[GHC] #14287: Early inlining causes potential join points to be missed

[GHC]

#14287: Early inlining causes potential join points to be missed
-------------------------------------+-------------------------------------
        Reporter:  jheek             |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.2.1
      Resolution:                    |             Keywords:
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 jheek):

 Actually sfilter3 does not produce optimal code for me. It is correctly
 recognized as a join point but it gets floated out of to be a top-level
 binding before the recursive case-of-case transformation jumps in.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/14287#comment:4>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler