[GHC] #7450: Regression in optimisation time of functions with many patterns (6.12 to 7.4)?

[GHC]

#7450: Regression in optimisation time of functions with many patterns (6.12 to
7.4)?
-------------------------------------+-------------------------------------
        Reporter:  iustin            |                   Owner:
            Type:  bug               |                  Status:  new
        Priority:  normal            |               Milestone:  7.12.1
       Component:  Compiler          |                 Version:  7.6.1
      Resolution:                    |                Keywords:
Operating System:  Unknown/Multiple  |            Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |               Test Case:
      Blocked By:                    |                Blocking:
 Related Tickets:                    |  Differential Revisions:
-------------------------------------+-------------------------------------

Comment (by simonpj):

 Ben,  good work.  I think you said that the culprit is `consDictBind`.
 Here are some ideas:

  * `snocDictBinds is only ever called with a singleton argument.  Nuke it
 in favour of `snocDictBind`.

  * The calls to `snocDictBind` in `specBind` have just called
 `flattenDictBinds` which stupidly throws away all the free varaible
 information cached in the `DictBind`.  Better to keep it!  To achieve that
    * `snocDictBind` should take a `DictBind` not a `CoreBind`
    * `flattenDictBinds` should return a `DictBind` not a `CoreBind`.  This
 is the place we are discarding the free var info. Instead, keep it!

 Does that make sense?

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/7450#comment:15>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler