From: Jaakko Jarvi (jajarvi_at_[hidden])
Date: 2003-10-10 17:24:03
First, the thing we were talking about in the previous email is
probably correct behavior according to the standard, so it is probably
not a g++ bug :) The Intel compiler has the same behavior.
We're currently figuring out again the behavior of different
compilers, and cannot repeat the same cases that motivated
lazy_enable_if. Compilers have been updated since we last tested.
We'll report later...
Jaakko & Jeremiah
In our last exciting episode Gabriel Dos Reis wrote:
> Lines: 30
> Jaakko Jarvi <jajarvi_at_[hidden]> writes:
> | In our last exciting episode David Abrahams wrote:
> | > > 1. lazy_enable_if
> | > >
> | > > Shouldn't be this named apply_enable_if. It very similar to the appropriate
> | > > MPL concept and seems logical to be named the same.
> | > I had the same thought, FWIW. I just couldn't make it sound right in
> | > the ear, though.
> | > On second thought, I wonder if we need it at all. Isn't
> | > mpl::apply_if<condition, nullary_metafunction, disabled>::type
> | > equivalent, where disabled is a struct with no nested ::type?
> | This wouldn't work. At the outermost level (apply_if), ::type always
> | exists, so this would lead to an error, not disabling the function, on
> | compilers (e.g. g++) where SFINAE is not applied when the invalid type
> | occurs nested in another instantiation.
> Jaakko --
> I'm trying to catch up with this thread. Please could you elaborate
> on the problem with g++ here?
> -- Gaby
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk