RE: [boost] directed graph template library

From: Moore, Paul (paul.moore_at_[hidden])
Date: 2000-04-20 05:37:14

From: David Bien [mailto:davidbien_at_[hidden]]
> 
> Anyway, I have a directed graph *implementation* - for which I 
> happened to define my own type names and interface names. This was 
> not meant as an affront to anyone associated with boost, I did not 
> even know about boost until about 5 days ago.

Please DON'T feel as if your work is unappreciated, or unwanted. There is a
lot of interest in this group (rightly so!) in getting "best of breed"
semantics and designs tied down solid. However, this does need to be backed
by a solid core of implemented code.

Personally, I don't have much feel for graph algorithms, for the simple
reason that I have never had a solid implementation to play with. I have
written code which feels like it would benefit from a drop-in graph
implementation, but it has never been important enough to make me write one.
(It's a bit like the old days, when everyone half-implemented associative
arrays, because the standard map<> class didn't exist).

I think that, as well as defining a "standard-level" interface and
implementation for things like graph algorithms, Boost should be offering
workable versions right now, to get people using the things. That's the
basis on which I submitted the rational number class - it works, and it can
be tried by people and we can get ideas from experience. It's also a bit
like the smart pointer stuff - it's there and usable, but we're still
hacking round the details.

I don't know whether I'm saying that Boost should adopt David's code, or
not. I don't have the experience to know. But I DO want to be sure that it
doesn't get lost... At the very least, maybe Boost needs to include a page
of links to other library implementations which are "in the spirit of" what
we are doing.

Paul.

PS I'm a worse case, as I'm stuck with MSVC, and many of the serious new
stuff is pushing C++ beyond what MSVC will cope with. So my hidden agenda is
to have access to things which I can use on MSVC, not just theoretical best
approaches which I can't use in practice :-( I get the feeling that David's
graph implementation is closer to working on MSVC than GGCL (but I can't
find where I got that impression now...)