avoid using curl for file:// urls since it's buggy

Utility/Url.hs

@@ -17,6 +17,7 @@ module Utility.Url (
 
 import Common
 import Network.URI
+import Utility.CopyFile
 
 type URLString = String
 
@@ -71,11 +72,13 @@ exists url headers = case parseURI url of
  - would not be appropriate to test at configure time and build support
  - for only one in.
  -
- - Curl is always used for file:// urls, as wget does not support them.
+ - For file:// urls, neither program works well, so we just copy.
  -}
 download :: URLString -> Headers -> [CommandParam] -> FilePath -> IO Bool
 download url headers options file
-	| "file://" `isPrefixOf` url = curl
+	| "file://" `isPrefixOf` url =
+		let f = drop (length "file://") url
+		in copyFileExternal f file
 	| otherwise = ifM (inPath "wget") (wget , curl)
   where
 	headerparams = map (\h -> Param $ "--header=" ++ h) headers

doc/bugs/get_fails_for_file:__47____47___web_remotes_if_the_file_is_empty.mdwn

@@ -19,3 +19,8 @@ Debian sid/experimental
     default repository version: 3
     supported repository versions: 3 4
     upgrade supported from repository versions: 0 1 2
+
+> Seems that curl doesn't write the file it was asked to 
+> output to in this case. Crazy. Does not affect empty 
+> http urls. Switched to using just cp for file://
+> [[done]] --[[Joey]]