------- Comment #46 from paolo dot carlini at oracle dot com 2009-11-03 15:33 ------- Created an attachment (id=18956) --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18956&action=view) Third draft, also regtests fine
This version restricts the copies to "simple" types (per your initial idea, here scalar types, similarly to std::fill, but more general types could be also added) and sets up some general infrastructure for optimizing k > 1 too. Maybe you like it better, I don't think we want to go much more beyond this, for 4.5.0. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41351
