#8226: Remove the old style -- # Haddock comments. -------------------------------------+------------------------------------ Reporter: Fuuzetsu | Owner: Type: task | Status: new Priority: normal | Milestone: 7.10.1 Component: Compiler | Version: 7.7 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: Unknown/Multiple Type of failure: None/Unknown | Difficulty: Unknown Test Case: | Blocked By: Blocking: | Related Tickets: -------------------------------------+------------------------------------ Comment (by adinapoli): (Rewriting comment that the tracker wiped :'( ) In a nutshell this is my first session of GHC hacking ever, so please be patient if - as I'm certain - my patch doesn't fully fix the issue or is going in the wrong direction. I have manually tested the fix modifying slightly the testcase attached to the relevant haddock ticket: http://trac.haskell.org/haddock/ticket/171 {{{ {- # LANGUAGE RecordWildCards #-} module Fail where data Foo = Foo { a :: String , b :: Int } showFoo Foo{..} = a main :: IO () main = print $ showFoo (Foo "a" 10) }}} As pointed out that one is _not_ a pragma (mind the space), so it needs to be interpreted as a normal comment, but Haddock chokes on it: {{{ -- pre-fix [nix-shell:ghc]$ ghc -haddock Fail.hs [1 of 1] Compiling Main ( Fail.hs, Fail.o ) Fail.hs:1:4: parse error on input ‘#’ }}} But after my patch it seems to accept it: {{{ [nix-shell:ghc]$ ghc-dev -haddock Fail.hs [1 of 1] Compiling Fail ( Fail.hs, Fail.o ) Fail.hs:10:9: Illegal `..' in record pattern Use RecordWildCards to permit this }}} I have ran "validate" before and after the patch and it seems no regression was introduces, but being the Lexer such a delicate part I really want to be cautious with it :) Thanks guys, hope I'm on the right track :) Alfredo -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/8226#comment:2 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler