On Tuesday 12 October 2010 22:37:06 Eric Niebler wrote: > On 10/12/2010 1:24 PM, Eric Niebler wrote: > > So it really seems to me that transform_expr is not necessary, but I > > may be wrong. > > I just confirmed this by trivially replacing uses of transform_expr with > appropriate uses of nary_expr in your phoenix_test.cpp (attached). > > The only difference is the type of the last expression. Since > pass_through transform leaves terminals alone, _1 ends up stored by > reference in the transformed expression, which is perfectly OK and saves > copies (admittedly trivial in this case).
You are right, transform_expr can be replace by nary_expr. transform_expr is a little less verbose but not as flexible, I admit. The motivation behind transform_expr to have something similiar to fold to resemble these "standard" algorithms like fold and transform without noticing they are already there. So I guess it would be a nice addition to the docs to have a mapping between algorithms people already know to how to express it with proto. _______________________________________________ proto mailing list proto@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/proto
