|
Boost : |
From: loisg_at_[hidden]
Date: 2000-12-12 19:41:49
The header files would not be mixed with the .cpp files. All .cpp
files go in the src directory, below all of the library's header
files.
boost_1_19/
boost/
my_library.hpp // #includes file[12].hpp
my_library/
file1.hpp
file2.hpp
src/
file1.cpp
file2.cpp
doc/
examples/
The single level 1 "convenience" header makes it easy to get started
using that library without a lot of detailed study, to learn which
classes are declared in which headers. Granularity can be adopted
later as needed. Of course a library author might decide to set up a
more complicated structure, but the user needn't worry about that
unless s/he wants the granularity of using the individual headers.
I'm suggesting that organization instead of one like this:
boost_1_19/
boost/
include/
my_library/
file1.hpp
file2.hpp
src/
my_library/
file1.cpp
file2.cpp
doc/
my_library/
index.html
file1.html
file2.html
examples/
my_library/
example1.cpp
example2.cpp
because the former system isolates each library into a single branch
of the directory tree. The latter is equally logical, but every new
library added requires updating several locations in the boost tree.
Lois
--- In boost_at_e..., Jens Maurer <Jens.Maurer_at_g...> wrote:
> Beman wrote:
> > Wasn't the point of Lois' suggestion that you could write either:
> >
> > #include <boost/graph/point.hpp> // splitters can write this
> >
> > or
> > #include <boost/graph.hpp> // lumpers can write this
> >
> > In other words, you get to view the headers for larger libraries
whichever
> > way you prefer. We don't dictate to the user which view they
choose.
>
> Lois' suggestion seems to entail that there are header files mixed
> with subdirectories (containing implementation .cpp) in several
levels
> of the directory structure. I would consider this sub-optimal.
>
> Jens Maurer
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk