Re: [boost] Boost.SQL, SOCI and beyond

From: Nicola Musatti (Nicola.Musatti_at_[hidden])
Date: 2006-10-05 06:39:24

Jeff Garland <jeff <at> crystalclearsoftware.com> writes:
> Nicola Musatti wrote:
[...]
> > Missing the TR2 deadline was unfortunate, but probably putting together 
> > a valid proposal deserved more time than was available, even with such 
> > an impressive starting point such as SOCI.
> 
> I think there was more than enough time.  I tried to talk everyone working on 
> the various database libraries to work together to bring a proposal.  I didn't 
> succeed in convincing them 

I'm not convinced, especially since as far as I know all of the involved people
work on this issue out of their spare time. Rather I believe the question is how
to ensure that developers can take profit of other libraries that are compliant
to Committee approved interfaces, before another 10 to 15 years go by, but
that's another story which probably is best discussed elsewhere.
 
[...]
> Well before I even consider engaging this discussion, why not help SOCI get 
> into boost instead of inventing a new interface?  We have several proposals 
> and half-finished libraries, but the real problem is that we don't have an 
> actual reviewed and accepted library to put into Boost.  Are you going to have 
> time to really develop this library from scratch?

Just to make this clear: I have absolutely nothing against a boostified version
of SOCI being accepted into Boost and eventually submitted to the Committee for
inclusion in a future TR or directly into the Standard.

However I am concerned about the suitability of some of the interface choices
made by the SOCI developers. I started writing my own library because I felt
that my arguments would carry more weight with a concrete implementation to
support them, however experimental, and because starting from scratch with Boost
compliance in mind was simpler than attempting a partial boostification of the
SOCI code base.

Unless the SOCI developers decide for some reason that they are not interested
in submitting their library to Boost anymore I see no reason for my library to
survive as anythong more than a personal pastime. Should that happen I feel
confident that I will be able to complete the ODBC backend I'm working on; more
than that is unlikely.

> In my view, SOCI has all the qualities to succeed as a Boost library including 
> a clean, good documentation, support for multiple databases, and several 
> developers to support it.  So maybe you should start by downloading, 
> installing, and reviewing the SOCI interface in comparison to what you are 
> thinking?

My example is the result of comparing SOCI's API with that of similar,
proprietary libraries I'm familiar with and trying to think about the best way
to transform those in a standard compatible API.

Cheers,
Nicola Musatti