Subj : Re: Can C++ local static objects be made thread safe?
To   : comp.programming.threads
From : David Schwartz
Date : Thu Feb 03 2005 06:26 pm


"Gianni Mariani" <gi2nospam@mariani.ws> wrote in message 
news:2JudnbZqFeq2oZ_fRVn-2g@speakeasy.net...

>>  A threading standard
>> must specifically restate any C++ standard guarantees that it provides in 
>> the context of threads. If we're talking POSIX, POSIX specifically 
>> requires accesses to shared data (which the static object is) to be 
>> protected by mutexes.
>
> Only if the shared data is modified.

    The thing is, it's not clear at what level the "modification" needs to 
take place. For example, consider a class that caches some information 
internally to save lookups. This can cause an access that is purely a "get 
information" operation to result in a modification to the internal state of 
the object. So, to be more precise, you must hold a mutex unless you can 
prove modification is impossible. In this case, you cannot. So a mutex is 
required.

    DS

.