On Thu, Mar 14, 2013 at 7:55 AM, Marc Glisse <marc.gli...@inria.fr> wrote: > On Wed, 13 Mar 2013, Diego Novillo wrote: > >> This patch adds an initial implementation for a new helper type for >> generating GIMPLE statements. > > > I hope you'll forgive the naive newbie question: why is the gimplifier used > so little outside of the gimplification pass? For instance, after a call to > fold in a gimple pass, we test valid_gimple_rhs_p and give up if it returns > false, instead of trying to gimplify whatever fold returned. The connection > with your patch is that generic trees are easier to build than gimple > statements, so for a long expression one could imagine building a tree and > having a single gimplification call at the end (the time wasted should be > quite limited).
It also creates more garbage and increase ggc overhead? David > > (Note that I encourage the efforts to simplify the gimple interface, I am > just taking the occasion to ask a vaguely related question) > > -- > Marc Glisse