Douglas Gregor <gregod@cs.rpi.edu> writes:
On Saturday 01 May 2004 11:45 am, Douglas Gregor wrote:
I think I have some free time as well... I'll need to get reacquainted with the existing result_of code (in the sandbox) before I know what to ask for.
It looks like the result_of implementation is in pretty good shape. I've expanded/fixed it to deal with a few cases I'd missed before. I'll check it in Two general issues remain: "support" for broken compilers and a matter of protocol.
Broken compilers can't use result_of: without partial specialization, it's completely unusable. Also, compilers that don't support SFINAE can't detect result_type. I'm not sure how limiting this will be, but obvious Borland, HP, and pre-7.1 Microsoft are all toast.
As for the protocol side of things: does anyone think we need to review a library that is (1) merely scaffolding for other libraries, (2) trivial, and (3) already blessed by the standards committee? Methinks not.
We probably don't need to. Maybe we should anyway, just to give the fast track review process some exercise? I don't really have a position on what the answer should be. -- Dave Abrahams Boost Consulting http://www.boost-consulting.com