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

16 Sep 2018

      #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):

 Took me a very long time to realise I could use `ghc -e ':t 1e100000000'
 -ddump-parsed` to spit out the parsed file.

 First I dug into the generated source for the parser trying to wrap my
 head around how it's built. Then I tried to write a program to spit out
 the parse result, but that was ultimately fruitless:

 {{{#!hs
     module Blah () where

     import Parser
     import Lexer
     import GHC
     import DynFlags
     import StringBuffer (stringToStringBuffer)
     import FastString
     import SrcLoc
     import Outputable

     main :: IO ()
     main = do
       dynFlags <- getProgramDynFlags
       let x = runParser dynFlags "putStrLn \"Hey\"" parseStatement
       case x of
         POk pstate res -> putStrLn $ ppr res
         _ -> putStrLn "Fail"

     runParser :: DynFlags -> String -> P a -> ParseResult a
     runParser flags str parser = unP parser parseState
       where
         filename = "<interactive>"
         location = mkRealSrcLoc (mkFastString filename) 1 1
         buffer = stringToStringBuffer str
         parseState = mkPState flags buffer location

 }}}

 That was pretty silly... anyway, I subsequently realised I could use flags
 to spit out stage output. I should have read the docs on hacking on GHC
 more! That yielded this:

 {{{#!shell
 ➜  ghc git:(master) ✗ ./inplace/bin/ghc-stage2 blah2.hs -ddump-rn -dppr-
 debug
 [1 of 1] Compiling Blah2            ( blah2.hs, blah2.o )
 blah2.hs:1:1:

     ==================== Renamer ====================
     nonrec {blah2.hs:3:1-24}
            main:Blah2.largevalue{v rpX}
            main:Blah2.largevalue{v rpX}
              = {blah2.hs:3:14-24}
                1e100000000 (base:GHC.Real.fromRational{v 02C})
            <>

 }}}

 So I'll dig into GHC.Real.fromRational next and see if I can't work out
 what's going.

-- 
Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15646#comment:4
GHC http://www.haskell.org/ghc/
The Glasgow Haskell Compiler