[boost] Re: In-memory databases

From: Arkadiy Vertleyb (vertleyb_at_[hidden])
Date: 2003-02-01 14:00:21

We did consider serialization.  You can take a look at test/persist.cpp and
persist.h.  They implement serialization of a tuple.  With the help of this
it's quite easy to serialize any relation (whether table or operator).  This
can be used either to implement persistable tables, or for client-server
data transfer.

There is yet another aspect.  In our library a table is separated from its
implementation (although right now it may be done not in the best possible
way).  Our current implementation is based on the sorted std::vector.
Ideally, there should be multiple implementations, memory and disk based
(talking about persistence).  Such differently implemented tables should be
able to join, etc.

Although, theoretically, table implementations are out of scope of the
library, we'll work in this direction if people consider this important.
However, our resources are quite limited.  Anybody interested is invited to
cooperate.

Arkadiy

"David B. Held" <dheld_at_[hidden]> wrote in message
news:b1h2ej$ibe$1_at_main.gmane.org...
> "Arkadiy Vertleyb" <vertleyb_at_[hidden]> wrote in message
> news:b1grsi$qh2$1_at_main.gmane.org...
> > Yes, we are currently working on such a database.
> > [...]
>
> After looking at the documentation, it looks like you have left out
> serialization/persistence completely.  That's ok, if that's the aim of
your
> project.  Did you plan to add that later?  It would be nice to see some
> progress on the serialization/persistence lib.  I think these could be
> put together into a pretty powerful library.  I think the harder part is
> writing the query engine, which it looks like you already have done
> (more or less).
>
> Dave
>
>
>
> _______________________________________________
> Unsubscribe & other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost
>