[GHC] #11731: Simplifier: Inlining trivial let can lose sharing

[GHC]

#11731: Simplifier: Inlining trivial let can lose sharing
-------------------------------------+-------------------------------------
        Reporter:  nomeata           |                Owner:
            Type:  bug               |               Status:  patch
        Priority:  normal            |            Milestone:
       Component:  Compiler          |              Version:  8.1
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):  Phab:D2064
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by nomeata):

 Sounds reasonable, but it raises a few questions:

 >  Indeed maybe the demand analyser should not even record signle-entry-
 ness in the Ids, since it is so fragile.

 Yes, I’m all for not storing information that we do not keep up-to-date,
 as it will just be too likely that a later pass uses them, and things
 break again in obscure places. But then this should also apply to the
 strictness-and-demand signature of functions, which should be stripped of
 any `1*` information! But that’s not too bad; the final, non-ww pass could
 attach full strictness signatures to exported functions (at least as long
 as we don’t have CSE as a STG-to-STG-transformation as proposed in #9291).

 What should be the consequence for `OneShot` annotations on lambda
 binders? Don’t all the same considerations apply here?

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11731#comment:23>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler