From: Jonathan Turkanis (technews_at_[hidden])
Date: 2005-09-08 19:07:35
David Abrahams wrote:
> The problem of building customization points for generic C++ libraries
> has no really satisfying solution at the moment. Any approach we
> choose will be a compromise. It isn't clear to me why this particular
> compromise should be deemed worse than any of the others.
I have a similar choice to make regarding customization points for Iostreams.
Currently, the user can customize boost::iostreams::read(),
boost::iostreams::write(), etc. for a type T by specializing the class template
boost::iostreams::operations. I'm considering switching to an ADL-based
mechanism in which users define functions boost_iostreams_read,
boost_iostreams_write, etc. I haven't documented the customization mechanism yet
because I'm waiting for agreement on the design of Boost.Range and I think it
would be beneficial to have a uniform policy.
Personally, I don't really care which solution is adopted, but I'm eager for the
problem to be resolved soon so library users can take advantage of
customization.
Jonathan