[GHC] #11770: Demand analysis: Wrong one-shot annotation due to fixed-point iteration

[GHC]

#11770: Demand analysis: Wrong one-shot annotation due to fixed-point iteration
-------------------------------------+-------------------------------------
        Reporter:  nomeata           |                Owner:
            Type:  bug               |               Status:  new
        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):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by simonpj):

 Yikes!  That's terrible!

 As well as that outright bug, there's a strange duplication:
  * the demand analyser goes to some effort to annotate one-shot info (e.g.
 `annLamWithShotness`)
  * but so does the occurrence analyser (see the `occ_one_shots` field of
 `OccEnv`)

 Let's do it in the occurrence analyser, as now.  So try this:

 * Get rid of all one-shot setting in the demand analyser, letting the
 occurrence analyser do the work. The demand analyser would then annotate
 only demand info.

 * BUT in `OccAnal.oneShotGroup` I'm suspicious of that `updOneShotInfo`
 which keeps one-shot info if it was there before. I don't think it's an
 actual bug, but I'd be much more comfortable with using `setIdOneShotInfo`
 instead.  That's an orthogonal change.

 Would you like to try those two changes (ideally independently) and see
 what effect it has?

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/11770#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler