#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 monoidal): The literal `1e100` means `fromRational (100...000 :: Rational)`, as specified in Haskell report. Constructing this number takes time and space proportional to the number of zeroes. It's not surprising it crashes badly when the exponent has 10 digits or more. On the other hand, the type of `10^1000000000` can be found quickly because the expression is not evaluated. The big integer is already created during parsing (as can be seen by compiling `main = print 1e1000` with `-ddump-parsed-ast`). With `-XNumDecimals`, we need to process the literal before we can tell its type: `1.234e3` is a valid `Integer` but `1.234e2` is not. I don't see any easy way to fix this and preserve backwards compatibility. Perhaps we could show a parse error when attempting to create an integer with an unrealistic exponent. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15646#comment:6 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler