On Tue, Oct 25, 2011 at 9:30 PM, Xinliang David Li <davi...@google.com> wrote: > > > On Tue, Oct 25, 2011 at 1:02 AM, Richard Guenther > <richard.guent...@gmail.com> wrote: >> >> On Mon, Oct 24, 2011 at 6:27 PM, Xinliang David Li <davi...@google.com> >> wrote: >> >> Well, you seem to keep not reading what I write. I am not opposed >> >> to adding -fopt-info/report nor to funnel messages to stdout/err. What >> >> I am opposed is the way you want to introduce them. I want you to >> >> fix what we dump into dump files, so that both -fopt-report and >> >> -fopt-info >> >> can be implemented by outputting selected pieces of the dump file >> >> to stdout/stderr. We already have -fdump-*-stats which supposedly >> >> could match -fopt-report, and the default -fdump-* should be what >> >> goes to -fopt-info (minus the function bodies, of course). >> > >> > That sounds good. What you propose seems like >> > >> > -fdump-pass-[ir_only|transformation|debug]-stderr >> > >> > and -fopt-info is a short cut for >> > -fdump-tree-all-transformations-stderr >> > -fdump-ipa-all-tranformations-stderr >> > -fdump-rtl-all-transformations-stderr >> >> Yes. Note that I don't like it the way the vectorizer does (with >> -fvectorizer-verbose=... the dump files are empty). The dump >> file content should be unchanged when redirecting (parts) to >> stderr, so we have to arrange to duplicate messages in two places. >> > Vectorizer dump is a good candidate for clean up when the dumping > infrastructure improvement is done. > FYI, for now, we will implement the opt-info for some passes in the simple > way in google branches, and later migrate to trunk when the dumping > infrastructure is improved.
I was hoping you would volunteer to improve the dumping infrastructure. Richard.
