RE: [boost] [Review] Boost Graphs Library (GGCL)

From: Beman Dawes (beman_at_[hidden])
Date: 2000-09-01 13:43:49

jsiek_at_[hidden] wrote:

 >I'm getting ready to reorganize the file structure of GGCL
 >to fit into boost.

You might want to relax over the weekend; the formal review period doesn't 
end until Sunday, IIRC:-)  Give the Review Manager (Dave) a chance to 
comment on issues raised during the review.

 > We have two options:
 >
 >1. move everything from ggcl/ to boost/
 >2. move everything from ggcl/ to boost/graph/
 >
 >I think 1) would be most convenient, but there are a awful lot of
 >files. 2) would keep boost/ from getting as cluttered.

There are just too many files. I strongly vote for putting them in 
boost/graph.

 > ...
 >
 >The following is a list of all the header files that would be
 >added to boost.
 >
 > ...
 >
 >Interface/Traits
 >----------------
 >boost/graph_traits.hpp
 >boost/property_accessor.hpp
 >boost/concept_checks.hpp
 >boost/graph_concepts.hpp
 >boost/container_traits.hpp
 >
 >
 >Heaps, Queues, etc.
 >----------------
 >boost/fibonacci_heap.hpp
 >boost/mutable_queue.hpp
 >boost/bucket_sorter.hpp
 >
 >
 >Misc
 >----
 >boost/limits.hpp	    (for compilers lacking c++ header limits)
 >boost/graph_as_tree.hpp
 >boost/graph_utility.hpp
 >boost/meta.hpp  (compile time IF, etc.)

Hum... Some of these are of general interest (meta.hpp, etc.) or are 
special in some other way (limits.hpp).

We need to talk more about headers with lots of potential use outside the 
context of ggcl. Such headers should live in boost, rather than 
boost/graphs. They should be documented separately, too.  For generally 
useful components, I'd like the docs in HTML.  They should really have one 
or more separate formal reviews, too.  Ideas?

--Beman