[GHC] #15646: ghci takes super long time to find the type of large fractional number

[GHC]

#15646: ghci takes super long time to find the type of large fractional number
-------------------------------------+-------------------------------------
        Reporter:  Johannkokos       |                Owner:
                                     |  JulianLeviston
            Type:  bug               |               Status:  new
        Priority:  normal            |            Milestone:  8.6.1
       Component:  GHCi              |              Version:  8.4.3
      Resolution:                    |             Keywords:  newcomer
Operating System:  Unknown/Multiple  |         Architecture:
 Type of failure:  Compile-time      |  Unknown/Multiple
  performance bug                    |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by JulianLeviston):

 Ok, so I tried to tackle this again today. The very first thing I did was
 to refactor `data FractionalLit` as suggested, then follow the type
 errors.

 Firstly I wasn't 100% sure how to best approach a workflow, so I tried
 using GHCi and loading the BasicTypes module in, but that failed because
 of missing dependencies, so then I tried to use `make fast -j3` in the
 `ghc/compiler` directory and that seemed to work out ok in terms of speed.
 I was initially worried it was going to take too long, but it's fine.

 So, adjusting `FractionalLit` is fine. I made the change, then it
 complained bout `mkFractionalLit` being incorrect, as I'd expect.

 I'm not 100% sure of the intent of `mkFractionalLit`, though — it looks
 like it's a simple constructor; when I compare it to `mkIntegralLit`, etc,
 and when I looked up the use-sites it appears to passing a Rational value
 still makes sense, so I thought I'd keep that. Then I realised I'm going
 to have to effectively implement something like an adjusted version of
 `fromBaseRational 10` from https://hackage.haskell.org/package/numeric-
 prelude-0.4.3/docs/src/Number-Positional.html#fromBaseRational to find the
 mantissa and exponent. Am I barking up the right tree here?

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15646#comment:26>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler