
> moerchendiser2k3 wrote:

>> class SetPointer
>> {
>> private:
>> void *ptr;
>> MY_LOCK lock;
>> public:
>> void SetPointer(void *p)
>> {
>> Lock(this->lock);
>> this->ptr = p;
>> }

> 3. You're locking, but never unlocking. The sequence should be: lock, do
> stuff, unlock.

Just FYI: C++ doesn't have try/finally, and such behavior is usually
emulated using a local object. When it goes out of scope, it is
automatically destroyed, meaning that the object destructor is called.
Whatever you would write in a "finally" clause, in C++ goes into a
destructor.

Of course C++ guys would never say they're "emulating" try/finally,
instead they declare RAII as *the* Only and Right Way :)

--
Gabriel Genellina

MRAB

3/20/2010 3:38:00 AM

Gabriel Genellina wrote:
> En Fri, 19 Mar 2010 23:31:23 -0300, MRAB <python@mrabarnett.plus.com>
> escribió:
>> moerchendiser2k3 wrote:
>
>>> class SetPointer
>>> {
>>> private:
>>> void *ptr;
>>> MY_LOCK lock;
>>> public:
>>> void SetPointer(void *p)
>>> {
>>> Lock(this->lock);
>>> this->ptr = p;
>>> }
>
>> 3. You're locking, but never unlocking. The sequence should be: lock, do
>> stuff, unlock.
>
> Just FYI: C++ doesn't have try/finally, and such behavior is usually
> emulated using a local object. When it goes out of scope, it is
> automatically destroyed, meaning that the object destructor is called.
> Whatever you would write in a "finally" clause, in C++ goes into a
> destructor.
>
> Of course C++ guys would never say they're "emulating" try/finally,
> instead they declare RAII as *the* Only and Right Way :)
>
Lock() doesn't look like a declaration, but a function call (although
it's been a while since I last did C++!).

In the past I've written some C++ code where try..finally... would've
been useful... *sigh*

Alf P. Steinbach

3/20/2010 4:15:00 AM

* MRAB:
> Gabriel Genellina wrote:
>> En Fri, 19 Mar 2010 23:31:23 -0300, MRAB <python@mrabarnett.plus.com>
>> escribió:
>>> moerchendiser2k3 wrote:
>>
>>>> class SetPointer
>>>> {
>>>> private:
>>>> void *ptr;
>>>> MY_LOCK lock;
>>>> public:
>>>> void SetPointer(void *p)
>>>> {
>>>> Lock(this->lock);
>>>> this->ptr = p;
>>>> }
>>
>>> 3. You're locking, but never unlocking. The sequence should be: lock, do
>>> stuff, unlock.
>>
>> Just FYI: C++ doesn't have try/finally, and such behavior is usually
>> emulated using a local object. When it goes out of scope, it is
>> automatically destroyed, meaning that the object destructor is called.
>> Whatever you would write in a "finally" clause, in C++ goes into a
>> destructor.
>>
>> Of course C++ guys would never say they're "emulating" try/finally,
>> instead they declare RAII as *the* Only and Right Way :)
>>
> Lock() doesn't look like a declaration, but a function call (although
> it's been a while since I last did C++!).

Right. But the OP is clearly a beginner who has yet to learn conventions (like
all uppercase reserved for macros) and good versus bad programming constructs
(like using void*), and who is even confusing C++ with Python. So it might just
be that 'Lock' is a macro, and then it can expand to anything.

> In the past I've written some C++ code where try..finally... would've
> been useful... *sigh*

Check out Marginean and Alexandrescu's ScopeGuard. Old DDJ-article. For MSVC you
may want to change use of __LINE__ in that code, to MS-specific __COUNTER__.

Fwiw. I can't remember ever needing 'finally' in C++ code.

It indicates a lack of encapsulation of entitites that should clean up
themselves. Simple smart pointers etc. can help. E.g., check out Boost library.

Cheers & hth., even if a little off-topic,

- Alf

comp.lang.python

Simple lock

moerchendiser2k3

Ben Finney

Chris Rebert

MRAB

Gabriel Genellina

MRAB

Alf P. Steinbach

x Login to ForumsZone