Re: [boost] [gsoc] Interest in BGL v2?

Subject: Re: [boost] [gsoc] Interest in BGL v2?
From: Michael Lopez (mlopez7_at_[hidden])
Date: 2011-03-31 09:01:01

>
>
> I won't argue that today, although I'm going in the opposite direction with
> my Metagraph project.  I'm making the pro- argument in my upcoming BoostCon
> paper/talk on MPL.Graph.
>
> I wonder if anyone can say something about the new graph concept hierarchy?
>  I don't see any origin docs yet.
>

The BGL concept hierarchy is being re-evaluated based on the rewrite of the
data structures and algorithms, meaning that the concepts come after we
implement most of the data structures and algorithms. We will then analyze
the library and define the concepts based on the types and algorithms we
have in place. After constructing the concept hierarchy, we will refine the
graph types, then the concepts, etc. This iterative method is the approach
we've chosen for Origin.

There was a time when we started to build the concept hierarchy before the
data structures were in place and even at the same time the data structures
were being implemented, but that wasn't working. We found that building data
structures to model concepts, at least initially, restricted us to the
design decisions we made with the concepts.

> I'd like to stay compatible and I'm also curious what was found lacking in
> the old graph concepts.  The unifying graph concepts in BGLv1 are really
> impressive for making all graph libraries adaptable to its interface &
> algos.
>
> So, having not constructed the concept hierarchy, I cannot yet tell you
what we are lacking.