Re: [boost] RTL: Relational Template Library

From: Arkadiy Vertleyb (vertleyb_at_[hidden])
Date: 2005-10-05 11:58:56

"Andreas Pokorny" <andreas.pokorny_at_[hidden]> wrote

> On Wed, Oct 05, 2005 at 09:17:18AM -0400, Arkadiy Vertleyb
<vertleyb_at_[hidden]> wrote:
> > The RTL was discussed on this list at some time in the past.  The early
> > version of RTL is described in the March 2004 issue of CUJ.  The main
> > difference from currently discussed RML is that we decided against using
> > SQL - like interface, and use the relational algebra-based interface
> > instead.
>
> The most important part for a library like that, is the ability to have
> the table stored in a file, and only partially mapped into memory. So

It's important, but I wouldn't say "most important".  For many tasks having
the ability to serialize tables should be enough.

> some tree datastructure like a B+tree is an important requirement for
> a database library. Any chance that this will happen soon? Maybe you
> remember that I started an attempt to write such a datastructure, but I
> lost myself in details and had to give up, as soon as exams took most
> part of my free time.

So, "Any chance that this will happen soon?" :)

>    From the interface point of view, I also prefer the relational
> algebra "syntax". It somehow looks more familiar to me.
>
> > The RTL can currently be compiled with VC7.1 or GCC 3.3+ (although with
GCC
> > there might still remain some naming conflicts between MPL and STL, that
> > show up with certain usages)
>
> Because gcc also considers structures when doing adl of a function call?

Yes.

> In some previous, and also private discussions you mentioned that you
> would like to have RTL rewritten, because you made some tradeoffs to
> support older versions of gcc and vc. Is that versions the overall
rewrite?

No, this is just the port to Boost 1.33 (vary little changes, in fact).  I
would not think it should be fully rewritten, though, just re-factored to
simplify the implementation.  A few design choices, made in the very
beginning, mostly related to the lack of partial template specialization in
VC6...

Nevertheless I wanted to once more attract people's attention to the library
because the interest to relational issues seems to pop-up, and again, I
don't believe to mimick SQL is the right way to go.  The SQL was designed as
yet another attempt to try to make computers understand a language that
looks more or less like plain English.  Now I don't think that even database
comunity considers it a way to proceed.

One of the main things that prevented us from moving forward, was not enough
interest, but now we can re-consider this -- the number of downloads seems
pretty high :)

Regards,
Arkadiy