comments only: point to relevant bug reports

[ghc-hetmet.git] / compiler / ghci / InteractiveUI.hs

diff --git a/compiler/ghci/InteractiveUI.hs b/compiler/ghci/InteractiveUI.hs
index e0fddac..7c64941 100644 (file)
--- a/compiler/ghci/InteractiveUI.hs
+++ b/compiler/ghci/InteractiveUI.hs
@@ -1536,14 +1536,17 @@ setUpConsole = do
        -- similarly for characters we write to the console.
        --
        -- At the moment, GHCi pretends all input is Latin-1.  In the
-       -- future we should support UTF-8, but for now we set the code pages
-       -- to Latin-1.
+       -- future we should support UTF-8, but for now we set the code
+       -- pages to Latin-1.  Doing it this way does lead to problems,
+       -- however: see bug #1649.
        --
        -- It seems you have to set the font in the console window to
        -- a Unicode font in order for output to work properly,
        -- otherwise non-ASCII characters are mapped wrongly.  sigh.
        -- (see MSDN for SetConsoleOutputCP()).
        --
+        -- This call has been known to hang on some machines, see bug #1483
+        --
        setConsoleCP 28591       -- ISO Latin-1
        setConsoleOutputCP 28591 -- ISO Latin-1
 #endif
@@ -1811,8 +1814,9 @@ findBreakByCoord mb_file (line, col) arr
 -- TERM to vt100 for other reasons) we get carets.
 -- We really ought to use a proper termcap/terminfo library.
 do_bold :: Bool
-do_bold = unsafePerformIO (System.Environment.getEnv "TERM") `elem`
-          ["xterm", "linux"]
+do_bold = unsafePerformIO mTerm `elem` ["xterm", "linux"]
+    where mTerm = System.Environment.getEnv "TERM"
+                  `Exception.catch` \e -> return "TERM not set"
 
 start_bold :: String
 start_bold = "\ESC[1m"