[boost] Serialization Library - Two challanges

From: Robert Ramey (ramey_at_[hidden])
Date: 2002-11-17 13:55:33

To boost members interested in the serialization library.

There has been much interest in adding things to the library
which I presonally believe don't belong there.  Much effort
has been expended in the design to keep the library
independent of the serialized data types and archive storage.
I believe that these are absolutely curcial to the ultimate
success of this library.  I propose the following challanges
to boost members.  Each challange should take no more
than one man day.  If it can't be done in a man day I want
to know. 

archive storage independence.
======================

Could some one who has XDR code implement a derivation 
of basic_[i|o]archive.  

class xdr_iarchive : public basic_iarchive
...
class xdr_oarchive : public basic_oarchvie
...

data type independence
=================
there has been talk of an in memory relational database. 
(I think).  I if serialization were implemented for this
class, I would have something I always wanted: A very small,
very fast, persistent, database for databases that fit in main
memory.  Given the size of main memory these days, this
woudl be very helpful to me.  This should use non-intrusive
template serialization.  It is a harder task than the first
but I still believe it can be done in one day.  You can
use the serialization of the stl collections to get a preview
of what's involved.

Robert Ramey

P.S.
If someone really want's a small interesting project along these lines
why not try to implement serialization for the relational algebra
package.  The net result of such an effort would be an extremely
fast, small database package for databases that fit in all of main
memory.  This would be extremely useful given the size of memory
available these days.

My experience in doing this for all the STL collections suggests
that this is about two days work.  I think it would be great
proof of concept