Re: [Boost-users] Re: Declaring boost.serialization archivers as members

From: Eyal Susser (eyal.susser_at_[hidden])
Date: 2005-05-16 11:06:22

By archivers I mean biary archives, text archives, and so on.  Code
snippet is a little tricky here, but the idea is more or less

CMyClass::CMyClass() : m_mystreambuf_facade(m_memberchar),
                                   m_an_iostream(m_mystreambuf_facade),
                                   m_binary_output_archiv(m_an_iostream)
{
}

//somewhere in a method

cMyBinArchive << cInstanceOfSomeClass;

//code that resets the write pointer if the streambuf_facade

//more actions

since I'm using a streambuf_facade, I control the offset from which
the stream reads.  I make sure I call a special reset method I added
to the streambuf after every serialization action
.  Since I reset the streambuf which holds the archive data, I expect
that data to change after each serialization, but it doesn't.

Hope this clarifies things in some way.

Thanks

On 5/11/05, Jeff Flinn <TriumphSprint2000_at_[hidden]> wrote:
> 
> "Eyal Susser" <eyal.susser_at_[hidden]> wrote in message
> news:dee7c1cf05051105557af2e389_at_mail.gmail.com...
> > Hi,
> >  After doing what I described in the subject, I found that when
> > marshalling data between processes, the same data gets sent over and
> > over again.  It's as if the archiver has some internal copy of the
> > stream, and keeps serializing that stream over and over.  Anyone else
> > have this problem, or is it some pitfall in my implementation?
> 
> What exactly do you mean by "boost.serialization archivers"? Can you show a
> short code snippet?
> 
> Jeff Flinn
> 
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>