Does Boost have a library that defines a file-based container, perhaps based on a B-tree or equivalent, for very large collections? Or does anyone know of such a thing, or has there been talk about developing one?
We have a few file-based collections with millions of entries. We don't especially want to install a database. The file is too large to load into memory. The file's format could be changed to make it compatible with the type of collection I'm thinking of, but for the moment, we're doing binary searches by reading the file on disc. As you might guess, this can be quite slow, especially when the file is on a DVD and not the HD.
Not in Boost now (though it is considered by author): http://i10www.ira.uka.de/dementiev/stxxl.shtml /Pavel