Subj : Re: Can C++ local static objects be made thread safe?
To   : comp.programming.threads
From : David Schwartz
Date : Wed Feb 16 2005 03:10 am


"Alexander Terekhov" <terekhov@web.de> wrote in message 
news:42132303.64B4747@web.de...

> David Schwartz wrote:
> [...]
>>     I am saying that. That is what makes them thread local.

> The mappings are local. Objects are not.

    That is one implementation. It is not the only implementation.

    There will always have to be restrictions on passing the address of 
thread-local variables between threads because the implementation manages 
their lifetime. What those restrictions are can certainly vary,

> [...]
>>     That's ridiculous. Providing no strictly-conforming code can tell the
>> difference, POSIX is silent. And strictly conforming POSIX code can't use
>> the '__thread' specifier.

> Sure it is extension and no "strictly-conforming code" can use. But
> extensions are not supposed to break the model, only extend it.

    That's not an argument. Either way, it's an extension.

>>     In any event, many implementations have at least one page that is not
>> shared by all the threads. This is a common way to, for example, tell 
>> which
>> thread you are.

> The stuff on those pages isn't addressable by C/POSIX object
> addresses. It's irrelevant implementation detail.

    Neither are variables that use '__thread' since '__thread' is not 
defined by C or POSIX.

    The requirement you claim exists simply does not exist. So it is 
impossible for an implementation to violate it.

    DS

.