character references are not recognized in emphasized text

[g9ks157k@…]

If I write /Hello&#A0;world!/ in a doc string, I don’t get a non-breaking space between “Hello” and “world!” in the output but I get “&#A0;” instead.

[SamB]

This appears to be an infelicity in the lexer (GHC's compiler/parser/HaddockLex.x):

<string,def> {
  $special                      { strtoken $ \s -> TokSpecial (head s) }
  \<\<.*\>\>                    { strtoken $ \s -> TokPic (init $ init $ tail $ tail s) }
  \<.*\>                        { strtoken $ \s -> TokURL (init (tail s)) }
  \#.*\#                        { strtoken $ \s -> TokAName (init (tail s)) }
  \/ [^\/]* \/                  { strtoken $ \s -> TokEmphasis (init (tail s)) }
  [\'\`] $ident+ [\'\`]         { ident }
  \\ .                          { strtoken (TokString . tail) }
  "&#" $digit+ \;               { strtoken $ \s -> TokString [chr (read (init (drop 2 s)))] }
  "&#" [xX] $hexdigit+ \;       { strtoken $ \s -> case readHex (init (drop 3 s)) of [(n,_)] -> TokString [chr n] }
  -- allow special characters through if they don't fit one of the previous
  -- patterns.
  [\/\'\`\<\#\&\\]                      { strtoken TokString }
  [^ $special \/ \< \# \n \'\` \& \\ \]]* \n { strtoken TokString `andBegin` line }
  [^ $special \/ \< \# \n \'\` \& \\ \]]+    { strtoken TokString }
}

Really, I guess the problem is that this is done in the lexer at all -- it should probably be done in HaddockParse.y instead.

[simonmar]

one possible reason for this is so that lone '\' characters don't cause a parse error. That used to be one of the most common reasons for accidental parse errors in Haddock docs. If you look back through the logs you'll probably see when that change was made.

[anonymous]

Milestone 2.5.0 deleted