On 05/08/2017 01:32 AM, Richard Biener wrote:
Note that I tried last stage3 (it ended up being too late) to get rid
of ASSERT_EXPRs
doing substitute-and-fold itself (basically copy-propagate them out at
this point rather
than as a separate thing later). This is because the ASSERT_EXPR uses interfere
with the single_use checks in match.pd patterns and thus are actually harmful.
The barrier I ran into was the ASSERT_EXPR use by the threader ... so
now you're making
us rely even more on those :/
Perhaps, but it's a short term thing -- Andrew and I want to get rid of
ASSERT_EXPRs too. I really think we all share that common goal.
I can certainly see how they muck up the single_use checks. They get in
the way of other things as well by hiding equivalency information.
jeff
