Re: [boost] Re: Threads

From: Lois Goldthwaite (loisg_at_[hidden])
Date: 2000-08-04 12:23:17

Well, if we really want to get down to basic definitions, the primary
requirement is an atomic test-and-set primitive, right?

Steve's points need to be addressed early on. IIRC, on Win32 all mutexes are
recursive. But on Solaris, it's quite possible for a thread to deadlock itself
by attempting to lock a mutex it already holds. You can implement the recursive
behavior on top of the other, but not vice versa.

Lois

William Kempf wrote:

> --- In boost_at_[hidden], Beman Dawes <beman_at_e...> wrote:
> > William Kempf wrote:
> >
> >  > ...
> >  >
> >  > ...  There are numerous ways to do
> >  >synchronization, but the most primitive form available to library
> >  >writers is the mutex.  Unless someone has some other idea, can we
> at
> >  >least start our discussion with this primitive?
> >
> > What base document do we look at for the definition of a mutex?
>
> I provided the definition in another post, so I won't state it
> again.  I admit that I should have given the definition of a mutex
> before asking the question you quote above.  However, the fact that
> you asked this begs a question to me.  Why are you asking what a
> mutex is?
>
> I fully agree that we need to have definitions for everything we
> discuss here, and I can see the need for a formal definition being
> given before terms are used here (and possibly revised as the
> discussion progresses).  However, there are several definitions that
> are well known, and a mutex is one such.  Other areas of boost have
> been discussed and worked on here with out formal definition of well
> known terms within their area of use.  Is the problem here that not
> enough people are familiar with the basic terms of cuncurrent
> programming, or is it something else that I'm missing.  I ask so that
> I know how to proceed in this discussion.
>
>