[GHC] #13208: Do two-phase inlining in simpleOptPgm

[GHC]

#13208: Do two-phase inlining in simpleOptPgm
-------------------------------------+-------------------------------------
        Reporter:  lukemaurer        |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.1
      Resolution:                    |             Keywords:  JoinPoints
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by bjmprice):

 What's the status of this ticket? I ask because in `Note [The simple
 optimiser]`, it is claimed that "It thereby guarantees to leave no un-
 reduced beta-redexes.", but in 8.6.3 the Haskell `beta = let f = \x -> x
 in f True` gives rise to the Core `beta = (\ (x_aVG :: Bool) -> x_aVG)
 GHC.Types.True`, which seems to contradict that claim.

 (This core is with `-ddump-ds`, which I think runs the simple optimiser.
 With `-ddump-ds-preopt` it is a let-binding, not a beta-redex.)

 I guess that the Note could mean that "all beta-redexes in the input are
 reduced", rather than "there are no beta-redexes in the output", but this
 isn't how I understood the text.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/13208#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler