From: Jeremy Siek (jsiek_at_[hidden])
Date: 2000-10-25 13:10:57
On Wed, 25 Oct 2000, Daryle Walker wrote:
> 3. I don't know whether or not to use your graph stuff. I think it can let
> me assign states and transitions to items (properties to nodes and edges,
> respectively). But I'm not sure if it can help with single transversals.
> Another problem is that the sub-library is big; I may not want to drag a lot
> of baggage for a specialized field. Making custom graph stuff (maybe with
> std::map's) could be easier.
I was assuming that the finite state automata algorithms would be
written in a generic fashion (template functions). Therefore, they
would be data-structure (graph class) independent and would not
"drag in" data-structure code from the graph library. This is the way the
graph algorithms are written... the user can use breadth_first_search()
without touching any other BGL component or header (other than the
graph_traits and property_traits classes).
If you want to write your own simple graph class based on std::map, that's
great. It's just that I'd like to *also* be able to use other graph
classes with the finite state automata algorithms.
> I wonder: if someone does your project, or if I do mine, could the regex guy
> use it for a new implementation?
That would certainly be a good test usage.
> P.S. Your forwarded messages mentioned some papers (listed below). Are
> they available in PDF format?
Sorry, I don't know.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk