From: Arkadiy Vertleyb (vertleyb_at_[hidden])
Date: 2006-05-10 05:59:32
"David Abrahams" <dave_at_[hidden]> wrote
> "Arkadiy Vertleyb" <vertleyb_at_[hidden]> writes:
> > However BOOST_TYPEOF, although does require user type registration, has
> > advantage of much nicer syntax. Compare:
> > result_of<minus_(minus_(int, _1_), _2_)>::type f = 3 - _1 - _2;
> > with
> > BOOST_AUTO(f, 3 - _1 - _2);
> Yes, of course it does. You don't think the advantages of auto are
> lost on me, do you?
> Inside my library's implementation I'm willing to buy lack of
> registration with a slightly uglier syntax, because asking people to
> register types is completely untenable for me.
> > It would be very nice to have lambda and bind types registered, so that
> > user at least have a choise.
> It's utterly irrelevant to me, since there's no way I'm going to ask
> users to register their types just so *I* can use typeof.
I didn't realise it's for *your* code. The examples you provided didn't
look to me generic enough to belong to a library. I do agree that usage of
BOOST_TYPEOF in generic code should be avoided as long as there is a
solution with no registration.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk