Do I need a lock here?

[Aaron Brady]

On Oct 28, 3:29 pm, jasiu85 <jasi... at gmail.com> wrote:
> On Oct 27, 10:12 am, "Diez B. Roggisch" <de... at nospam.web.de> wrote:
>
>
>
> > jasiu85 schrieb:
>
> > > Hey,
>
> > > Please take a look at the code of the two threads below:
>
> > > COMMON_DICT = {}
>
> > > def thread_1():
> > >     global COMMON_DICT
> > >     local_dict = prepare_dict()
> > >     COMMON_DICT = local_dict
>
> > > def thread_2():
> > >     global COMMON_DICT
> > >     local_dict = COMMON_DICT
> > >     use_dict(local_dict)
>
> > > Do I need a lock to protect the COMMON_DICT dictionary? AFAIK bytecode
> > > operations are atomic and in each thread there's only one crucial
> > > bytecode op: STORE_NAME in the first thread and LOAD_NAME in the
> > > second one. So I suspect that everything will work just fine. Am I
> > > right?
>
> > Depending on what you mean by "right".
>
> > The above is not enough to judge what is really happening. But depending
> > on the execution order, thread_1 overwrites the reference in COMMON_DICT
> > *after* thread_2 has altered it.
>
> > But it won't crash or anything. If that's right for you.
>
> > Diez
>
> The second thread doesn't alter the dictionary, it only reads it. Even
> if the first thread puts a reference to a new dictionary into the
> COMMON_DICT variable while the second thread is reading the old
> dictionary, such sematics is fine for me. I'm seeking for a kind of
> Producer/Consumer pattern. So the first thread produces some data in a
> form of a dictionary. The second thread reads it from time to time.
> It's sufficient for me if a dictionary is "lost" because the first
> thread overwrites the COMMON_DICT variable before the second thread
> tries to read it. It's also sufficient for me if the first thread
> updates COMMON_DICT variable while the second thread is reading the
> previous copy of the dictionary. So from my point of view the most
> critical parts are these two lines:
>
> COMMON_DICT = local_dict # in the first thread
> local_dict = COMMON_DICT # in the second thread
>
> Can these two instructions by any chance interfere with each other in
> a way that will crash either of the threads?
>
> I hope I made myself a little bit more clear :).
>
> Thanks!!
>
> Mike

I vote for OK, in this /very/ /particular/ /case/, but what would you
accept as evidence?

These pointer values start out different and become the same, but both
are dictionaries.

10490912 10490912
10490912 10490912
10490912 10490624
10490912 10490624
10490624
10490624 10490624
10490624 10490624
10490624 10490624